National Academies Press: OpenBook

Identification of Research Needs Related to Highway Runoff Management (2004)

Chapter: Chapter 3 - Review of Published Literature and Potential Research Needs

« Previous: Chapter 2 - Department of Transportation Research Preferences
Page 57
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 57
Page 58
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 58
Page 59
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 59
Page 60
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 60
Page 61
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 61
Page 62
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 62
Page 63
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 63
Page 64
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 64
Page 65
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 65
Page 66
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 66
Page 67
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 67
Page 68
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 68
Page 69
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 69
Page 70
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 70
Page 71
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 71
Page 72
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 72
Page 73
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 73
Page 74
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 74
Page 75
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 75
Page 76
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 76
Page 77
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 77
Page 78
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 78
Page 79
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 79
Page 80
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 80
Page 81
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 81
Page 82
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 82
Page 83
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 83
Page 84
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 84
Page 85
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 85
Page 86
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 86
Page 87
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 87
Page 88
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 88
Page 89
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 89
Page 90
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 90
Page 91
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 91
Page 92
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 92
Page 93
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 93
Page 94
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 94
Page 95
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 95
Page 96
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 96
Page 97
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 97
Page 98
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 98
Page 99
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 99
Page 100
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 100
Page 101
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 101
Page 102
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 102
Page 103
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 103
Page 104
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 104
Page 105
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 105
Page 106
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 106
Page 107
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 107
Page 108
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 108
Page 109
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 109
Page 110
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 110
Page 111
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 111
Page 112
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 112
Page 113
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 113
Page 114
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 114
Page 115
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 115
Page 116
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 116
Page 117
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 117
Page 118
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 118
Page 119
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 119
Page 120
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 120
Page 121
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 121
Page 122
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 122
Page 123
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 123
Page 124
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 124
Page 125
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 125
Page 126
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 126
Page 127
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 127
Page 128
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 128
Page 129
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 129
Page 130
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 130
Page 131
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 131
Page 132
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 132
Page 133
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 133
Page 134
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 134
Page 135
Suggested Citation:"Chapter 3 - Review of Published Literature and Potential Research Needs." National Academies of Sciences, Engineering, and Medicine. 2004. Identification of Research Needs Related to Highway Runoff Management. Washington, DC: The National Academies Press. doi: 10.17226/13791.
×
Page 135

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

57 CHAPTER 3 REVIEW OF PUBLISHED LITERATURE AND POTENTIAL RESEARCH NEEDS As mentioned previously in section 1.4, Research Method- ology, more than 900 of the most relevant annotated cita- tions from the 2,500-plus documents incorporated into the research database were reviewed, and nearly 300 full docu- ments were obtained. During the review process, citations were categorized according to the following broad research areas: Evaluation of Stormwater Control Facilities and Pro- grams, Watershed-Based Approaches, Highway Runoff Char- acterization and Assessment, and Receiving Waters Impact Assessment. After a brief review of some of the major syn- theses of highway runoff/urban stormwater quality research, each of these subsections and subcategories are discussed. Instead of including a discussion of every document in the review, only a selected subset of the most comprehensive documents (i.e., results either were included in the abstract, or the full document was acquired successfully, or both) have been summarized. Nonetheless, all of the categorized docu- ments were considered in identifying potential research gaps and needs. 3.1. BRIEF REVIEW OF RECENT MAJOR SYNTHESES OF HIGHWAY RUNOFF/URBAN STORMWATER QUALITY RESEARCH Several researchers have attempted to compile and sum- marize highway runoff/urban stormwater quality research and data, including BMP evaluation studies and performance data. A few of the most notable efforts are described below. 3.1.1. National Highway Runoff Water-Quality Data and Methodology Synthesis The National Highway Runoff Data and Methodology Syn- thesis (NDAMS) is an effort by the U.S. Geological Survey (USGS) and FHWA to compile and readily make available highway runoff research and guidance information. Three volumes were published recently (July 2003) as a result of the NDAMS: • Volume I—Technical Issues for Monitoring Highway Runoff and Urban Stormwater, FHWA-EP-03-054; • Volume II—Project Documentation, FHWA-EP-03-055; and • Volume III—Availability and Documentation of Pub- lished Information for Synthesis of Regional or National Highway-Runoff Quality Data, FHWA-EP-03-056. Volume I is a compilation of “expert chapters” designed to address different technical issues for monitoring highway runoff and urban stormwater. Volume II provides an over- view of the bibliographic database design, the project, the catalog of available information, the efforts to evaluate avail- able information, the project quality-assurance and quality- control (QA/QC) program, and the directory structure and files on a CD-ROM accompanying the volume. Volume III is a report describing the NDAMS report–review process and summarizes and interprets the results of the metadata review process. As a product of this synthesis, a bibliography of more than 2,600 relevant references with more than 1,300 selected abstracts (or previa—an abstract written by some- one other than then author) and 252 reviewed and classified references were compiled and are available as an online searchable database (http://ma.water.usgs.gov/FHWA/biblio/ default.htm). NDAMS noted that FHWA, USGS, the Environmental Protection Agency (EPA), and many state highway depart- ments and universities have sponsored or conducted research on the nature and impacts of highway runoff on water qual- ity, but a centrally available composite of these data was still lacking, and the existing data present conflicting information. Existing data and studies from FHWA, USGS, state depart- ments of transportation (DOTs), and other sources were com- piled and evaluated to determine whether the information needs of highway managers, practitioners, and researchers are being met and whether this information will meet future needs. The primary goal of the effort was to determine whether the quality of highway runoff and processes contributing to water quality constituents in highway runoff could be char- acterized adequately nationwide, based on published informa- tion. FHWA sought to check the validity of the existing data and procedures to assess and predict pollutant loadings and impacts from highway stormwater runoff as a first step toward indicating whether current guidelines for highway runoff

quality are up-to-date and technically supportable. FHWA wanted a catalog of existing studies and available data as well as indications of the robustness of the data, the sufficiency of the data to characterize pollutant loadings and impacts from highway and urban stormwater runoff, and the changes in atmospheric deposition around the country since the mid- 1980s. To this end, and to assess the suitability of available data to validate runoff quality models developed by FHWA, a catalog of available data and investigations was developed. 3.1.2. International Stormwater BMP Database Probably the most widely recognized resource on BMP effectiveness data is the award-winning International Storm- water Best Management Practices (BMPs) Database at http:// www.bmpdatabase.org/. The database was known formerly as the National ASCE/EPA Stormwater BMP database (the name of the database was changed to recognize and acknowledge data contributors outside of the United States). This database provides access to BMP performance data in a standardized format for roughly 200 BMP studies conducted over the past 15 years. This database is intended to provide a consistent, scientifically defensible set of data on BMP designs and related performance. The database team has made an extensive effort to assess the quality of the data entered for consistency and accuracy. However, in compiling such a large set of data, and because of limits on resources for data QA/QC, the developers acknowledge that some data may contain errors. The database may be searched on or downloaded from the website and also is available on CD-ROM. The database was developed by ASCE’s Urban Water Resources Research Council under a cooperative agreement with EPA. In 2003, new structural BMP storm event and analysis tables were added to the data search page. These tables allow for parameter-specific searches on all structural BMPs analyzed by the project team. A summary of the number of data records currently resid- ing in the BMP database is shown in Table 3-1. The database is relational in design, and Table 3-1 describes the number of 58 records present in various tables within the database. Note that some of the sites in the database include multiple BMPs (e.g., filters and detention), and therefore the total number of sites is less than the total number of BMPs. A summary of data stored in design tables and associated flow and water quality records is provided in Table 3-2. The event monitoring summary information provided in Table 3-2 does not include grab samples or flow measurements and pre- cipitation data that are not associated with measured or cal- culated event mean concentrations. As demonstrated in Table 3-2, retention ponds are the best-populated BMP category in the current database. At the other extreme, some BMPs have only a single site (e.g., infiltration trenches) in the database. A summary of the geographical distribution of BMPs is provided in Table 3-3. Based on analyses of the data stored in the BMP database, the database team has come to the fol- lowing conclusions with regard to evaluating BMP perfor- mance (Strecker et al., 2000): • Substantially more data are needed for many BMP types to be able to explore meaningfully design versus performance. • Removal percentages are not very useful for character- izing performance, unless looked at much more care- fully (e.g., with treatability information). As a result, BMP performance requirements generally should not be specified in terms of percent removal. • Effluent quality among BMPs of the same type is much more consistent then percent removal and is thought to be a better way of characterizing efficiency; although, at an individual site, it is important to test whether the BMP had a statistically significant effect on water quality. • As effluent quality is fairly consistent, some BMP types may have been mischaracterized as less effective because of cleaner influent. For example, BMPs that rely on set- tling as a primary removal mechanism cannot have high percent removals where suspended solids concentra- tions are low in the influent. The influent data in the dry- extended detention ponds in the database are relatively Category Records in Database General Test Site Information 158 Sites Sponsoring and Testing Agencies 60 Agencies Watershed Information 167 Watersheds Nonstructural BMPs Information 28 BMPs Structural BMPs Information 170 BMPs Monitoring Stations 557 Monitoring Stations Precipitation Data 3,396 Precipitation Events Flow Events Monitored 6,563 Flow Events Monitored Water Quality Sampling Event Data 8,588 Water Quality Events Monitored Water Quality Laboratory Analyses 122,265 Analyses Conducted TABLE 3-1 Summary of records found in ASCE/EPA BMP Database (as of August 2002)

cleaner than other BMPs, and therefore they have been reported as achieving lower percent removals. In fact their effluent quality is relatively close to wet ponds and wetlands. • Long-term trends in receiving water quality, coupled with biological assessments, probably would be a much better gauge of the success of the implementation of BMPs, especially on an area-wide basis. Strecker et al. (2003) reanalyzed the data in the currently expanded database and found that in addition to effluent quality, BMPs are best described by their ability to reduce runoff volumes (i.e., how much stormwater can the BMP prevent) and process variable flows (i.e., how much is treated by the BMP). Table 3-4 compares the ability of some BMPs to reduce stormwater runoff volumes. Notice that biofilters and detention basins have the ability to reduce significantly runoff volumes, making them effective controls for reducing 59 overall runoff volumes; wetland channels and basins do not have this ability. Even with the expanded database there are still many limitations: • The data are for storm event means only, making it impos- sible to do intra-storm processes analyses; although, for the sites that have included more detailed data, intra- storm data may be available. Some grab samples are included in the data sets. • Source reports of the data must be obtained by contact- ing the project team and requesting a paper copy of the source information. • Few cost data are available. • The only summary results on the web site are cross- sectional performance data for a selected BMP based on aggregating the individual storm event data for some or all of the events. BMPs Type Number of BMPs in Category, Including Design Information Precipitation Records for BMPs Type1 Flow Records for BMPs Type1 Water Quality Records for BMPs Type1 Detention Basin 24 129 229 4,209 Grass Filter Strip 32 227 385 6,251 Media Filter 30 187 327 6,144 Porous Pavement 5 5 5 55 Retention Pond 33 378 817 14,293 Percolation Trench and Dry Well 1 3 3 21 Wetland Channel and Swale 14 53 113 1,241 Wetland Basin 15 221 681 7,320 Hydrodynamic Devices 16 169 309 6,186 Total 170 1,372 2,869 45,720 1 Only events that included the collection of event mean concentrations have been included in the summary statistics presented in this table. TABLE 3-2 Summary of data by structural BMP type (as of August 2002) State Number of BMPs State Number of BMPs Alabama 13 North Carolina 6 California 41 Ohio 1 Colorado 4 Oregon 3 Florida 24 Texas 19 Georgia 2 Virginia 29 Illinois 5 Washington 20 Maryland 4 Wisconsin 10 Michigan 5 Other 2 Minnesota 7 Total 198 New Jersey 3 TABLE 3-3 Total number of BMPs by state (as of August 2002)

• All of the interpretive results depend on the storm event definition. • The database is not structured to handle time series data efficiently. • Relatively few studies have data on bypass flows versus those treated. The database research team is actively pursuing new data sets compatible with its requirements. To facilitate this com- patibility, the site recommends use of Urban Stormwater BMP Performance Monitoring: A Guidance Manual for Meeting the National Stormwater BMP Database Requirements, of which there have been more than 25,000 downloads to date. In addition to the detailed BMP monitoring and data analy- sis guidance provided in the manual, the guidance manual includes lists of required and recommended data elements, or metadata, that should be reported with BMP performance data, including specific information for individual BMP types. The manual also includes sample BMP metadata forms to ensure that all of the necessary data elements are recorded. References to other BMP performance guidance manuals are included in Appendix B. The database team’s tasks have not included and will not include recommendations of one BMP type over another; however, the team does report on the performance charac- teristics of BMPs based on the entered data and information in the database. Peer-reviewed performance assessment tech- niques are included. The database is intended to provide a data-exchange tool that permits characterization of BMPs based solely on their measured performance using the same protocols for measurements and reporting information. 3.1.3. Caltrans’ New Technology Report One of the best consolidated sources of evaluative infor- mation on BMP technologies, and on new and ultra-urban BMPs in particular, is the California Department of Trans- 60 portation’s (Caltrans) Caltrans’ New Technology Report. The report, issued twice (with revisions, updates, and new infor- mation) in 2003, consolidates and standardizes information on new technologies developed or identified as part of the Cal- trans process for BMP identification, evaluation, and approval. Since 1996, the Caltrans stormwater program has evaluated and approved a wide range of BMPs for use on Caltrans facil- ities and has approved more than 110 separate BMPs. New technologies—including the latest innovations in permanent stormwater treatment and control and the existing technolo- gies used but not selected (approved) previously as a BMP by municipal or DOT stormwater management programs— are evaluated and described for practitioners in a standard- ized format. New technologies are identified from the liter- ature, consultants, manufacturers, regulators, third parties, Caltrans personnel, or through the agency’s formal New Prod- uct Review Process. Fact sheets for new technologies sum- marize constituent removal effectiveness and the advantages and constraints of each type of new technology presented in the report. Many of the new technologies identified are appro- priate for ultra-urban environments. Favorable evaluations of promising BMPs often lead to pilot studies for gathering definitive performance data. So far, there have been 121 full-scale and small-scale pilot studies. Five of the most recent pilot studies addressed low impact development (LID) areas such as bioretention and constructed wetlands, Direct Flow Inclined-Screens Gross Solids Removal Devices (GSRD), Forward Sloping Screens GSRD, and Reverse Sloping Screens GSRD. Austin filters with alterna- tive media and infiltration basins with alternative media are being considered for pilot testing. Successfully piloted tech- nologies may be approved and listed in the statewide man- agement plan (SWMP) as a permanent BMP to be used where applicable by all Caltrans project engineers as part of signif- icant construction and retrofit projects. Also, fact sheets are developed for each newly identified technology that is not approved as a BMP by Caltrans. Each fact sheet presents summary information to be used by Cal- TABLE 3-4 Comparison of average of mean outflow to mean inflow for selected BMP types contained in the BMP Database BMPs Type Mean Monitored Outflow/Mean Monitored Inflow for Events Where Inflow is Greater Than or Equal to 0.2 Watershed Inches Detention Basins 0.70 Biofilters 0.62 Media Filters 1.00 Hydrodynamic Devices 1.00 Wetland Basins 0.95 Retention Ponds 0.93 Wetland Channels 1.00

trans Soil Water Assessment Tool (SWAT) members to evaluate the potential applicability, as well as specific advan- tages and constraints, of a given BMP to various DOT facil- ities, including for design parameters, operations, mainte- nance, treatment effectiveness, and costs. All fact sheets use a standard format to facilitate comparison among various BMP types. Each fact sheet is divided into a standard series of discussion topics. These topics, and the relevant information included under each topic, are discussed below. 1. BMP Description. A description of the BMP is pre- sented at the top of each fact sheet. The description pro- vides a summary of the BMP configuration and a gen- eral overview of the treatment process, how the BMP operates, and considerations that need to be addressed for promoting maximum treatment effectiveness and functionality. 2. Constituent Removal. The relative degree each BMP is able to remove selected groups of constituents from stormwater runoff is provided. The groups of constituents examined were selected based on the likelihood of occur- rence at transportation facilities at levels that would require treatment consideration. For each of the selected constituent groups [total suspended solids (TSS), TDS, total metals and dissolved metals, pathogens/BOD, nutri- ents, litter, and pesticides], a level of confidence in the available data and a general assessment of the BMP’s ability to remove various categories of pollutants are provided. The fact sheets report relative removal effi- ciencies (high, medium, or low) for each of the nine general categories of constituents, derived from the lit- erature review. Constituent removal was quantified by first calculating the average removal percentage for all constituents within a given category (sediment, nutri- ents, pesticides, metals, pathogens, and litter) and then defined using the following criteria: • High: average removal percentage was equal to or greater than 75%, • Medium: average removal percentage was between 40 and 75%, or • Low: average removal percentage was less than or equal to 40%. The fact sheets provide notes with additional infor- mation regarding how the removal assessment was assigned to a given BMP. A caveat is that the level of confidence in the con- stituent removal data found in the literature depended on the type and amount of information. Assessing con- stituent removal from stormwater BMPs is not a precise science. In fact, the National BMP Database project found that percent removals are not an accurate mea- surement of performance (Strecker et al., 2000). Water quality monitoring studies have demonstrated the wide variability in water quality concentrations in storm- 61 water runoff; with more consistent effluent quality from BMPs, percent removal becomes a function of how polluted the inflow is. To ensure that data of the highest quality are pro- duced, storm event monitoring requires that samples be collected according to standard protocols. The criteria applied for defining the confidence level were • High: The information came either from a Caltrans research study or from a study that met the Caltrans QA/QC monitoring protocols. • Medium: Constituent removal rates were established from the results of a scientific monitoring study or studies conducted independently of equipment man- ufacturers, and the BMP technology has a docu- mented history of application for treating stormwater; or the treatment process was a known technology for treating other types of wastewater discharges; or the BMP technology provided no discharge to surface waters under design conditions. Constituent removal was assumed to be 100% removal, although it was recognized that certain large storm events would not receive treatment. • Low: The BMP monitoring program used to quantify the removal percentages and the applied monitoring protocols could not be substantiated. 3. Caltrans SWMP Category. Caltrans has developed the following categories for BMPs: • Category I BMPs: Technology-based pollution pre- vention BMPs to meet the maximum extent practi- cable (MEP) requirements for designing and main- taining roadways and related facilities; – Group A: The BMPs applicable to all maintenance operations, and – Group B: The BMPs used in the design of new facilities or major renovations of existing facilities. • Category II BMPs: Controls to meet BCT/BAT requirements for construction projects; and • Category III BMPs: Treatment BMPs to meet MEP requirements. 4. Key Design Elements. Elements that have been high- lighted by vendors in the literature or as a result of test- ing. Ancillary facilities assumed to be used in conjunc- tion with the new technology also are listed in this section. An example would be including a detention basin downstream of a chemical treatment technology to capture flocculated particles. 5. Schematic Figure. If appropriate, a schematic figure is provided to depict graphically a typical design plan or cross-section, or both, with the major components identified. 6. Capital, Operational, and Maintenance Cost Assess- ments. Assessments pertaining to the costs of building, operating, and maintaining each BMP also are pro- vided, with the level of confidence in the available data and a general assessment of the BMP’s overall costs.

The level of confidence in the costs to build and oper- ate a BMP depends on the type and amount of infor- mation found in the literature. Using the cost informa- tion developed for municipal stormwater programs was not considered by Caltrans to be directly relevant to Caltrans facilities. The right-of-way costs and con- struction costs of major highway transportation proj- ects are typically much greater than the typical sub- urban street or arterial road that might be constructed by a municipal public works department. Furthermore, operations and maintenance costs of facilities along major freeways can be much more expensive than simi- lar municipal facilities because of limited access and the need to provide traffic control. The criteria applied for defining the confidence level of the cost estimates were • High: Unit cost information was available from a facility designed and constructed by Caltrans or a sim- ilar state transportation department. • Medium: Cost information was available from sev- eral similar facilities constructed under municipal stormwater programs. • Low: No cost information was available from a sim- ilar BMP facility that could be verified independently. Construction costs were extrapolated from available pricing information. The cost-effectiveness for each BMP was assessed in terms of its equivalent uniform annual cost (EUAC) relative to a detention basin. A four-quadrant system was used as a tool to rate each BMP. The cost estimates were defined by first calculating the typical range of costs for constructing or operating a BMP on a per acre basis. The acre represented the drainage area served by the BMP. Operation and maintenance costs based on the BMP’s design life were then added. The EUAC for a particular BMP was estimated and then compared qualitatively to that of a detention basin. If the EUAC was higher than a detention basin, it was marked as a higher cost using the quadrant rating key. The benefit of the BMP was evaluated relative to the performance of a typical detention basin. If the constituent removal was greater than that of a detention basin, the BMP was marked as having a greater benefit. 7. Issues and Concerns. Issues and concerns presented information to be considered in maintenance and in project development, with a standard set of topics in each category facilitating comparisons between vari- ous BMPs. Under the maintenance category, the stan- dard topics include • Requirements: summarizes routine maintenance tasks required to keep the BMP functional; • Nuisance Controls: identifies whether the BMP has the potential to create odors, breed mosquitoes, or attract pests; • Traffic Safety: identifies the level of potential traffic control during BMP servicing; and 62 • Staffing/Equipment: identifies the level of staff, and their skills, required to perform the maintenance, as well as any specialty equipment. For the project development category, the topics include • Right-of-Way Requirements: identifies relative space requirements to install the BMP; • Siting Constraints: identifies siting considerations and limitations, such as soil types, slope of the land, dis- tance from existing infrastructure or other natural features, and regulatory requirements; • Design Complexity: identifies major components and equipment requirements and operational controls or limits; and • Retrofit Potential: identifies the potential for retro- fitting existing Caltrans facilities. 8. BMP-Specific Advantages and Constraints. BMP- specific advantages and constraints lists additional advantages and constraints of the BMP that were not covered in the previous sections, including hydrologic characteristics and regionally specific weather condi- tions, experiences from actual installations, and expan- sion of particular points discussed in previous sections of the fact sheet. 9. Sources of Information. Sources of information are provided when appropriate (e.g., vendor contact infor- mation is provided for proprietary technologies). 3.1.4. Urban Wet Weather Flow Literature from 1996 through 2002 Clark et al. (2003) compiled and organized wet weather flow (WWF) literature reviews that were published origi- nally in the annual literature review issues of Water Envi- ronment Research from 1996 through 2002. The document includes approximately 3,350 references from the 7 years alone. Over this 7-year period, the field of urban WWF research expanded dramatically, in part due to increased interest in the United States due to the Clean Water Act (CWA) NPDES stormwater permit program and increased awareness of the seriousness of urban WWFs throughout the world. The document is organized according to the following primary topic categories: characterization, pollution sources, monitoring and sampling, surface-water impacts, groundwater impact, decision-support systems, regulatory policies and financial aspects, and control and treatment technologies. Each section is divided further into subcategories. For instance, highway and other roadway runoff is a subcategory of pollu- tion sources. 3.1.5. Center for Research in Water Resources: Highway Runoff Literature Review Barrett et al. (1995) conducted a literature review that evaluated the impact of highway construction and operation

on surface water quality and on recharge of groundwater aquifers. The types of barriers for containment and reten- tion of sediment and pollutants from runoff and the effec- tiveness of each device were discussed. The report also addressed the quantity and quality of highway runoff from different types of road surfaces, drainage and conveyance systems, and various types of highways. In addition, meth- ods and strategies for the handling and control of highway runoff and effectiveness of pollution control devices were reviewed. 3.1.6. Identification of Research Needs Clearly there is a need to compile major syntheses of stormwater runoff research, such as the major examples pre- sented above and this current research effort. The NDAMS effort, with the generation of the bibliographic database is a good starting point for such a compilation of research litera- ture. However, since only 252 references out of the 1,300 abstracts in the database were reviewed and classified, a more extensive effort appears to be needed. Also, the classi- fication of documents could be extended and refined to include subcategories within the major categories. The International Stormwater BMP Database currently contains primarily BMP design and monitoring data with no direct link to published literature. A research project that attempts to link an extensive bibliographic database (such as a refined NDAMS database) to a water quality and BMP per- formance database (such as the International Stormwater BMP Database) would result in a very useful tool for storm- water practitioners. Such a project would likely require the participation of several state and federal agencies, as well as private organizations, in order to produce a user-friendly database. However, this type of project could be limited only to highway runoff-related studies, substantially reducing the size of the final database and the costs associated with its development. 3.1.7. Primary References Barrett, M. E., Charbeneau, R. J., Collins, E. R. III, Malina, J. F. Jr., Ward, G. H., and R.D. Zuber. A Review and Evaluation of Liter- ature Pertaining to the Quantity and Control of Pollution from Highway Runoff and Construction. CRWR Online Report 95-5. Center for Research in Water Resources (1995) 186 pp. Caltrans. Caltrans New Technology Report. Report CTSW-RT- 03-010 (February 2003) 130 pp. Clark, S., Pitt, R., Field, R., Fan, E., Heaney, J., Wright, L., and S. Burian. Annotated Bibliography of Urban Wet Weather Flow Literature from 1996 through 2002. Compilation of Annual Literature Review Issues of Water Environment Research, www. epa.gov/ednnrmrl/repository/wef_lit_rvws/index.htm (2003) 380 pp. Granato, G. E., Zenone, C., and P. A. Cazenas (eds.). National High- way Runoff Water-Quality Data and Methodology Synthesis, 63 Volume I—Technical Issues for Monitoring Highway Runoff and Urban Stormwater. FHWA-EP-03-054, Prepared by the USGA for the Federal Highway Administration (2003) 479 pp. Granato, G. E., Dionne, S. G., Tana, C. K., and T. L. King. (2003). National Highway Runoff Water-Quality Data and Methodology Synthesis, Volume II—Project Documentation. FHWA-EP-03- 055, Prepared by the USGA for the Federal Highway Adminis- tration (2003) 22 pp. Granato, G. E. National Highway Runoff Water-Quality Data and Methodology Synthesis, Volume III—Availability and Documen- tation of Published Information for Synthesis of Regional or National Highway-Runoff Quality Data. FHWA-EP-03-056, Prepared by the USGA for the Federal Highway Administration (2003) 71 pp. Strecker, E. W., Quigley, M. M., and B. R. Urbonas. A Reassess- ment of the Expanded EPA/ASCE National BMP Database. Proc., National Conference on Urban Stormwater—Enhancing Programs at the Local Level, Chicago, IL (February 17–20, 2003) pp. 555–573. Strecker, E. W., Urbonas, B. R., Quigley, M. M., Howell, J., and T. Hesse. Urban Stormwater BMP Performance Monitoring: A Guidance Manual for Meeting the National Stormwater BMP Database Requirements. Prepared for the U.S. EPA and ASCE, (February 2002) 216 pp. Strecker, E. W., Quigley, M. M., Urbonas, B. R., Jones, J. E., and J. K. Clary. Determining Urban Stormwater Best Management Practice Removal Efficiencies. Final Data Exploration and Eval- uation Report, ASCE/EPA Assistance Agreement Number CX 824555-01 (2000). 3.2. EVALUATION OF STORMWATER CONTROL FACILITIES AND PROGRAMS Over the past 30-plus years, stormwater researchers have evaluated the performance of stormwater BMPs. These eval- uations have come in many forms and permutations. For instance, some evaluations simply may investigate the pollu- tant removal effectiveness of a BMP by monitoring the influ- ent and effluent concentrations or loads, or both, and com- paring results at the storm event, seasonal, or annual scales. More advanced evaluations have included attempting to associate performance with specific site conditions or design variables, evaluating methods to improve pollutant removal in existing drainage systems, and characterizing the water quality achieved rather than the removals. Furthermore, some BMP evaluations may have looked beyond the pollutant removal effectiveness through the use of surrogate perfor- mance measures, such as the hydraulic regimes (hydraulic residence, bypass volume, etc.), retention of previously cap- tured pollutants, maintenance requirements, or even biolog- ical indicators. Because of the increasing use of stormwater BMPs by state DOTs, the evaluation of stormwater control facilities and programs will be an area of increasing interest for DOT stormwater managers. The survey of state DOTs revealed that a large portion of transportation agencies across the country implement stormwater control practices. Table 3-5

shows the number of states that currently implement standard stormwater control practices, and Table 3-6 shows the num- ber of states that currently implement specific controls. Of the nearly 900 documents and abstracts reviewed, the project team identified more than 400 studies that generally evaluated stormwater control facilities and programs. These evaluative studies were further categorized according to the primary subtopic area of the study. As studies that evaluate different BMP types often have similar objectives, the pri- mary subtopic areas were based on the primary objectives of the study rather than on specific BMP types to assist in iden- tifying research gaps and needs with regard to BMP evalua- tions. This section is organized according to the following primary topic areas: • General Evaluation, • Gross Pollutant Removal, • Hydraulic Assessment, • Pollutant Retention, • Methods to Improve Pollutant Removal in Existing Stormwater Systems, • Erosion and Sediment Control, • Design Variables, and • Unit Processes. 3.2.1. General BMP Evaluation Although there are several different ways to evaluate BMPs, the most common methods monitor the effluent and influent water quality. The literature shows significant variation among the methods used to collect and analyze such water-quality data and make inferences about BMP performance. ASCE, in cooperation with the EPA, attempted to develop a standard BMP performance monitoring protocol through the publica- tion of Urban Stormwater BMP Performance Monitoring: A Guidance Manual for Meeting the National Stormwater BMP Database Requirements (GeoSyntec Consultants, 2002). How- ever, the methods and terminology recommended in the man- ual have yet to be adopted widely by the stormwater commu- 64 nity. Many state DOTs have evaluated BMP performance, but without agreed-upon methods and terminology, it is difficult to compare meaningfully the evaluations. The survey of state DOTs revealed that 15 of the respond- ing DOTs have conducted studies or have prepared reports that evaluate the effectiveness and efficiency or performance, or both, of source control or treatment control stormwater management measures at DOT facilities. As more DOTs begin monitoring the effectiveness of their BMPs, more sta- tistical summarizations and comparisons of performance data should be completed; however, this will not be easy if different methods are used to evaluate and report BMP per- formance. There is a need for reaching a consensus on stan- dard BMP performance measures and terminology. As discussed in Guidance Manual for BMP Monitoring, efficiency is a measure of how well a BMP or BMP system removes or controls pollutants, performance is a measure of how well a BMP meets its goals for the stormwater the BMP is designed to treat, and effectiveness is a measure of how well a BMP system meets its goals in relation to all storm- water flows (ASCE/EPA, 2002). In other words, performance and effectiveness are project and site specific, in that they are measures relative to specific goals, such as meeting a prede- termined effluent quality. Efficiency, on the other hand, is an absolute measure, independent of effluent quality expecta- tions. Cost efficiency is not included in any of these mea- sures. However, a systems analysis approach to stormwater management should consider capital investments and opera- tion and maintenance costs when evaluating the overall effi- ciency of various treatment alternatives. Efficiency as defined above will be the only measure discussed herein. In addition to clarifying differences in terminology as dis- cussed above, the primary research questions with regard to measures of BMP performance appear to be • What are the various measures of BMP performance? • What are the positive and negative attributes of each? 3.2.1.1. Historical BMP Efficiency Calculation Methods Many publications reporting efficiency values do not provide adequate information (such as the equation or even the parameter values) to determine the method used. When removal efficiency calculation methods are reported, values can be based on a number of different methods. Historically, the methods used to calculate BMP efficiency include the efficiency ratio (ER), summation of loads (SOL), regression of loads (ROL), mean concentration (MC), efficiency of individual storm loads (ISL), and reference watersheds and before-and-after studies, plus various alterations of the afore- mentioned methods. The efficiency method used most often is the ER method, which has serious shortcomings, as do all of the others listed. Recent research indicates that BMPs are effluent-limited, and the effluent concentration of some BMPs, including extended detention basins, has little to no TABLE 3-5 Categories of stormwater control practices in use at state DOTs Stormwater Quality Practices Used Number of States Using Number of States Indicating that They Do Not Use Temporary erosion soil control 50 0 Permanent stormwater facility 40 10 Stormwater retrofit 17 33 Stormwater monitoring 26 24 Water quality BMPs in operations and maintenance 35 15

dependence on the influent concentration. Therefore, effi- ciency calculation methods using influent concentrations tend to overestimate the efficiency of the BMP when influ- ent concentrations are high and to underestimate the effi- ciency when the influent concentrations are low. The para- graphs below describe briefly the most common and currently accepted methods used to calculate efficiency. A more com- plete description of historically used BMP efficiency calcu- lation methods can be found in Guidance Manual for BMP Monitoring (ASCE/EPA, 2002). Efficiency Ratio Method. As indicated above, the ER method is the most commonly used method and is accepted by vari- ous organizations including the EPA (U.S. EPA, 2002). As opposed to a storm-by-storm estimate, this method tends to minimize the influence of clean influent concentrations at underestimating BMP efficiency by averaging the event mean 65 concentrations (EMCs) of several storms. By definition, the ER is the ratio of the difference between the average EMCs of the inlet and outlet to the average inlet EMC. It can be expressed mathematically as Summation of Loads. The SOL efficiency calculation is used often to evaluate long-term performance of a BMP. Summing loads over a period greater than the residence time reduces the inherent outlet independence to inlet concentrations for detention–storage BMPs. This method, recommended by the Bay Area Stormwater Management Agencies Association (BASMAA, 1996), has been used by Texas DOT to evaluate the treatment effectiveness of the DOT’s highway runoff controls (Keblin et al., 1997). By definition, the SOL is equal ER = −1 average outlet EMC average inlet EMC Innovative Technique or Technology Number of States Using Practice Number of States Not Using Practice Water quality inlets 16 32 Constructed wetlands 32 16 Grassed/vegetated swales and buffer strips 43 5 Wet ponds 31 17 Dry ponds 39 9 Wet vaults/tanks 9 38 Dry vaults/tanks 6 41 Porous/permeable pavement designs 5 42 Oil and water separators 27 21 Silt fences 31 16 Infiltration basin/trench 32 15 Sand filter 15 32 Low impact design 11 36 Hydrodynamic ultra-urban BMPs 9 38 Filtration ultra-urban BMPs 14 33 Natural stream channel design and stabilization 24 22 Herbicide alternatives for roadside vegetation maintenance 23 23 Gross solid separators (trash) 16 30 Dry weather diversion 9 38 Flocculating agents 10 37 *Not all state DOTs responded to every question TABLE 3-6 Stormwater control technologies in use at state DOTs*

to the difference in the sums of the inlet and outlet loads divided by the sum of the inlet loads over a specified time period. Individual loads are calculated by multiplying the EMC by the entire flow volume of the storm. The SOL can be calculated mathematically as Regression of Loads. In this method a least squares linear regression of the influent and effluent loads is conducted, with the regression line constrained at the origin. Percent removal then is defined as the compliment of the slope (β) of the regression line, or mathematically as ROL = 1 − β Over a large range of loadings, there is sufficient evidence to demonstrate that outlet concentrations are not correlated linearly to inlet concentrations. Therefore, this method is not a recommended method. Endorsement of the ROL method could not be found during a review of current literature (i.e., within the last 5 years). Mean Concentration. The MC method generally is not a recommended method; however, when flow-weighted data or storm volumes are not available (such as from grab sam- ples, which are required when sampling for oil and grease), this method may be of some value. Data transfer is not advis- able when this method is employed because of the general lack of storm information. The MC equation is identical to the ER equation except that average outlet concentrations are used instead of average EMCs. Thus, an inherent assumption of this method is that the grab sample is representative of a flow-weighted composite sample. Efficiency of Individual Storm Loads. The ISL method is a ratio of the loads removed to the loads entering a BMP dur- ing a single storm event. The mean ISL efficiency of several individual events is then considered to be the average effi- ciency of the BMP. This method weights all storms equally and does not account for pollutant storage and release during successive storm events. The most serious shortcoming of this method is the assumption of effluent quality dependence on influent quality, particularly when applied to detention storage BMPs that have residence times greater than the storm event duration. Reference Drainage and Before-and-After Studies. These two methods differ slightly; however, the inherent assump- tions are essentially the same—characteristics of the refer- ence drainage, or the study drainage, before BMP imple- mentation are the same as the study drainage with the BMP. The reference drainage method assumes spatial transferabil- ity of drainage characteristics, and the before-and-after method SOL = −1 sum of outlet loads sum of inlet loads 66 assumes temporal transferability of drainage characteristics. Because of site constraints or poorly defined inlet or outlet structures (such as infiltration facilities), these two methods are often the only methods available for analysis. Fairly recent examples of reference drainage studies include those of Legret and Colandini (1999), where the effectiveness of porous pavement at removing heavy metals is evaluated, and of Sansalone (1999), where the effectiveness of a partial exfiltration trench (PET) on a highway shoulder is evaluated. The major difficulty of this method is the large number of parameters that need to be consistent between the two drainages in the reference watershed method or constant in time in the before-and-after studies. 3.2.1.2. Innovative Approaches and Variations of the Historical Approaches In the past few years of stormwater BMP data collection and assessment, the shortcomings of the historical BMP effi- ciency calculation approaches have become more apparent. Also, inconsistent use of the several available methods has led to a wide range of efficiency values for BMPs, as well as inappropriate transfer of data. In an effort to overcome some of these shortcomings, several stormwater quality profes- sionals have proposed alternative methods for calculating efficiency. Most of the newer methods are variations on the historical approaches; however, some innovative methods have been proposed. All of the more recently proposed meth- ods attempt to address the shortcomings of the historical approaches and to increase the transferability of BMP data. During a review of the most recent literature on BMP per- formance monitoring protocols, four promising alternative methods were found. Effluent Probability Method. The effluent probability method was the recommended method in the ASCE/EPA National Stormwater BMP Database Guidance Manual. This method evaluates statistically the influent and effluent EMCs to determine if the differences in concentrations are statisti- cally significant and, subsequently, to discover trends or characteristics in the two data sets by analyzing cumulative distribution functions or standard parallel probability plots. Useful information, such as ranges in influent values that yield the greatest percent removal, is provided by this method. Because of the relatively large amount of analytical infor- mation generated, as well as the relatively more complicated analysis as compared to the historical approaches, BMP effi- ciency estimates may be more difficult to include in a user- friendly database. Instead of a single efficiency value, a range of values at specific influent conditions or graphical plots would need to be reported. Nonetheless, this method pro- vides the most innovative assessment of BMP efficiency and has the potential to generate widely transferable data.

Flow-Dependent Removal Efficiency Method. This method, which is a variation of the ER method, uses partial EMCs. Storm hydrographs are bracketed into storm sampling peri- ods, during which flow-weighted composite samples are col- lected. Partial EMCs are then averaged according to average inflow rates. Therefore, instead of reporting a single effi- ciency value, as in the ER method, the removal efficiency is calculated at various inflow rates. This method has been pro- posed by the Environmental Technology Evaluation Center and David Evans and Associates (EvTEC and DEA, 2000) and endorsed by the Washington State Department of Envi- ronmental Quality (WADOE, 2002). A serious assumption of this method is that the flow rate is essentially steady during the storm sampling period. Also, as with the ER approach, this approach neglects the fact that outflow concentrations are often independent of influent concentrations over the course of a storm event. Minimum Influent Concentration. The concept of this method is similar to the previous method, in that it is an attempt to discretize the efficiency according to influent con- ditions. In this method, influent concentration, instead of flow rate, is used to provide an efficiency ratio with a lower limit on the influent quality. Specific guidance on this method has not been found. However, it was referred to in Storm- water Best Management Practice Demonstration Tier II Pro- tocol for Interstate Reciprocity as an appropriate BMP per- formance claim (Tier II Protocol, 2001). For example, a stormwater BMP performance claim could be “The Model X system can capture and treat the first half-inch, 24-hour storm for a 10-acre runoff area. Under these conditions, a TSS removal rate of 85%, ±5% (at a 95% confidence interval), can be achieved with inflow TSS concentrations greater than 100 mg/L.” Placing a lower limit on influent quality addresses the fact that the efficiency of BMPs tends to decrease as influent concentrations are reduced. Pollutant Flux Ratio. This method is a variation of the SOL method with average event flow rates being used instead of flow volumes. This modification results in a pollutant flux ratio instead of a loads ratio. This method likely would result in similar values as the SOL method, depending on how event flow volumes are calculated. The WADOE guidance docu- ment does not provide any preference to either method (WADOE, 2002). However, the appropriateness of summing flux values from individual storms is questionable. 3.2.1.3. Identification of Research Gaps and Needs Of all the BMP efficiency calculation methods analyzed, the most promising methods are the effluent probability and the minimum influent concentration removal efficiency meth- ods. The former provides greater detail of the actual perfor- mance of a BMP. The latter provides an easier-to-understand and transferable measure of BMP efficiency. Neither method 67 can be used adequately to estimate the efficiency of BMPs without well-defined inlets and outlets, such as infiltration- type facilities or even source controls. As mentioned above, reference watersheds and before-and-after studies have been used to estimate infiltration performance. In other perfor- mance estimates, infiltration is considered to be 100% effi- cient at removing pollutants and is, therefore, based solely on the amount of stormwater infiltrated. This type of efficiency measure is appropriate for some pollutants such as suspended solids; however, it is inappropriate for highly mobile pollu- tants, particularly when groundwater resources are threatened. Alternative estimates could be based on changes to ground- water quality or in soil concentrations. For large infiltration operations, efficiency could be based on changes to local groundwater quality. For small infiltration operations, such as roadside ditches, calculating the percent removal before the introduction to the groundwater would be necessary. Changes in soil concentrations would provide an idea of the pollutant removal; however, the transport and chemical trans- formations of the pollutant of concern would need to be assessed adequately. Based on the review of the many methods for evaluating and reporting the efficiency of BMPs, consensus and clear guidance are needed. 3.2.1.4. Primary References GeoSyntec Consultants. Urban Stormwater BMP Performance Mon- itoring: A Guidance Manual for Meeting the National Storm- water BMP Database Requirements. American Society of Civil Engineers and U.S. Environmental Protection Agency (Febru- ary 2002). EOA, Inc. Guidance for Monitoring the Effectiveness of Stormwater Treatment Best Management Practices. BASMAA Report (1996). EvTEC and DEA. EvTEC Evaluation Plan for Ultra-Urban Stormwater Technologies. WA Department of Transportation (March 2000). Keblin, M. V., Barrett, M. E., Malina Jr., J. F., and R. J. Charbeneau. The Effectiveness of Permanent Highway Runoff Controls: Sedi- mentation/Filtration Systems. CRWR Online Report 97-4, http:// www.ce.utexas.edu/centers/crwr/reports/online.html (1997). Legret, M., and V. Colandini. Effects of Porous Pavement with Reservoir Structure on Runoff Water: Water Quality and Fate of Heavy Metals. Water Science and Technology, Vol. 39, No. 2 (1999) pp.111–117. Sansalone, J. J. In Situ Performance of a Passive Treatment System for Metal Source Control. Water Science and Technology, Vol. 39, No. 2 (1999) pp. 193–200. Tier II Protocol. Stormwater Best Management Practices Demon- stration Tier II Protocol for Interstate Reciprocity. Endorsed by California, Massachusetts, New Jersey, Pennsylvania, and Vir- ginia (August 2001). U.S. Environmental Protection Agency. ETV Verification Protocol Stormwater Source Area Treatment Technologies. Draft Report 4.1, Washington, DC (March 2002). Washington State Department of Ecology. Stormwater Treatment Facility Performance Evaluation Guidance Document. Draft

Report, Washington State Department of Environmental Quality (April 2002). 3.2.2. Gross Pollutant Removal The subtopic of gross pollutant removal generally refers to stormwater treatment facilities or programs designed to reduce the amount of trash, debris, or large sediments discharged from constructed road surfaces or storm drain systems. Ero- sion controls, which are designed to hold soils in place either temporarily during construction or permanently on cut–fill slopes, will be addressed in a separate subsection. Some com- mon research questions with regard to gross pollutant removal include • How effective are source controls—such as street sweep- ing, catch basin cleaning, and public education—at reducing trash and debris in transportation facility runoff? • How effective are trash racks, screens, or other GSRD at removing and retaining bulk pollutants? Several state DOTs have participated in recent studies that try to answer these research questions. Caltrans has taken a lead role with impetus from new TMDLs for trash; however, survey results showed only 16 DOTs use gross solid separa- tor devices. The surprisingly low implementation of this sim- ple treatment technology indicates a potential need for edu- cation and outreach in this area. To investigate the characteristics of litter in freeway storm- water and the effectiveness of various BMPs, Caltrans con- ducted a 2-year litter management pilot study in the Los Angeles area (Lippner et al., 2001). New litter sampling and monitoring protocols were devised to characterize litter and to test BMP effectiveness. Twenty-four freeway catchments were monitored. Half of the catchments were treated with one of five BMPs; the others were controls. Tested BMPs included increased street sweeping frequency [the results of which were included in Lippner and Moeller (2000), dis- cussed at the end of this section], increased frequency of manual litter pick up, a modified drain inlet, a bicycle grate inlet, and a litter inlet deflector developed during the study. Litter discharges were quantified by weight, volume, and count and were further classified by composition. Roughly half the freeway stormwater litter was found to consist of paper, plastic, and Styrofoam. With the exception of cigarette butts, the origins of most litter items could not be identified because of their small size. Of the five BMPs tested, only increased litter pick up and the modified drain inlet demon- strated some reduction of litter in stormwater runoff, although the data are highly variable. Increased frequency of sweeping, the bicycle grate, and the litter inlet deflector did not effec- tively reduce litter in stormwater discharges monitored dur- ing the study, although the trash bags placed on the outfall to evaluate what trash was leaving the system were effective. 68 Street-sweeping efficacy studies have been conducted by several researchers with variable results. One of the conclu- sions of the EPA-sponsored Nationwide Urban Runoff Pro- gram (NURP)—in which more than $30 million was expended in an intensive 3-year investigation of urban runoff quality at 28 locations throughout the United States (U.S. EPA, 1983)—was that street sweeping was generally an ineffective technique for improving the quality of urban runoff. Similar results were reached by Pitt (1979) and Pitt and Shawley (1982). Nevertheless, recent studies suggest that street sweep- ing programs can be optimized to reduce significantly pollu- tant washoff from urban streets (Sutherland and Jelen, 1997) Lippner and Moeller (2000) conducted a paired watershed study to evaluate how end-of-pipe litter discharges were affected by street sweeping frequency and the type of sweeper used. The study included field-tests of vacuum, regenerative air, and high-efficiency and mechanical broom sweepers to determine which sweeper type would be most appropriate for sweeping frequency analysis. Results of the tests found that while the high-efficiency and regenerative air sweepers left the pavement cleaner than the broom sweepers, large material often was lodged in the air intake hoses of regenerative air sweepers or was pushed in front of the suc- tion head of the air machines rather than being sucked up. Also, Caltrans was concerned that the maximum operating speed of the high-efficiency sweeper precluded it from being used in freeway applications; thus, the agency chose a broom sweeper (Mobil model M-8A) for the sweeping frequency study. The analysis indicated that litter reduction from sweep- ing monthly as compared to weekly was not statistically sig- nificant at the 95% confidence level. Analysis of conven- tional water quality constituents such as metals, nutrients, oil and grease, TSS, and coliform bacteria showed that increas- ing sweeping from monthly to weekly actually may have increased the concentrations of hardness, total and dissolved copper, dissolved nickel, and total petroleum hydrocarbons (diesel). The cause of this is unknown, however, it could be due to the abrasive action of the sweeper on the road surface, the pollutant sorption ability of street litter no longer available once removed, or simply the random variability of the data. In another study, Smith (2002) evaluated the street sweep- ing effectiveness of mechanized street sweepers for particu- late removal. The first mechanized street sweeping had no observable effect on subsequent storm loads of suspended sediment. Following the second sweeping, a net increase of the suspended-sediment load was observed at one station, and a net decrease of the suspended-sediment load was observed at the second station; however, these effects were only temporary. The highway was swept a third time after continuous monitoring was terminated. The particle-size dis- tribution in sweeper samples for the size fraction <4 mm in diameter was similar to the particle-size distribution in bot- tom sediment in the catch basin. The concentration of parti- cles >0.5 mm in diameter was higher in sweeper samples than in samples from the oil–grit separators, allowing for the

conclusion that the sweepers were successful in removing the larger particles. The Wisconsin DOT Bureau of Highway Operations con- ducted a research project to study the effectiveness of a high- efficiency street sweeper used on an urban freeway section to control the quality of stormwater runoff from the pave- ment surface (Martinelli et al., 2002). The research process used a paired basin approach on a test section that was swept once per week and on a control section that was not swept during the study period. The results of the study indicated with a 90% confidence interval that there was a difference of 1% and 280% in suspended-sediment concentration (SSC) between the control and test sites. This upper limit indicates that the control site may have had higher average baseline conditions than the test site, which is one of the problems with paired watershed studies. To eliminate the spatial variability, a before-and-after study could have been conducted (though this approach introduces temporal variability). Alternatively, a calibrated simulation model—such as the Simplified Particulate Transport Model (SIMPTM) developed by Sutherland and Jelen (1993) or the Source Loading and Management Model (SLAMM) devel- oped by Pitt and Voorhees (2002)—could have been used. These models have been calibrated and applied successfully by some researchers to estimate loads and concentrations, as well as to evaluate the BMP effectiveness, including street sweeping. Please refer to section 3.2.10, BMP Modeling, for descriptions of studies that have calibrated and successfully applied these models. Catch-basin cleaning is considered a source control BMP designed to reduce the potential for stormwater bypass and resuspension of previously captured pollutants and subsequent discharge to receiving waters. Dammel et al. (2001) conducted the Drain Inlet Cleaning Efficacy (DICE) Study for Caltrans to evaluate whether catch-basin cleaning improves the water quality of highway stormwater runoff. The runoff water qual- ity was monitored and analyzed to determine any difference in water quality between stormwater discharge from a drain- age system with clean drain inlets versus discharge from unclean systems. Results from 4 years of monitoring have not indicated a statistically significant difference between cleaned and uncleaned catchments for all 21 monitored cases. The DICE Study is continuing with additional sampling sites and with the sampling of litter and other macro debris from the flow stream added to the list of monitored constituents. As additional data become available, efforts will be made to deter- mine if cleaning drain inlets has a measurable impact on the water quality of effluent emanating from Caltrans freeways. Another Caltrans study tested three nonproprietary in-line devices that could be incorporated into existing or future highway drainage systems to remove trash from stormwater discharges, subsequently meeting the waste load allocation of the trash TMDL (Endicott et al., 2002). The pilot study included conceptual design of trash removal devices, site selection, development of device design criteria, construc- 69 tion, monitoring, and assessment of the performance of each device. The peak runoff generated by a 25-year storm event was set as the minimum hydraulic design criteria for the pilot GSRD. Design criteria to address operation and maintenance concerns included adequate parking and access for mainte- nance and monitoring vehicles; no lane closures for servic- ing or monitoring a device; minimized shoulder closures for major device maintenance activities; maintenance equipment, limited to equipment commonly available in the Caltrans maintenance fleet; and an annual maintenance cycle for removal of accumulated gross solids. For this study, device effectiveness was defined as the per- centage of total litter captured by the device. The pilot GSRD removed a combination of gross solids, including solids, veg- etation, and litter. Removal efficiencies for gross solids ranged from approximately 82 to 100% on a wet mass (weight) basis and from approximately 55 to 100% on a wet volume basis. Removal efficiencies for litter ranged from approxi- mately 66 to 100% on a dry mass (weight) basis and from approximately 66 to 100% on a dry volume basis. Key findings from this pilot study include the follow- ing: (1) GSRD are sensitive to gross solids loading rates; (2) design loading rates must consider total gross solids (solids, vegetation, and litter), because the simple screening technologies used in these devices do not automatically seg- regate the litter component regulated under the TMDL from overall gross solids; (3) litter is a relatively small component of gross solids on both a total mass and total volume basis; (4) gross solids loading rates require further study to define the average and range of expected values; (5) screen blind- ing, and subsequent bypass, is the most common cause for a device to exhibit a low level of effectiveness for litter removal; and (6) gross solids storage and screen blinding prevention must be considered individually during design. Virtually every municipality and several state DOTs have public education and outreach programs that discourage lit- tering. However, the effectiveness of this type of source con- trol is difficult to evaluate and therefore done rarely. Caltrans has embarked on an extensive Public Education Litter Mon- itoring Study (PELMS) to implement and assess a public education program targeted at reducing stormwater litter pol- lutants (Caltrans, 2002). Public education media rollout for the Public Education Resource Study (PERS) started in mid- February 2002, so the effectiveness of the program has yet to be determined. Stormwater litter monitoring is one of several methods that will be used to gauge public education effec- tiveness. Other methods include public opinion surveys and an assessment of roadside litter collection before-and-after program implementation. 3.2.2.1. Identification of Research Needs Based on the literature review pertaining to gross pollutant stormwater control facilities and programs, litter removal using solids separation devices has been demonstrated by a

number of researchers, and the effectiveness of these devices at removing large particles (>5 mm) is well documented. A smaller amount of literature is available on the effectiveness of source controls such as public education and catch-basin cleaning, particularly with regard to the overall effects of catch-basin bypass. The effectiveness of street sweeping technologies has not been demonstrated clearly, even with numerous studies, but it appears that mechanical sweepers may be better at removing larger pollutants, and air machines may be better at removing fine particulates. The likely cause for street sweeping studies being inconclusive is that the overall reductions in runoff loads and concentrations caused by street sweeping are rela- tively small in comparison to the high degree of noise or vari- ability of the data. Such noisy data require many more sam- ples to detect differences than are collected typically. Most studies do not have the resources to collect and analyze this many samples. One potential research need is identification of a uniform definition of gross solids (and the related components) for the purpose of standardizing data reporting. There is an ASCE/ EWRI committee working on this issue, but protocols devel- oped for highway situations may be appropriate to help stan- dardize BMP performance. 3.2.2.2. Primary References Caltrans. Caltrans Public Education Litter Monitoring Study 2001–2002. Preliminary Report CTSW-RT-02-021 (2002). Dammel, E. E., Berger, B. J., Regenmorter, L. C., and G. S. Lippner. Evaluating Drain Inlet Cleaning as a Storm Water Best Manage- ment Practice. International Water Association 5th International Conference, Milwaukee, WI (June 10–15, 2001). Endicott, J. D., Berger, B. J., and S. S. Stone. Design and Perfor- mance of Non-Proprietary Devices for Highway Runoff Litter Removal. Global Solutions for Urban Drainage, 9th International Conference on Urban Drainage (September 8–13, 2002). Lippner, G. S., and G. Moeller. Study Quantifies Broom Sweeper Litter Pickup Ability. American Sweeper, Vol. 8 (2000). Lippner, G. S., Johnston, J., Combs, S., Walter, K., and D. Marx. Results of California Department of Transportation Litter Man- agement Pilot Study. In Transportation Research Record 1743, TRB, National Research Council, Washington, DC (2001) pp. 10–15. Martinelli, T. J., Waschbusch, R., Bannerman, R., and A. Wisner. Pollutant Loadings to Stormwater Run-Off from Highways: The Impact of a Sweeping Program. Wisconsin Department of Trans- portation, Division of Transportation Infrastructure Develop- ment, Bureau of Highway Operations (2002) 94 pp. Pitt, R. E. A Demonstration of Nonpoint Pollution Abatement through Improved Street Cleaning Practice. Final Report, U.S. Environmental Protection Agency (1979). Pitt, R. E., and G. Shawley. A Demonstration of Nonpoint Pollu- tion Management on Castro Valley Creek. Report, Alameda County Flood Control and Water Conservation District, Hay- ward, CA (1982). Pitt, R., and J. Voorhees. http://www.winslamm.com (2002). 70 Smith, K. P. Effectiveness of Three Best Management Practices for Highway-Runoff Quality along the Southeast Expressway, Boston, Massachusetts. Water-Resources Investigations Report, U.S. Geo- logical Survey (2002) 170 pp. Sutherland, R. C., and S. L. Jelen. Simplified Particulate Transport Model—Users Manual, Version 3.1. (1993). Sutherland, R. C., and S. L. Jelen. Contrary to Conventional Wis- dom: Street Sweeping Can Be an Effective BMP. Advances in Modeling the Management of Stormwater Impact, Vol. 5 (1997). U.S. Environmental Protection Agency. Final Report of the Nation- wide Urban Runoff Program. Water Planning Division, Wash- ington, DC (1983). 3.2.3. Hydraulic Assessment One method of evaluating the performance and applica- bility of a stormwater control facility is to analyze the flow rates and volumes of stormwater into, within, and out of a facility. Hydraulic residence times and the volumes of over- flow, or bypasses, often are used as sizing criteria before BMP construction and as performance measures after con- struction. The degree of short-circuiting, which is a function of BMP design, has a direct effect on the hydraulic residence time within a stormwater BMP. The hydraulic conductivity, or flow-rate capacity, for flow-based BMPs and the design volume for detention-based BMPs have a direct effect on the bypass or overflow volume for a given rainfall-runoff event. Assessment of these hydraulic phenomena sometimes is con- sidered when evaluating the performance of individual BMPs. Other times, researchers have reported only the treated efflu- ent, but not the amount that was bypassed. In evaluating the overall performance of a stormwater management program within a watershed, which may include a combination of nonstructural source control and structural treatment control BMPs, it often is desirable to evaluate the amount of water stored and released slowly, evapotranspired, or infiltrated throughout the watershed. Thus, this distributed BMP approach was coined LID (see section 3.2.9). BMP design guidelines and criteria frequently include recommended or required hydraulic residence times. For instance, the City of Portland BMP design manual specifies that the outlet of stormwater quality ponds be designed such that the pond drains to the permanent pool volume in 12 hours (Woodward-Clyde, 1995). For extended detention, which is believed to provide a higher level of treatment, the hydraulic residence time may be increased by 24 to 48 hours for deten- tion facilities. Hydraulic residence time also is sometimes a design criterion for other BMP types, such as vegetated swales, where it is recommended that stormwater be in con- tact with biofiltration media 5–9 minutes (Water Environ- ment Foundation and ASCE, 1998). Despite these recom- mended design parameters, the relationship between BMP performance and hydraulic residence times, as well as other hydraulic characteristics of BMPs, is not understood clearly. As such, some of the common research questions with regard

to the hydraulic assessment of individual stormwater control facilities include • How do hydraulic residence times and bypass volumes relate to BMP performance? • What design variables influence hydraulic residence times and short-circuiting? • What methods are available to evaluate, improve, and maintain stormwater infiltration? • What is the potential for storage and reuse of urban stormwater? 3.2.3.1. Hydraulic Residence and Bypass A couple of studies have shown that typically the longer the hydraulic residence time, the better the overall pollutant removal performance (Kulzer, 1989; Driscoll, 1986). With regard to particulate settling, Galli (1992) noted that several researchers have found that a large portion of suspended par- ticulates (30–70%) settle out within the first 6–12 hours of detention. Fine silt and clay-sized particles settle out over a much longer period, on the order of days and weeks. With regard to increasing the hydraulic residence within stormwater BMPs, Newberry and Yonge (1996) studied the factors that influence the hydraulic residence on highway grass strips. They found that a change in flow rate has a greater effect on hydraulic residence than an equivalent per- cent change in slope and that a lower degree of soil com- paction would allow for more subsurface flows and a longer average hydraulic detention time. A tracer study by Price and Yonge (1995) found that the installation of a baffle at the inlet of a detention basin would reduce short-circuiting and increase hydraulic residence. Increased sediment and adsorbed metals removal also were observed. As a flood-control precaution or to protect the BMP, water quality BMPs often are designed to bypass stormwater runoff that exceeds their design capacity. Sometimes, bypass occurs unintentionally when filter media or inlet structures become clogged or blocked. Ultra-urban BMPs—such as storm drain inlet filters, oil–grit separators, and infiltration facilities—are the most susceptible to bypass. The overall effect of bypass on BMP performance, and ultimately receiving water qual- ity, is relatively unknown. Consequently, several BMP per- formance evaluation protocols require an evaluation of BMP performance with the inclusion of bypass volumes. Very few studies were found that evaluated the effects of bypass on BMP performance. A study by Greb et al. (1998) evaluated the effects of bypass in a Stormceptor® and a multi-chamber treatment train (MCTT) installed in public works maintenance yards. For the Stormceptor, 11 out of 45 storms bypassed the unit; the total water volume that bypassed equaled approximately 9%. The reported percent removal including the bypass vol- ume was 22% for TSS, as compared to 25% when excluding the bypass volume. For the MCTT, the overall TSS removal 71 efficiency was impossible to measure directly because of the water loss problems. However, it was estimated that the over- all percent removal of TSS including bypass was 78%, as compared to 98% when excluding bypass. Keblin et al. (1997) studied the effects of bypass on the removal efficiency of the Texas DOT’s Seton Pond facility in Austin, which includes a sedimentation basin and a sand filter. Of the 10 storms that were monitored, the authors observed that 20% of the total runoff volume bypassed the facility, resulting in the total TSS load removal efficiency being reduced from 89% excluding bypass to 71% including bypass. 3.2.3.2. Hydraulics of Infiltration Facilities Rather than maximizing residence time, infiltration facili- ties usually have the goal of maximizing percolation rates. Common types of infiltration facilities include porous pave- ment, infiltration basins and trenches, sand filters without an underdrain, and PETs. Sand filters or other media filters that have an underdrain are not considered infiltration facilities. Infiltration practices are one of the most valuable urban storm- water BMPs, because they help to reduce not only storm- water pollutants but also stormwater volume, which increases groundwater recharges and reduces the potential for scour and bank erosion in receiving waters. Livingston (2000) provides a comprehensive review of the successes and failures of stormwater infiltration, a summary of the lessons learned about the use of infiltration practices, and a list of recommen- dations of when and how they should and should not be used. The review of highway stormwater literature revealed sev- eral studies that evaluated the hydraulics of porous pavement (Nawang et al., 1993; Goforth et al., 1984; Dempsey and Swisher, 2003; Bond et al., 1999; Pratt et al., 1995; Wada et al., 1997; and Backstrom and Bergstrom, 2000). The model simulation study by Wada et al. (1997) found that the construction of permeable pavements with infiltration pipes (a perforated pipe within a gravel bed beneath the pavement) significantly increased the percolation rate of the pavement. The study by Backstrom and Bergstrom (2000) that evalu- ated the hydraulics of porous pavement in cold climates found that when porous asphalt was exposed alternatingly to melting and freezing over a 2-day span (conditions similar to the snowmelt period), the infiltration capacity was reduced by approximately 90%. Based on the results of this study and previous studies, the infiltration capacity of porous asphalt was estimated to be 1–5 mm/min for snowmelt conditions. These results have serious implications with regard to the use of porous pavements in cold climate areas. An increasing concern, especially with the implementation of EPA’s UIC regulations is that if infiltration rates are too high, many pollutants could be introduced to groundwater. This issue is addressed in section 3.4, Highway Runoff Char- acterization and Assessment.

3.2.3.3. Identification of Research Needs Based on the literature review addressing the hydraulic assessment of stormwater control facilities in relation to BMP performance, the most pressing gaps appear to be in the eval- uation of the characteristics and effects of short-circuiting and bypass or overflow (e.g., ponds or wetlands discharging over the low-flow outlet or bioswales when depths and velocities for good treatment are exceeded). The influence of hydraulic residence time on BMP performance has been studied well, and it has been confirmed that detention time is correlated positively with pollutant removal (at least for particulate- bound pollutants). However, no studies were found that inves- tigated the nature of the correlation (linearly, asymptoti- cally, and others). Also, hydraulic residence usually is calculated simply by dividing the permanent pool volume by the average outflow discharge rate of a BMP. The true hydraulic residence time depends on the flow path through the system, which requires some means of estimating the velocity field of the system, such as the use of tracers, ultra- sensitive velocity meters, or two-and three-dimensional hydrodynamic models. 3.2.3.4. Primary References. Backstrom, M., and A. Bergstrom. Draining Function of Porous Asphalt During Snowmelt and Temporary Freezing. Canadian Journal of Civil Engineering, Vol. 27(3) (2000) pp. 594–598. Bond, P. C., Pratt, C. J., and A. P. Newman. A Review of Storm- water Quantity and Quality Performance of Permeable Pave- ments in the UK. Proc., 8th International Conference on Urban Storm Drainage (1999) pp. 248–255. Dempsey, B. A., and D. M. Swisher. Evaluation of Porous Pave- ment and Infiltration in Centre County, Pennsylvania. Water World & Environmental Resources Congress 2003, Philadelphia, PA (June 2003). Driscoll, E. D. Detention and Retention Controls for Urban Runoff. In Urban Runoff Quality: Impact and Quality Enhancement Technology, ASCE (1986) pp. 381–393. Galli, J. Analysis of Urban BMP Performance and Longevity in Prince George’s County, Maryland. Final Report, Prince George’s County Department of Environmental Resources, Watershed Pro- tection Branch, MD (1992) 203 pp. Goforth, G. F., Diniz, E. V., and J. B. Rauhut. Stormwater Hydro- logical Characteristics of Porous and Conventional Paving Sys- tems. U.S. Environmental Protection Agency, Washington, DC (1984) 302 pp. Greb, S. R., Corsi, S., and R. Waschbusch. Evaluation of Storm- ceptor® and Multi-Chamber Treatment Train as Urban Retrofit Strategies. Proc., National Conference on Retrofit Opportunities for Water Resource Protection in Urban Environments, Chicago, IL (February 9–12, 1998). Keblin, M. V., Barrett, M. E., Malina Jr., J. F., and R. J. Charbeneau. The Effectiveness of Permanent Highway Runoff Controls: Sedimentation/Filtration Systems. CRWR Online Report 97-4, Center for Research in Water Resources, Bureau of Engineering Research, The University of Texas at Austin, (1997) 126 pp. 72 Kulzer, L. Considerations for the Use of Wet Ponds for Water Qual- ity Enhancement. Office of Water Quality, Municipality of Met- ropolitan Seattle (1989) 90 pp. Livingston, E. H. Lessons Learned About Successfully Using Infil- tration Practices. Proc., of National Conference on Tools for Urban Water Resource Management and Protection, Chicago, IL (February 7–10, 2000) pp. 81–96. Nawang, W. M., and S. Saad. Stormwater Infiltration Investigation Using Porous Pavement. Proc., 6th International Conference on Urban Storm Drainage, Niagara Falls, Canada, Vol. I, pp. 405–410. Newberry, G. P., and D. R. Yonge. Retardation of Heavy Metals in Stormwater Runoff by Highway Grass Strips. Washington State Department of Transportation, Olympia, WA (1996) 81 pp. Pratt, C. J., Mantle, J. D. G., and P. A. Schofield. Research into the Performance of Permeable Pavement, Reservoir Structures in Controlling Stormwater Discharge Quantity and Quality. Proc., 2nd International Conference on Innovative Technologies in Urban Storm Drainage (1995) pp. 337–344. Price, F. A., and D. R. Yonge. Enhancing Contaminant Removal in Stormwater Detention Basins by Coagulation. In Transportation Research Record 1483, TRB, National Research Council, Wash- ington, DC (1995) pp. 105–111. Wada, Y., Miura, H., Tada, R., and Y. Kodaka. Evaluation of an Improvement in Runoff Control by Means of a Construction of an Infiltration Sewer Pipe under a Porous Asphalt Pavement. Water Science and Technology, Vol. 36(8–9) (1997) pp. 397–402. Water Environment Foundation and American Society of Civil Engineers. Urban Runoff Quality Management. Manual and Report, WEF Manual No. 23; ASCE Manual, Report and Engi- neering Practices No. 87 (1998). Woodward-Clyde. City of Portland Stormwater Quality Facili- ties—A Design Guidance Manual. Report, Environmental Ser- vices, City of Portland, Clean Water Works (1995). 3.2.4. Pollutant Retention Pollution retention is another import criterion for evaluating the performance of stormwater quality control facilities. Dur- ing large storm events, pollutants may be flushed out of sedi- mentation systems, particularly in-line systems such as catch basin sumps, and be discharged inadvertently into receiving waters. Changes in water chemistry also may have an effect on pollutant mobility. For instance a decrease in pH or a change in oxidation-reduction potential, or both, may cause solid-phase pollutants to become soluble, and therefore mobile. Dry weather flows into BMPs may have a different chemistry, or the BMP, through biochemical processes, may alter water chem- istry to the point that pollutants are released and remobilized. Resuspension of pollutants within stormwater BMPs can cause the BMPs to be a source of pollutants, which may translate into negative percent removals in BMP evaluation studies if mobi- lization occurs during storm events. If pollutants are mobilized via dry weather flows, BMP studies that focus on stormwater event monitoring alone would not detect this. Potential research questions with regard to pollutant reten- tion are • What is the potential for resuspension of previously cap- tured sediment?

• What conditions influence pollutant mobility in BMP systems, and how can these conditions be reduced? • What need is there for more continuous monitoring of wet BMPs to assess the potential for pollutant remobi- lization between storm events? • How sequestered are captured pollutants in BMPs? Smith (2002) investigated the sediment retention of oil–grit separators and a deep-sumped catch basin. Despite the pres- ence of bypass pipes at the inflows to the separators and the fact that the depth of bottom sediment retained in the catch basin was less than 25% of the sump depth, previously cap- tured sediments from the separators and the catch basin sump were resuspended during several monitored storm events. For the separators, resuspension of sediments was detected at and above rainfall intensities of 0.04 in. per 5-min interval and flows >0.46 ft3/s. The amount of resuspended sediment estimated for the separators represented about 8% of the suspended-sediment loads retained at the end of the monitor- ing period. The estimated quantity of suspended sediment that bypassed the separators was 16–20% higher than the amount of sediment resuspended (<0.062 mm in diameter). For the catch basin sump, the frequency of cases in which resuspension was detected did not increase with an increase in captured sediment. The estimated amount of resuspended sediment represented 18% of the final mass of retained sediment in the sump. Results of experiments conducted by Clark et al. (2001) to determine if four potential filter media (sand, activated car- bon, peat moss, and compost) could retain previously trapped pollutants indicated that permanent retention of heavy met- als (copper, lead, iron, and zinc) may occur even in an anaer- obic environment. However, retention of nutrients may not occur under these conditions. In a BMP performance study by Yu and Stopinski (2001), four ultra-urban BMPs—three oil–grit separators (Isoilater, Stormceptor, and Vortechs Stormwater Treatment System) and a bioretention area—were evaluated. Monitoring results indicated that resuspension of sediment from the bioretention occurred during three of the larger monitored storm events, presumably because of minimal vegetation establishment before the study’s onset. Negative removals for total nitro- gen also were observed in three events of the Stormceptor monitoring. However, these events did not correspond to large events. In fact, the largest negative removal occurred during the smallest storm event. The authors hypothesized that the negative total nitrogen removals were due to a decrease in the amount of aeration inside the BMP, which would limit the oxidation of ammonia. Analysis of accumu- lated sediment depths in the oil–grit separators showed that the Isoilator unit lost captured sediment during 5 out of 15 storm events, with the highest loss of sediment (21.8 cm) occurring during the largest monitored rain event (87.6 mm on 3/31/00). The observed sediment depth never reached the manufacturer’s recommended clean-out depth (34.5 cm and 73 43.2 cm, respectively). Similarly, sediment accumulations in the Stormceptor and Vortechs units were monitored. The Stormceptor unit showed consistent accumulation, except 10.2 cm were lost during the large 87.6 mm rainfall event. The Vortechs unit did not show sediment accumulation dur- ing the study, which was attributed to the unit not being installed properly. Because of the increased use of porous pavement systems in LID designs, the pollutant retention capacity of porous pavement is of particular interest to stormwater managers. A study by Dierkes et al. (2002) investigated the pollutant reten- tion capabilities of four different systems of paving stones: pavers with infiltration joints, porous concrete pavers with a filter-layer, greened (grass) porous pavers, and pavers with greened infiltration joints. All four systems showed very high pollution retention capacities for cadmium, copper, lead and zinc, but the greened systems and the porous pavers were more efficient than the system with the infiltration joints. Copper and lead were retained more effectively than cad- mium and zinc in all of the pavement systems. In another study, the pollutant retention of the subbase of porous con- crete pavers was investigated. Differences in pollution reten- tion capacities between the subbase materials existed, with the highest pollutant retention capacities being reached by crushed stones with high contents of CaCO3. Overall, the pH in the porous concrete effluent of all system configurations showed that the buffering capacities of concrete are very high, so there is little danger of a mobilization of previously captured metals from porous concrete paving systems. 3.2.4.1. Identification of Research Needs When evaluating the performance of stormwater control facilities, it is important to consider not only the pollutant removal capacity under a variety of hydrological and influ- ent quality conditions but also the pollutant retention capac- ity over long time periods and under both storm and low-flow conditions. Few studies were found that investigated the potential for leaching or resuspension of previously captured pollutants. The studies that were found indicate that resus- pension of sediments in catch basin sumps and oil–grit sep- arators may be significant. Resuspension also may occur in bioretention areas before complete establishment of vegeta- tion. Heavy metals do not appear to go easily into the dis- solved phase once captured, but nutrients do, particularly if there is a change in the oxidation-reduction potential. The pH of the stormwater likely has some effect on the solubility of captured metals; however, concrete and other construction materials containing high concentrations of CaCO3 have a high buffering capacity and tend to raise the pH of storm- water on contact. Therefore, a slight decrease in the pH of rainwater is not expected to cause a substantial increase in dissolved metals concentrations, especially if the stormwater flows over or through porous concrete.

This is a research topic area requiring a more detailed lit- erature review before substantiating a research need, but it is more appropriately discussed under the section Highway Runoff Characterization and Assessment. For the subtopic of pollutant retention, it appears that the primary research needs and gaps are in identifying the conditions—such as pH, oxidation-reduction potential, hardness, and organic con- tent—that affect desorption or dissolution, or both, of cap- tured pollutants in stormwater treatment systems. 3.2.4.2. Primary References Clark, S., Pitt, R., and P. Brown. Effect of Anaerobiosis on Filter Media Pollutant Retention Linking Stormwater BMP Designs and Performance to Receiving Water Impact Mitigation. Proc., Engineering Foundation Conference, Snowmass, CO (August 19–24, 2001) pp. 494–498. Dierkes, C., Kuhlmann, L., Kandasamy, J., and G. Angelis. Pollu- tion Retention Capability and Maintenance of Permeable Pave- ments. Proc., 9th International Conference on Urban Drainage, Portland, OR (September 8–13, 2002) pp. 444–445. Smith, K. P. Effectiveness of Three Best Management Practices for Highway-Runoff Quality along the Southeast Expressway, Boston, Massachusetts. Water-Resources Investigations Report 02-4059, U.S. Geological Survey (2002) 170 pp. Yu, S. L., and M. D. Stopinski. Testing of Ultra-Urban Stormwater Best Management Practices. Final Report VTRC 01-R7, Vir- ginia Department of Transportation (January 2001) pp. 43–48. 3.2.5. Methods to Improve Pollutant Removal in Existing Stormwater Systems Caltrans is conducting a multiyear study in Los Angeles and San Diego to examine the technical feasibility, costs, and operation and maintenance requirements of retrofitting struc- tural BMPs into existing highway and related infrastructure (Currier et al., 2001). Thirty-three locations are being retro- fitted with 39 BMPs using 12 different types of BMP tech- nologies. Automated monitoring stations have been installed upstream and downstream of each BMP to determine removal efficiencies from flow weighted composite samples. Con- stituents monitored in the runoff include suspended solids (e.g., sediment), metals, nutrients, and organics (e.g., gaso- line). To date, most projects have been sited, designed, con- structed, and monitored for at least one year. The purpose of the program has been to identify the problems and solutions that occur with structural BMP retrofits and to collect opera- tion, maintenance, and performance data for the BMPs. Results have indicated the existence of substantial construc- tion, maintenance, and cost challenges in retrofitting existing infrastructure with conventional structural BMP technology. Water quality monitoring results have indicated that average pollutant removal efficiencies are consistent with published values. The information collected on completion of the study will enable more accurate prediction of BMPs cost and per- formance for treating highway runoff. 74 3.2.5.1. Flood Control Retrofits for Water Quality Enhancement Only eight states have conducted studies or prepared reports on the retrofitting of existing stormwater management mea- sures at DOT facilities. Before the CWA, stormwater management primarily involved protecting people and property from floods through the construction of flood conveyance and detention facilities. Over the past few decades, however, the emphasis of storm- water management has broadened to include quality control and quantity control. In response to these dual stormwater management objectives, existing flood control basins often are retrofitted for water quality enhancement. The primary research question with regard to retrofitting flood control basins is how can detention facilities be modified to provide water quality benefits without compromising flood control objectives? Walesh (1991) presented approaches for retrofitting exist- ing stormwater detention facilities to improve quantity con- trol, add quality control, improve operation and maintenance, reduce safety hazards, enhance aesthetic attributes, and add recreation features. A matrix was used to illustrate retrofitting objectives for stormwater detention facilities versus available retrofitting measures. Examples presented were all based on actual facilities. Barth (2000) discusses the conversion of existing detention facilities (dry detention basins) into more functional treatment practices. The author states that the modification of older basins into stormwater wetlands or wet ponds is perhaps the easiest retrofit option for the following reasons: (1) stormwater is already managed in a distinct location, (2) there is already some resident acceptance and understanding of stormwater management, and (3) it usually involves minimal impacts to secondary environmental resources. Modification options include (1) excavating the pond bottom, (2) raising the embankment, (3) modifying the outlet structure, or (4) increas- ing the flowpath by using baffles, berms, and other treatments. The conversion of a dry detention pond at Villanova Uni- versity in Pennsylvania to a constructed wetland was pre- sented by Traver (2000). The steady, year-round base flows necessary for wetland establishment previously were piped through an underdrain below the detention basin. In the design and construction of the extended detention wetland, multiple meanders and gravel berms were placed to maxi- mize water storage. A sediment forebay was installed off- center to allow for sedimentation of small to medium-size storms, but to be bypassed by larger storms, so that resedi- mentation was avoided and flood protection was maintained. Several wetlands plants were sown throughout the site to allow for competitive selection and maximum nutrient uptake. The outlet was modified slightly to sustain the wetland water surface elevation and to maintain the original flood control hydraulics of the original detention basin design. The site is being monitored for both water quality and water quantity data.

Decker and Guo (2003) evaluated the feasibility of retro- fitting with the installation of a new subsurface flow gravel- bed wetland system to enhance water quality treatment of two existing dry detention basins within a single-family res- idential development in Morris Township, New Jersey. An overall model of the entire project area was prepared using the U.S. Army Corps of Engineers HEC-1 Model with the Natural Resources Conservation Service Methodology to facilitate calculation of peak flows and hydrographs, routing through the detention basins, and combination of hydro- graphs. The model was calibrated and verified based on pre- viously measured storm events. Based on the modeling analy- sis, the researchers concluded that the initial preferred retrofit plan should be rejected for the following reasons: • The loss of flood storage due to the filling inside the basin would result in an increase in peak flows downstream. • The flat slope of the existing basins prevented the pro- vision of a positive slope from the inlet to the outlet or the installation of any peninsulas to increase particle flow distance. • Cost of an underground concrete forebay system would be excessive. • Introduction of the forebay would result in additional friction and head loss that would cause an increase in flood elevations upstream of the upper basin. • The proposed 762-mm diameter overflow pipe from the flow splitter would be required to be raised to provide a positive slope to the outlet. Raising the pipe would increase the upstream hydraulic grade line and would cause flooding upstream of the upper basin. Additional alternatives to minimize hydrologic–hydraulic and environmental impacts for retrofitting at the upper basin in combination with enlarging or modifying the lower basin were evaluated and ultimately rejected because of excessive costs, site constraints, or adverse environmental impacts. An alternative site for the subsurface flow wetland between the two flood control basins was recommended. 3.2.5.2. Coagulants Methods to improve pollutant removal in existing storm- water systems can be borrowed from technologies used at municipal treatment facilities. One of the most common tech- nologies is the addition of coagulants—such as aluminum sul- fate (alum), ferric chloride, and lime—to enhance coagula- tion and sedimentation rates. As compared to other coagulants, alum has been shown by several researchers to provide a high pollutant removal rate and a stable end product, as long as pH is monitored and adjusted as needed (Harper et al., 1999; Escobar et al., 1998). A study by Price and Yonge (1995) evaluated four coagulants (alum, ferric chloride, and two pro- prietary cationic inorganic coagulants: SWT 848 and SWT 75 976) for their ability to enhance removal of sediment and metals. Results indicated that alum and SWT 976 were inef- fective at destabilizing the sediment suspension and initiat- ing floc formation in the relatively short rapid-mixing period. Ferric chloride and SWT 848 exhibited rapid floc formation and good solids settling characteristics, but ferric chloride was sensitive to dose, requiring dose optimization for each of the four test flow rates; SWT 848 did not exhibit dose sensi- tivity over the range of flow rates studied. In another study, one by Babin et al. (1992), researchers used lime and alum in an urban stormwater pond to reduce pH concentrations in the water column and to precipitate out particulate matter. Of the two chemical treatments, the researchers found that a lime–alum mixture was better at controlling macrophytes and shoreline filamentous algae, but alum was better at controlling planktonic algal growth and turbidity. A combination of both chemicals, lime (which ele- vates pH) and alum (which lowers pH), is used to maintain pH within a desirable range (6–10). Overall, water quality can be improved through the application of alum–lime mix- tures; however, these applications will have to be applied routinely throughout the open-water season because of con- tinuous nutrient inputs from point and nonpoint sources. The Southwest Florida Water Management District, under its Stormwater Research Program, conducted a study to deter- mine the feasibility of using an in-line alum injection facility as a stormwater treatment retrofit (Carr, 1999). The water quality constituents analyzed during the study included var- ious forms of phosphorous and nitrogen, and several metals. Individual storm data revealed that event mean percent loads were reduced. Reductions were observed in total phosphorus (37.2%), ortho-phosphorus (42.7%), ammonia (24.5%) and nitrate-nitrite (52.2%). A detailed analysis of the potential for aluminum toxicity to various fish and benthic species was conducted, and concentrations of monomeric species of dis- solved aluminum were measured at the inflow and outflow of the injection facility at levels that have been shown to be toxic or to affect adversely golden shiners, striped bass, rain- bow trout, and Daphnia magna (a zooplankton). An innovative coagulant and adsorbent that has not been used widely for stormwater treatment is chitosan, a biopoly- mer of shrimp and crab shells that is manufactured by Vanson HaloSource, Inc. Similar to alum, chitosan causes coagula- tion of fine sediment particles, which then allows for gravity settling, biofiltration, sand filtration, or cartridge filtration. FHWA used chitosan to reduce turbidity and enhance sand fil- tration in runoff from a road-widening project from a section of Big Salt Lake Road on Prince of Wales Island, Alaska (Nat- ural Site Solutions, 2002). Application of chitosan enhanced settling and reduced turbidity in sedimentation tanks by more than 90% and enhanced sand filtration that further reduced turbidities to less than 5 NTU. In another highway construc- tion project, Washington State DOT used chitosan for the treatment of construction site runoff from the Washington State I-90 Sunset Interchange Issaquah Project (Washington

State DOT, 2003). Chitosan, when added to settling pond effluent, caused the fine sediment particles to bind together and was removed with the sediment during sand filtration. Chitosan also removes phosphorous, heavy minerals, and oils from the water. Other areas are considering the use of coagulants for post- construction runoff, including the Lake Tahoe area, where fine particulates and nutrients have been identified as reduc- ing the lake’s clarity. In areas where sands are used for win- ter traction and the applied sands are ground into fine mate- rials, coagulants may be one of the only approaches for achieving desired suspended sediment levels. 3.2.5.3. Identification of Research Needs The literature review addressing methods for improving pollutant removal in existing stormwater control facilities indicates a few potential research gaps. With regard to retro- fitting flood control facilities to include water quality treat- ment, there appears to be a need for detailed design guidance that includes cost–benefit comparisons between retrofit alter- natives, with consideration of the overall feasibility and potential impacts to flood protection. Sponsoring research to evaluate whether other less conservative flood control meth- ods could be employed safely is another option. These meth- ods could include using more refined continuous simulation approaches to assess flood detention needs. With regard to coagulants, the literature reviewed for this study as well as the plethora of literature available in the area of wastewater management, showed that further research in this area likely is not a high priority. However, soil amend- ment recommendations for more passively improving perfor- mance in BMPs need to be developed. In selected locations, coagulant use may be necessary to achieve water quality goals. For these areas, more detailed design guidance for highway situations may be valuable. The potential impact of coagulants on receiving waters may warrant further research, particularly for fairly new products or products—such as chitosan—not used widely for stormwater treatment. 3.2.5.4. Primary References Babin, J., Prepas, E. E., and Y. Zhang. Application of Lime and Alum to Stormwater Retention Lakes to Improve Water Quality. Water Pollution Research Journal of Canada, Vol. 27, No. 2 (1992) pp. 365–381. Barth, C. A. Stormwater Retrofits: Tools for Watershed Enhance- ment. In Watershed Protection Techniques (2000) pp. 712–715. Carr, D. W. An Assessment of an In-line Alum Injection Facility Used to Treat Stormwater Runoff in Pinellas County, Florida. Proc., 6th Biennial Stormwater Research and Watershed Man- agement Conference (September 14–17, 1999) pp. 68–79. Currier, B., Taylor, S. M., Borroum, Y., Friedman, G., Robison, D., Barrett, M., Borroum, S., and C. Beitia. California Department of Transportation BMP Retrofit Pilot Program. Presented at 80th 76 Annual Meeting of the Transportation Research Board, Wash- ington, DC (January 7–11, 2001). Decker, T. R., and Q. Guo. Drainage Evaluations of a Proposed Stormwater Detention Basin Retrofit. Water World and Environ- mental Resources Congress 2003, Philadelphia, PA (June 2003). Escobar, I. C., Randall, A. A., and F. E. Marshall, III. Alternative Methods in Stormwater Management. Proc., of the 25th Annual Conference on Water Resources Planning and Management (1998) pp. 580–585. Harper, H. H., Herr J. L., and E. H. Livingston. Alum Treatment of Stormwater: the First Ten Years. New Applications in Modeling Urban Water Systems, Monograph 7 in the series, Published by CHI, Guelph, Canada (1999). Natural Site Solutions, LLC. Storm-Klear Liqui-Floc Project Bul- letin. Federal Highway Administration, Prince of Wales Island, AK (2002). Price, F. A., and D. R. Yonge. Enhancing Contaminant Removal in Stormwater Detention Basins by Coagulation. In Transportation Research Record 1483, TRB, National Research Council, Wash- ington, DC (1995) pp. 105–111. Traver, R. G. Creating a Wetland Stormwater Best Management Practice—A Retrofit. Proc., Joint Conference on Water Resources Engineering and Water Resources Planning and Management, Minneapolis, MN (July 2000). Walesh, S. G. Retrofitting Stormwater Detention Facilities for Quality and Quantity Control. Proc., International Conference on Urban Drainage and New Technologies, Dubrovnik, Yugo- slavia (June 19–22, 1991) pp. 283–290. Washington State Department of Transportation. Washington State I-90, Sunset Interchange, Issaquah Project. Project Summary Report, http://www.wsdot.wa.gov/projects/I90SunsetInterchange/ default.htm (2003). 3.2.6. Erosion and Sediment Control Erosion prevention reduces the amount of sediment gen- erated from the land surface. Once erosion occurs, sediment- control practices are necessary to limit the downstream movement of the sediment. The review of erosion and sedi- mentation controls included in this section is limited to stud- ies that have evaluated the effectiveness of stormwater BMPs designed to control the detachment (erosion controls) and transport (sediment controls) of sediment from road surfaces, right-of-ways, and banks of receiving waters. The effects of scour, sedimentation, and turbidity on receiving waters are discussed in section 3.5.2. Characterization of highway con- struction runoff is discussed in section 3.4.7. This review, as with the preceding and subsequent sections, does not attempt to exhaust the literature on the subject, but instead provides a brief summary of some key studies. Potential research questions with regard to erosion and sedimentation controls include • How effective are temporary soil stabilization and ero- sion controls at keeping particulates in place? What is the minimum grain size effectively held in place?

• Are vegetated erosion controls more cost-effective than nonvegetated controls? • What are the differences in erosion control effectiveness between native and nonnative vegetation? • What are the variables that affect seed germination? • What are some alternatives to riprap for in-stream chan- nel stability BMPs? The most effective erosion prevention measure is almost always minimization of the amount of land being disturbed. Once land is disturbed, erosion prevention controls must be implemented. Some common controls include erosion mats, compost and mulch, and hydroseeding. A study by Miller et al. (2002) evaluated the effectiveness of composted yard waste mulch, sod, and erosion mats in controlling erosion and establishing permanent vegetation along Florida highways. The study found that the composted mulch can effectively control erosion but does not necessar- ily facilitate the growth and establishment of turf grass or other vegetation. The composted mulch can provide slope stability for periods of at least 18 months, and probably longer, with or without vegetative growth. Lack of sufficient rainfall during the study period severely limited establish- ment of vegetation (and also limited erosion) in compost mulch-treated plots. Sod and erosion control mat treatments had greater turf grass and vegetative cover, but all treatments effectively controlled erosion during the study. Seeding exper- iments indicated that seed incorporation into composted yard waste mulch may not be necessary during periods of abundant rainfall; however, it is necessary during periods of low rainfall. Erosion control mats can be seeded either above or below the mat without affecting seed germination. In a 5-year research project by Banovich and Outcalt (2002), three test zones were established to evaluate various erosion control materials and methods on cut and fill slopes of US 40 on the west side of Berthoud Pass, a high altitude (∼10,000 ft) Colorado pass. Snowmelt runoff and severe spring and summer rain storms frequently washed away the easily eroded sandy soil, preventing vegetation from estab- lishing itself on the slopes, some of which were steeper than 11. The results of the study showed that all of the cellular confinement materials and soil retention blankets were suc- cessful in holding and reinforcing the plants’ root systems. The average density of plant shoots in the test areas (blankets and geocell materials with seeding, fertilizer, and mulch) ranged from -20% to 276% of the density in the control sec- tions (seeding, fertilizer, and mulch only). Based on observations of the surface conditions and on quantities of plant material on the slopes, it appeared that all of the blankets and cellular confinement products provided reinforcement to the scarp-forming area of the cut slopes. Based on the plant counts in the six test areas, the effective- ness of the products ranks as follows from most effective to least: Enviro Grid, Geoweb, Armater Geocell, Enkamat 20-S, Multimat, Pyramat. The Pyramat blanket in one of the zones 77 did not conform to irregularities in the slope as well as the other products. This is the only zone where the plant count was lower than the count in the control section. Failures occurred in the Armater test section when the anchor system failed and where the product was placed over a large boulder. Polyacrylamides (PAMs) are water-soluble synthetic poly- mers widely used in furrow irrigation to reduce erosion and turbidity. McLaughlin (2002) evaluated PAM, in the labora- tory as well as in the field, for construction site erosion and turbidity control. A laboratory screening was conducted for 11 PAMs on 13 sediment sources from North Carolina DOT construction sites. In addition, field tests were performed for two PAMs at two rates, with and without straw mulch and seeding, on a 21 fill slope, a 41 cut slope, and a 41 fill slope. The results indicated that no one PAM is effective for turbidity reduction on all sediment sources but that several are promising for many soils. Superfloc A-100 ranked among the top three flocculants for 10 of the 13 sediment sources. Some PAMs are equally effective but at different doses, some as low as .075 mg/L, or a few grams per 1,000 ft3 of water. Tests of PAMs with and without mulching on 21 slopes at North Carolina DOT construction sites resulted in erosion rates that were 20 times greater on bare soil plots after the first seven events, with or without PAMs, compared to those mulched with straw and seeded to grass. During the eighth and last event, in which more than 6 cm of rain was recorded, rates of more than 50 tonnes/ha were recorded for a single, intense storm event for the bare soil plots compared to 3–9 tonnes/ha on the mulched and seeded plots. PAMs at the highest rate (11 kg/ha) were effective in reducing erosion and turbidity on the 41 cut slope with a clay loam texture, but the effect declined with each storm event. On the sandy 41 fill slope, there was no evidence of any effects of PAMs, even at an application rate of 20 kg/ha. Nwankwo (2001) evaluated the effectiveness of PAMs at controlling erosion from three highway construction projects around Wisconsin. Comparison of CFM 2000, PAM, with other erosion-control products that are used currently by Wisconsin DOT, showed that this product is effective in con- trolling erosion, is applied easily, and, at a material and instal- lation cost of approximate $1,250/ha ($500/acre), is rela- tively inexpensive when compared to the $11,250/ha ($4,500/ acre) for Wisconsin DOT Class 1 Type A erosion mats. Also, when the manufacturer’s recommended application rate is followed, the product was found to be environmentally safe. The performance of CFM 2000, PAM, in controlling erosion is based on the fact that it binds soil together into particles of a larger size; the binding makes the soil more resistant to col- lapse, dispersion, and shear forces. Soil infiltration rates also appear to increase with the use of PAMs, resulting in more available water for the seeds to germinate, lower runoff, and less soil detachment from erosion. CFM 2000, PAM, performed comparably to erosion mats and better than mulch and seed on slopes of 21 or less in controlling erosion before the establishment of permanent

vegetation. The combinations of the polymer, seed, and mulch performed the best for erosion control and vegetative growth. From the data of on the CTH N test plots it follows that Test Plot 2 (Class 1 Type A erosion mat plus seed) and Test Plot 3 (PAMs, mulch, and seed) produced the smallest amounts of eroded soil of all five test sections after 6 months of obser- vation. Initial indications also showed that Test Plot 3 pro- duced not only the most seed germination and the densest vegetation but also the tallest grass plants. Although field observations 8 months after the products were placed showed no significant difference in plant height, the test plot with PAMs, mulch, and seed appeared to have the denser vegetation. The Georgia DOT recently completed a research study to evaluate the effectiveness of using PAMs in erosion control and runoff turbidity reduction on Georgia’s DOT construc- tion projects and to establish BMP guidelines for Georgia DOT. A report has not yet been prepared. Caltrans (2002) initiated a series of laboratory experi- ments under a variety of rainfall regimes and erosion control treatments to identify and select plant species that demon- strate initial fast growth and potential long-term erosion con- trol. The plants examined included native and nonnative nat- uralized species. Preliminary results indicated the benefits of using jute netting for optimum vegetation cover. Results also indicated that the type of vegetation cover (grass, legume) was affected by the erosion control treatment. The greatest water quality improvements were seen with the use of bound fiber matrix, jute, and straw. Initial results indicated that sedi- ment amounts decreased with the hydroseeding of native seeds as compared to plug planting. Soil roughening and using crimped straw were shown to be the most effective forms of erosion control on test plots in combination with vegetation. Also, results showed that native vegetation was affected neg- atively when fertilizer was applied on the test plots. California Polytechnic State University, in conjunction with Caltrans, investigated soil stabilization treatments and burial depth influences on the germination capabilities of seven native California plant species and annual ryegrass (Chiaramonte et al., 2003). Six treatments—gypsum, gyp- sum and wood fiber, guar tackifier, bona fide fiber matrix, wood fiber, and no treatment—were applied hydraulically to the soil surface. One hundred seeds of eight plant species (Lotus scoparius, Lupinus succulentus, Artemesia califor- nica, Eriogonum fasciculatum, Escholzia californica, Bro- mus carinatus, Achillea millefolium, and Lolium multiflo- rum) were hand planted into each treatment. Eriogonum fasciculatum, Artemesia californica, and Lotus scoparius experienced less than 18% germination for all treatments. Lupinus succulentus experienced less than 13% germination for all treatments. Lolium multiflorum (ryegrass), with the highest germination rate for all species, had higher than 86% germination rates for all treatments. The bona fide fiber matrix treatment resulted in the lowest overall germination percentages, and gypsum and wood fiber treatment resulted 78 in the highest overall germination percentages. The depth resulting in the greatest germination percentage was the 0.25- inch burial depth. The use of nonnative species for roadside erosion and sed- iment control has become an issue in many states because of the related invasive and aggressive establishment. Com- monly used species like reed canary grass, sweet clover, perennial rye, smooth brome, and crown vetch have led to weed problems in many areas, in some instances even lead- ing to the plant being listed on states’ noxious weed lists. This issue has prompted FHWA to prepare the handbook Roadside Use of Native Plants (http://www.fhwa.dot.gov/ environment/handbook.htm). Landphair et al. (2001) evaluated the benefits and perfor- mance of native plant materials compared to an introduced species commonly used in the erosion control mixes for the stabilization of roadsides in Texas. The study found that wildflower-only mixes did not prove successful; there was some germination in the first year of planting, but the vege- tation appeared to be gone by the second year of the project. A recent check of the plots, however, revealed a greater per- sistence than was evident in 1999 and 2000. Bermuda grass was very aggressive in the first few years of planting. Where researchers originally planted native grasses and forbs, the latter began to gradually displace the Bermuda grass. This displacement likely can be attributed to shading of the low- growing invaders and the fact that mowing was being done at this time. Native grasses will continue to increase if mowing is not permitted. However, stands of natives will still require some cultural management, such as mowing or burning, to main- tain their vitality and to prevent the invasion of woody species (if woody species are not desired). The erosion control prop- erties of native grasses do not appear to be as effective as the grass mixes currently used by Texas DOT. This is probably a function of their clump-forming growth habit and the slow- developing nature of the native species. This finding argues in favor of the practice of using nurse grasses with the native prairie species. The vegetation reached at least 70% cover by the second year. However, the aesthetics of the natives prob- ably would not meet expectations during some parts of the year. Finally, there was no evidence that the native plant materials made any more significant difference than the other plant materials in the rate of surface erosion or contributed to any increase in tensile strength of the surface soil layer. How- ever, in 2 or 3 years, the larger natives, such as witchgrass and Little Bluestem, will develop more mature root systems that may indeed show some increase in soil shear strength. Riprap is used commonly for roadway protection at streams and often is used at the expenses of increased water temperature and decreased quality of stream habitat due to riparian vegetation removal. Researchers at Oregon State University investigated the potential for integrating riparian vegetation into stream bank protection designs (Klingeman et al., 2002). Based on the research, it appears that vegetation

may be incorporated safely into riprap projects at the time of project construction. However, allowing vegetation to grow in existing riprap requires caution because the riprap systems were not designed with this in mind, which introduces more uncertainty and the possibility of failure. Examination of some revetments that have growing vegetation suggests that riprap rock displacement does occur and that adjacent rocks are pushed up along the trunk. However, rock displacement does not diminish the riprap integrity when the tree is part of an extensive mass of vegetation growing in the riprap, as the flow resistance provided appears to diminish the local veloci- ties at the vegetated riprap. Isolated trees in riprap have not yet been observed, so judgment is reserved on such conditions. NCHRP Project 24-19, Environmentally Sensitive Channel- and Bank-Protection Measures, includes the development of selection criteria; design guidelines; and techniques for the type, size, and placement of environmentally sensitive channel- and bank-protection measures. The selection crite- ria, guidelines, and techniques are based on engineering and environmental considerations. Vegetated riprap and riparian habitat are among the many different research areas. 3.2.6.1. Identification of Research Needs With regard to temporary vegetation controls, there is seemingly sufficient research with respect to the erosion con- trol effectiveness of compost/mulch, erosion control mats and blankets, and cellular confinement technologies. Also, there is sufficient guidance in this area (see Appendix B for a list of guidance manuals). The effectiveness of erosion controls at removing fine particulates does not seem to be covered ade- quately in the literature. However, the use of PAMs or other flocculants in conjunction with temporary vegetation controls holds promise for controlling erosion of fine particulates. The application of PAMs as a highway erosion control BMP is fairly new, so there are a limited number of studies available in the literature with regard to highways. However, the two studies cited above indicate that the use of PAMs is indeed an effective erosion control BMP. In fact, the use of PAMs is one of the recommended BMPs in the California Stormwater Quality Association’s Construction Handbook (www.cabmphandbooks.com). Furthermore, there are several studies in the realm of irrigation and agricultural practices that demonstrate its effectiveness and environmental safety (http:// www.nwisrl.ars.usda.gov/pampage.shtml). Therefore, further research on the effectiveness of PAMs is not warranted, unless, as stated above, the research involves the use of PAMs to enhance the effectiveness of other BMPs. With regard to native versus nonnative vegetation, it appears that more research on how to increase germination and survival rates, as well as overall soil coverage, of native vege- tation is needed. The two studies presented both indicate that native species are not as effective at establishing themselves after being applied hydraulically to slopes. The Texas DOT study suggests that the native species are not as effective at 79 controlling erosion as the grass mixes currently used; however, this likely is due to the density of vegetation establishment. With regard to bank protection research, the primary research needs identified by Klingeman et al. (2002) are (1) to evaluate and compare different types of vegetation for riprap planting; (2) to study the necessary top elevation for conventional riprap as a function of velocity, turbulence, and flow duration; (3) to compare terraced versus sloping riprap in terms of hydraulic performance and planted vege- tation success; (4) to evaluate alternative current deflectors that have a lesser effect on aquatic habitat than riprap, but are effective in preventing bank erosion; and (5) to conduct more-detailed inspection of riprap where vegetation is now growing or has grown in order to better understand its impacts to bank stability. 3.2.6.2. Primary References Banovich, M., and W. Outcalt. Evaluation of Slope Stabilization Methods (US 40 Berthoud Pass). Final Report, Colorado Depart- ment of Transportation Research Branch (2002) 22 pp. Caltrans. Rainfall Simulation: Evaluating Hydra Seeding and Plug Planting Technologies for Erosion Control and Improved Water Quality. Vegetation Establishment and Maintenance Study, Experiments: RS2 and RS3, Central Coast District 5 (2001–2002) 131 pp. Chiaramonte, M., Scharff, M., Hallock, B., and M. Curto. Effects of Erosion Control Treatments on Native Plant and Ryegrass Establishment. Proc., International Erosion Control Association (IECA) 34th Annual Conference and Exposition, Las Vegas, NV (February 24–28, 2003). Klingeman, P., Pyles, M., Hibbs, D., and B. Kauffman. Roadway Applications of Vegetation and Riprap for Streambank Protec- tion. Final Synthesis Report, Oregon Department of Transporta- tion Research Group (2002). Landphair, H. C., Schutt, J. R., and J. A. McFalls. Native Vegeta- tion or Bermuda Grass? Testing the Erosion Control and Engi- neering Properties of Roadside Vegetation. Project Summary Report, Texas Department of Transportation, Austin (2001) pp. 1504–1505. Nwankwo, K. N. Polyacrylamide as a Soil Stabilizer for Erosion Control. Final Project Report, Wisconsin Department of Trans- portation, Milwaukee (2001) 29 pp. McLaughlin, R. A. Measures to Reduce Erosion and Turbidity in Construction Site Runoff. Research Project Report, North Car- olina Department of Transportation (2002) 31 pp. Miller, G. L., Black, R. J., and G. Kidder. Erosion Control Along Florida Roadways. Florida Department of Transportation, Envi- ronmental Management Office (2002) 88 pp. 3.2.7. Design Variables Affecting BMP Performance The performance of stormwater control facilities is believed to be affected in large part by design variables such as geom- etry, surface area, outlet control structure, and vegetation den- sity and type. Because of site- and project-specific constraints,

BMPs of the same type that follow the same design criteria and guidelines may end up vastly different in terms of per- formance. In fact, differences in BMP designs likely account for a large amount of the variability observed among various BMP performance findings, such as those compiled in the ASCE/EPA BMP Database (www.bmpdatabase.org). The reanalysis of the BMP database conducted by Strecker et al. (2003) indicated that some design parameters (e.g., the cap- ture volume of a BMP relative to monitored storm volume for volume-based BMPs) are found to be statistically signif- icant with regard to performance. Design requirements and recommendations provided in BMP design manuals (see Appendix B for a brief list of available design manuals) often are based on a limited num- ber of studies, the majority of which are conducted in labo- ratories where only a limited number of design configura- tions are investigated under strictly controlled conditions. Alternatively, they have been based on good engineering judgment. As discussed above in section 3.2.3, Hydraulic Assessment, the positive association of BMP performance and hydraulic residence has been well documented, so BMP design criteria often are composed in terms of detention or contact time (see section 1.1.2), and design guidelines often are intended to increase hydraulic residence. Some potential research questions with regard to design variables and BMP evaluation include • How do BMP geometry or specific surface area, or both, affect pollutant removal? • Other than overall size relative to incoming storms, what are the most influential design variables affecting the performance of a BMP? A study by Barrett et al. (1997) investigated the impacts of swale length, water depth, and season of the year on removal efficiency of a highway swale in Austin, Texas. Results indi- cated that swale length and water depth affect the removal of constituents. TSS removal efficiency was found to be reduced as water depth increased. The reduction in removal efficiency confirmed expectations, since the filtration action of the grass blades was expected to be lower for higher water depths. Removal of other constituents was not correlated as strongly with water depth. Pollutant removal efficiency increased with length, but the increment of increased efficiency dimin- ished as runoff proceeded down the swale. This trend was evident especially for TSS, chemical oxygen demand, total phosphorus, and metals. The majority of removal occurred in the first 20 m of the swale for these constituents. A study related to the one by Barrett et el. investigated the effect of a swale underdrain on the removal efficiency (Walsh et al., 1997). During nine experiments, simulated highway runoff was sampled on the surface of a swale and from the swale’s underdrain pipe after percolating through a top layer of grass sod, 16 cm of topsoil, and 6 cm of gravel. Concentrations of constituents in runoff that had percolated 80 through the soil in the swale generally were lower than the concentrations in surface runoff after 40 m of treatment by the swale. The underdrain water quality demonstrated the fil- tering capability of the soil and reflected water quality of recharge for groundwater in situations with shallow soils. Petterson et al. (1999) studied the effects of specific sur- face area (i.e., the ratio of the pond area and the impervious catchment area, m2/ha) on the pollutant removal efficiency of four stormwater ponds in Sweden. Each pond had a different specific surface area, but the depths were similar (1.2–1.7 m). The results of the comparison showed that the removal effi- ciency of TSS, phosphate, copper, lead and zinc increased up to a certain level of surface/impervious area, 250 m2/ha, and above this level the increase was not as significant. Nitrogen showed a less significant, but similar trend. However, the pollutant removal efficiency of nitrogen was low for all of the ponds. In a study by Horner (1990), the pollutant removal effec- tiveness of laboratory model-scale sedimentation pond designs was evaluated. The results of the laboratory tests demon- strated that the following design features, in concert, maxi- mized actual water residence time to promote sedimentation: (1) length/width ratio of 5:1; (2) series arrangement of two chambers rather than a single pond of equivalent size and shape; and (3) use of a perforated riser outlet. To verify these results in a full-scale application, a sedi- mentation pond was designed according to the laboratory find- ings, constructed in a highway right-of-way, and monitored for pollution-control performance. Another sedimentation pond without these design features was tested for comparison. Sam- ples were analyzed for solids, metals, phosphorus, and organic content. Results demonstrated that the ponds designed accord- ing to the laboratory findings were both more efficient in pol- lutant removals and less costly (per unit area served) than the pond to which they were compared. 3.2.7.1. Identification of Research Gaps and Needs Based on the review of literature, it is apparent that some of the primary design variables affecting BMP performance— such as outlet structures, baffles, berms, and vegetation den- sity, in addition to the volume a system is able to capture— are those that control flow. Design features specific to individual types of BMPs, such as specific surface area for detention facilities and flow length for the swales, also are significant factors to consider when evaluating and compar- ing BMP performance. These design variables are related directly to physical treatment mechanisms of sedimentation and filtration. Other design variables that are more related to the bio- and geo-chemical treatment mechanisms—such as vegetation and soil type—also may be important design fac- tors. However, no studies were found that compared BMP performance according to these variables, indicating a poten- tial research gap.

One of the problems addressing research gaps in this area is that in order to provide verification with field studies, a large number of BMP studies with different BMP design attributes are needed. One purpose of the National BMP Database is to provide a repository for design and performance data to facil- itate future research on BMP design versus performance. Recent ASCE database project efforts have included conduct- ing this type of analysis, but given the relatively small number of BMPs in most BMP categories and the large number of design parameters, short-term research in this area could be premature. 3.2.7.2. Primary References Barrett, M. E., Keblin, M. V., Walsh, P. M., and Malina, J. F. Jr. Evaluation of the Performance of Permanent Runoff Controls: Summary and Conclusions. Online Report 97-3, Center for Research in Water Resources (1997). Horner, R. Improving the Cost Effectiveness of Highway Construc- tion Site Erosion and Pollution Control. Final Report GC 8286- Task 7, Construction Erosion Control-II, Washington State Trans- portation Commission, Federal Highway Administration (January 1990). Petterson, T. J. R., German, J., and G. Svensson. Pollutant Removal Efficiency in Two Stormwater Ponds in Sweden. Proc., 8th Inter- national Conference on Urban Storm Drainage, Vol. 2, Sydney, Australia (August 30–September 3, 1999) pp. 866–873. Strecker, E. W., Quigley, M. M., and B. R. Urbonas. A Reassess- ment of the Expanded EPA/ASCE National BMP Database. Proc., National Conference on Urban Storm Water—Enhancing Programs at the Local Level, Chicago, IL (February 17–20, 2003) pp. 555–573. Walsh, P. M., Barrett, M. E., Malina, J. F., and R. J. Charbeneau. Use of Vegetative Controls for Treatment of Highway Runoff. Research Report 2954-2, Center for Transportation Research (1997) 125 pp. 3.2.8. Unit Processes As noted in the premeeting report for the 2002 TRB Research Needs Meeting (Sansalone, 2000), with regard to modeling of BMP unit processes, sedimentation and infiltra- tion appear to be well covered in the literature. However, other BMP water quality treatment unit processes—such as sorption processes (absorption and adsorption), phytoremediation, solar radiation, and volatilization—need to be studied further before reliable BMP performance models can be developed. The lack of information on the modeling of BMP treatment trains appears to be a knowledge gap, as does the decrease in the treatment efficiency of BMPs as a function of time. As part of NCHRP Project 25-20(01), Evaluation of Best Management Practices for Highway Runoff Control, the proj- ect team led by Professor Wayne Huber of Oregon State University will evaluate the performance of approximately 20 different types of highway BMPs from a unit processes perspective. From a unit processes point of view, stormwater 81 controls can be partitioned into approximately four to nine fundamental process categories (adapted from Metcalf and Eddy, 2003): • Sedimentation—as in ponds, basins, or small storage devices—including the possibility of resuspension; • Filtration and adsorption, trash racks, screens, sand fil- ters, compost filters, soil, and vegetation; • Infiltration, in which filtration is accompanied by removal (redirection to the ground) of surface water runoff, including porous pavement and concrete blocks; • Hydrodynamic devices, as in swirl concentrators or other secondary current devices; • Biological treatment and uptake, within storage devices or in combination with infiltration, as in bioswales and wetlands; • Oil–water separators and devices that rely upon density differences; • Chemical treatment to enhance flocculation, use of alum or for disinfection, and use of chlorine; • Reduction in runoff volumes via evapotranspiration; and • Combinations of all or some processes (e.g., in ponds, wetlands, and swales). The final report will include a section that identifies gaps and research needs with regard to BMP performance charac- terization and statistical assessment. 3.2.8.1. Identification of Research Needs Based on the fact that (1) Project 25-20(01) was initiated to begin filling highway stormwater performance evalua- tion and assessment research gaps identified by earlier investigators and (2) the final report will include the iden- tification of additional research gaps and needs with regard to unit process evaluations, it is deemed premature at this time to include such an identification in this report. How- ever, based on the opinion of the 25-20(01) project team, the most likely gap will be treatability data and information that can be used to characterize the fundamental removal processes (unit processes) in action within a given BMP, as well as the simple lack of monitoring data of several dif- ferent BMP types. 3.2.8.2. Primary References Sansalone, J. The Role of Water in Ecologically Sustainable Trans- portation. In Transportation in the New Millennium, Committee on Hydrology, Hydraulics, and Water Quality, TRB, National Research Council, Washington, DC (2000). Metcalf and Eddy, Inc. Wastewater Engineering—Fourth Edition. McGraw-Hill, New York (2003).

3.2.9. Low Impact Development/ Distributed BMPs LID technologies provide tools that can promote with max- imum efficiency the dual goals of environmental protection and transportation system improvements. LID technologies are based on using the cumulative effects of multiple, redun- dant, decentralized stormwater management techniques to meet quantified stormwater management thresholds. LID is designed to create a multifunctional–multibeneficial use in every aspect of the urban landscape to manage runoff and, where possible, to restore or maintain effectively the natural hydrologic and water quality regimes. The water quality and economic benefits of LID have been demonstrated successfully for residential, commercial, and industrial development applications in the United States, Europe, and the Pacific Rim nations (see www.epa.gov/ owow/nps/lid for a list of studies). In many cases, LID has been shown to be more cost-effective as it makes multifunc- tional use of the landscape to manage runoff on site, and, therefore, reduce conventional drainage infrastructure. For linear transportation systems, LID can allow transportation agencies to maximize the use of existing rights-of-way for stormwater management, reducing the need to procure addi- tional land to meet stormwater management objectives, and thereby can reduce project costs. This relatively new approach shows tremendous potential, particularly in highly urbanized areas, for new development and retrofit projects. At the present time, a design manual exists for suburban residential development in Prince Georges County, Maryland (see Appendix B for this and other LID guidance references). Communities and resource agencies across the nation rapidly are adopting LID practices as a new alternative to help meet regulatory requirements and resource protection goals. Since LID is a relatively new practice, many special con- siderations need to be addressed for linear transportation sys- tems. Some of the characteristics of linear transportation sys- tems that present challenges for LID methods are extensive cut-and-fill situations that cross multiple streams, drainage divides, limited rights-of-way, multiple project phases, and limited maintenance resources. Key planning and design LID strategies that have been used for urban retrofits and green development include impact avoidance, minimization, strate- gic timing and routing of runoff, uniform distributed inte- grated management practices, and pollution prevention. LID stormwater control practices include combinations of dis- charge dispersal, infiltration, retention, bioretention, filtration, impervious disconnection or removal, detention, amended soils, water reuse, and increasing surface roughness. Inte- grating LID design principles and practices can be incorpo- rated into every aspect of a highway right-of-way (medians, shoulders, swales, pipes, inlets, streetscapes, slopes, green space, and others) to create a more hydrologically functional transportation system, instead of using drainage infrastruc- ture solely for stormwater conveyance. 82 Economic benefits of LID come from reduced costs that result from either downsizing or eliminating end-of-pipe treatment systems. Without accurate methods to simulate the retention capacities of LID systems, end-of-pipe controls will continue to be full-sized, and much of the economic ben- efit from using LID will be lost. The standard base models used to estimate runoff volumes and rates may not be well suited for evaluating LID systems, since stormwater routing options may be unaccounted for, and input parameters are “hard wired” into 16 standard scenarios that fail to simulate the hydraulic response of engineered LID systems. With regard to engineered roadsides and sheet-flow dispersion, a new or modified set of parameters may need to be developed. The newest version of SWMM5, EPA’s stormwater analysis model developed by Wayne Huber and others, will have routing and continuous simulation options that will be useful for designing LID systems in the future. Huber is one of the principal investigators in NCHRP Project 25-26, Develop- ment of a Low-Impact Development Design and Construc- tion Manual for Transportation Systems. One of the gaps in LID technologies knowledge is a long-term understanding of hydrologic effects. For example, will LID-type approaches provide enough hydrological control in regions that tend to have back-to-back storms? The Low Impact Development Center in Maryland (www. lowimpactdevelopment.org) has developed a series of case studies as part of its LID training courses on the economic and environmental benefits of LID for residential and com- mercial development. Future LID projects planned in Seattle (SeaStreets Project, www.ci.seattle.wa.us/util/SEAstreets) and Pierce County (Village on the Hylebos) will monitor closely design and construction costs. Evaluations of the eco- nomic feasibility of LID will determine whether or not those practices will gain general acceptance in the development industry or will be limited to features of boutique develop- ment projects. LID practices are likely to be incorporated into stormwater designs on SR 405 expansions, SR 16 HOV, and other Washington State DOT projects. Detailed eco- nomic analysis of the cost and benefits of using those LIDS should be performed, if that is feasible. The Low-Impact Development Design and Construction Manual for Transportation Systems is under development as part of NCHRP Project 25-26. The manual will be an effective tool in designing and constructing LID facilities with consis- tency, which in turn leads to effective technologies that can be monitored and compared. The product will include practical design standards and practices that meet identified regulatory requirements and resource protection goals. The anticipated criteria that will be used to develop the LID methods will include regional applicability, highway safety, spatial and temporal requirements, soil characteristics, pollutant removal effectiveness, hydrologic benefits, lifecycle maintenance requirements, and resultant costs. A series of conceptual design standards will be developed for practical field evalua- tion and optimization. Several DOTs, including Washington

State DOT, Maryland State Highway Administration, Virginia DOT, and Caltrans expressed interest in piloting LID technol- ogy because of its potential for addressing the escalating envi- ronmental requirements that are projected. NCHRP Project 25-26 will make recommendations on modeling programs that can simulate effectively LID sys- tems and will develop an applicable model. Washington State DOT has an insufficient amount of funding to update its MGSFlood continuous model to allow it to route water to structures in series. These structures could be modeled as leaky, where water losses can be incorporated. As time pro- gresses, LID practices can be input as significant leaky struc- tures, but future field evaluations will be needed to quantify accurately the water losses in LID systems. This will take a significant amount of time and money. The SeaStreets pro- ject in Seattle found an approximate reduction of 40% in the volume of runoff, which could be incorporated into assess- ments of BMP performance. The Friends of the Rappahannock and the Low Impact Development Center (www.lowimpactdevelopment.org) are developing guidance and strategies for rural communities in Virginia to incorporate LID into their local resource protec- tion and regulatory programs (Weinstein and Tippett, 2003). The first part of this effort includes evaluating state and local codes to determine what, if any, necessary legislative, code, or local regulations need to be modified to include LID. Iden- tifying areas in the town and land uses that are appropriate for LID technologies follow this effort. The next step will be to develop materials for developers and plan reviewers to help guide them through the development process when the use of LID is appropriate. The final step will be to design and implement a small demonstration project that showcases LID features, such as rain gardens, soil amendments, permeable pavers, and infiltration devices. 3.2.9.1. Identification of Research Needs Pilot projects conducted by several researchers have demon- strated the potential of LID to meet regulatory requirements, but substantial work needs to be conducted on developing LID design strategies, performance standards, and specifica- tions. LID’s decentralized approach to stormwater manage- ment technology has tremendous potential to supplement or in some situations to replace completely conventional cen- tralized stormwater BMP approaches; however, LID’s appli- cability, efficacy, and long-term economic sustainability have yet to be determined or documented for transportation sys- tems. One long-term research need is documenting the type of hydrologic losses that can be achieved via LID regionally and under various soil, slope, and vegetation conditions. 3.2.9.2. Primary References Weinstein, N., and J. Tippett. Low Impact Development Strategies for Rural Communities. Proc., National Conference on Urban 83 Stormwater-Enhancing Programs at the Local Level, Chicago, IL (February 17–20, 2003) pp. 497–501. Rushton, B. T. Low-Impact Parking Lot Design Reduces Runoff and Pollutant Loads. Journal of Water Resources Planning and Management, Vol. 127, No. 3 (June 2001) pp. 172–179. U.S. Environmental Protection Agency. Low Impact Development (LID): A Literature Review. Report EPA-841-B-00-005, Office of Water (October 2000) 41 pp. 3.2.10. BMP Modeling Software modeling tools have become a vital part of storm- water management. The complexity of BMP models ranges from simple spreadsheet calculations to multitiered combi- nations of sophisticated models. Problems and limitations that need to be overcome in this area include a lack of consistent BMP monitoring and performance data and a better understanding of BMP treatment processes. Current sources of data used in modeling applications include the NWS, EPA, USGS, and the USDA. The monitoring and per- formance data limitations encountered in the modeling of BMPs could be overcome through the growth and the evolu- tion of the ASCE/EPA BMP database. As more and more studies are added to the BMP database, BMP properties in relation to various external variables can be determined more accurately using statistical analysis and other mathematical techniques. Areas of interest and possible knowledge gaps with respect to BMP modeling include • Unit processes in BMPs (sedimentation, infiltration, absorption, adsorption, biodegradation, photolysis, vola- tilization, etc.); • Prediction of BMP longevity and BMP effectiveness as a function of time; • Modeling of BMP treatment trains; • Modeling of distributed BMP systems such as LID; • Development of stochastic BMP performance models; and • Factoring the effects of maintenance on BMP perfor- mance. Spreadsheets are simple programs that are used widely by engineers for a variety of technical applications. Numerous spreadsheet models have been created for water quality and hydraulic models and, more recently, for BMP modeling. Hayes et al. (2003) discuss the application of the Integrated Design and Evaluation Assessment of Loadings (IDEAL) model to BMP design and evaluation issues. IDEAL is a process-based stochastic spreadsheet model that is capable of predicting and routing runoff and pollutant loadings. IDEAL is limited in the number of pollutants that can be simulated satisfactorily, though it can model vegetative strips, dry deten- tion ponds, and wet detention ponds. New models are sometimes created to tackle new issues. Likewise, existing models are often modified, updated, and enhanced to address new problems or to take advantage of

new technologies and more efficient algorithms. SWMM, in use since 1970, has been used primarily for hydrologic and hydraulic modeling; however, the increasing emphasis on water quality and environmental regulation compliance has been a driver for the development of new tools and the enhancement of old ones. Huber (1996) presented a discus- sion about the use of SWMM in BMP modeling, including a discussion of model enhancements, and a case study to illus- trate the use of the model. According to Baxter (2002), both SWMM and the Better Assessment Science Integrating Point and Non-Point Sources (BASINS) have BMP modeling capabilities. SWMM is particularly proficient at predicting pollutant loads. BASINS is based on an extensible, open architecture. Integration with geographic information sys- tems (GIS) enhances the visualization of input data and model results. BASINS version 3 includes PLOAD, HSPF, and SWAT, all of which are all capable of modeling various BMPs. PLOAD is a nonpoint-source loading model. The HPSF BMP module can interface with an Access database of 34 standard BMPs, including detention ponds, infiltration systems, and manufactured systems. HSPF also allows custom BMPs to be specified and modeled. SWAT is capable of simulating a variety of agricultural practices, including tillage and pesti- cide application. GIS is rapidly becoming an indispensable tool in storm- water management. In the past, GIS packages were used mostly for the post-processing of model output, mainly as a visualization tool. However, a number of researchers are beginning to integrate stormwater modeling into GIS, taking advantage of the inherent libraries, routines, and underlying programming interfaces of GIS packages. Xue et al. (1996) created a mechanism-based BMP model and successfully linked the model to Arcview 2.1 using ArcView’s built-in macro language AVENUE. The integrated model had a user- friendly interface, and a sample simulation was provided to illustrate the functionality of the tool. Melancon et al. (2000) outline the application of a GIS- based BMP model to simulate the use of BMPs in the mitiga- tion of bacteria-contaminated runoff. The model was devel- oped using Arc/Info and ArcView GIS software packages. Results from the model were used to determine the source of bacteria loads, and the model was found to be capable of esti- mating flow and load conditions with reasonable accuracy. BMP models are used mainly in the context of water qual- ity or flood control design; however, there are models that incorporate additional optimization parameters such as cost. Heatwole et al. (1985) present a model capable of analyzing the cost and water quality implications of selectively apply- ing various BMPs throughout a basin and comparing differ- ent scenarios. Using the model, the authors discovered that the cost for the maximum level of BMP treatment was four times as high as the cost for a 90% improvement in water quality using the four most economical BMPs. 84 Detention systems and infiltration systems appear to be the most extensively studied BMPs in terms of BMP modeling. There are numerous studies that have used existing software packages like SWMM, HPSF, and BASINS to model deten- tion ponds. Boyd et al. (1994) discuss the use of MOUSE in on-site detention design in the City of Wollongong, Aus- tralia. Wu and Ahlert (1985) discuss a trajectory model that is used to investigate sedimentation processes in detention ponds. The model used a normal distribution of sediment particle sizes to calculate sediment trapping efficiencies for various length-to-width ratios. Wong et al. (1996) discussed the use of the P8 Urban Catchment Model and dynamic pro- gramming in detention pond design. The methodology was applied to the Marley Creek watershed to obtain the most economical system of detention ponds that would meet water quality and flow goals set for the watershed. The Detention Outlet Channel Dynamic Program (DOCP) is an optimiza- tion model that helps determine least-cost locations and sizes of detention basins. Bennett (1983) demonstrates the capa- bilities of DOCP in an application of the model to the Brays Bayou Watershed in Houston. In a study presented by Lam and Palmer (1996), two existing detention facilities that were constructed originally for flood-control purposes were retrofitted to provide water quality treatment. The sediment removal capabilities of the retrofitted ponds were analyzed using QUALHYMO and STORM. Dynamic wave routing was accomplished with OTTHYMO, QUALHYMO, and BOSS-DAMBRK. Petterson et al. (1998) used data from an open stormwater detention pond to verify the FEM model. The authors found that the 2-dimensional and 3-dimensional analysis was in agreement with the observed data. Modeling infiltration systems has been an active area of study. James et al. (1997) present a discussion of the use of SWMM and HSPF shallow groundwater routines as a foun- dation for developing alternate approaches to infiltration BMP modeling. To gain a better understanding of the clogging phenomena of infiltration BMPs, Gautier et al. (1999) inves- tigated an infiltration basin and two groundwater recharge basins. Other studies that have made attempts to model porous pavement structures include Loughreit et al. (1996) and Goforth (1983). Morita et al. (1996) describe a conjunctive flow model that overcomes some of the simplifying assump- tions made in the development of other models. The model accounts for the interaction between surface flow and sub- surface flow, and the authors provide an example to demon- strate the capabilities of the model. Debo (1994) discuss the development of a model used in the design and analysis of infiltration basins, infiltration trenches, dry wells, porous pavement, and vegetated swale with check dams. Bishop and Scheckenberger (1994) describe the use of HPS-F in the design of a constructed wetland, which was to serve as a BMP to mitigate runoff for a proposed freeway interchange. The HPS-F analysis provided information that

was used in the bathymetric design and also the plant species distribution for the wetland. SLAMM, a recently calibrated urban runoff model, was used to compare the cost-effectiveness of using combina- tions of source area and regional stormwater treatment prac- tices (Bannerman et al., 2003). Model results indicated that individually the Delaware Perimeter Sand Filter, Stormcep- tor, Multi-Chamber Treatment Tank, bioretention, porous pavement, and infiltration trenches could reduce the solids load to Lake Wingra by 7 to 19% and that high-efficiency street sweeping could reduce annual solids load by 17%. By modeling various street sweeping–treatment control prac- tices, it was found that nine different combinations would be able to achieve the 40% reduction goal. For example, a 42% reduction in solids load to Lake Wingra is estimated for the combination of high-efficiency street sweeping on all the streets and Delaware Perimeter Sand Filters on all the park- ing lots. Alternatively, the 40% reduction could be achieved by using regional detention ponds with a total of 20 acres of permanent pool area. However, it was estimated that the annual cost of the source area practices range from about $573,000 to $1,504,000, while the range for the detention ponds is $963,000 to $1,840,000, assuming a 20-year life span. The least expensive combination of source area prac- tices would only increase the annual stormwater utility bill for the Madison taxpayers by about $6, while the most likely detention pond alternative will increase the utility bills by about $18. Through the use of the SIMPTM computer simulation, Kurahashi and Associates (1997) evaluated the effectiveness of new high-efficiency pavement sweepers in combination with conventional sediment-trapping catch basins to determine if the combination technology provided pollutant-reduction benefits that were comparable to those of wet vaults. The results of the simulation study showed that pollutant removals obtained with high-efficiency sweeping at a weekly frequency in combination with normal catch basin inlets cleaned annu- ally are comparable to removals obtained by wet vaults. How- ever, the model assumes that all of the sources of pollutants can be described by a build-up–wash-off function, which is not true. Therefore, the findings regarding a BMP that acts on this function can overstate grossly the BMP performance. In fact, high-efficiency sweeping appears to be more effec- tive than wet vaults in the removal of highly dissolved pol- lutants (copper, zinc, and phosphorus), but wet vaults appear more effective than high-efficiency sweeping in the removal of TSS and sediment-bound pollutants such as lead. The use of high-efficiency pavement sweepers in combination with conventional sediment-trapping catch basins would result in substantial savings for the Port of Seattle compared to the use of wet vaults (estimated lifecycle costs of $2 million for high-efficiency sweepers in combination with conventional sediment-trapping catch basins versus $18 million for wet vaults) if their treatment were equal. Street sweeping has never 85 shown the pollutant reductions in outfall discharges that the modeling of this type has—a result that is likely due to the pollutant source assumption. 3.2.10.1. Identification of Research Needs With regard to the modeling of BMP unit processes, sedi- mentation and infiltration appear to be well covered in the lit- erature. However, other BMP water quality treatment unit processes—such as sorption processes (absorption and adsorp- tion), biodegradation, photolysis, and volatilization—still need to be studied further before reliable BMP performance models can be developed. There also is a lack of information on the modeling of BMP treatment trains. A better under- standing of BMP longevity and of the decrease in treatment efficiency as a function of time is required, so that the opti- mization models used in selecting cost-effective BMP systems can provide better estimates of BMP lifetime costs and bene- fits. Another area that could be explored addresses how the sources of pollutants are represented in models. Many models still use a build-up–wash-off approach as the only way the pol- lutants get into stormwater; however, this approach should be used with caution as it can lead to faulty results if the BMP acts directly on that function. With any attempt to predict or model environmental processes, there is a general need for accurate and represen- tative data for parameter estimation and model calibration. Thus, the ability to measure and analyze accurately unit treat- ment processes is essential for the development of reliable models that can evaluate or predict BMP performance. As Sansalone (2000) stated in his TRB Millennium Paper, “. . . the future for ecologically sustainable transportation will require the ability to gather sufficient temporal and spatial measurements for increasingly sophisticated and integrated hydrologic, hydraulic, and water quality treatment models.” The development of a review of modeling approaches and guidance on their selection and application is a potential research topic. 3.2.10.2. Primary References Bannerman, R., Fries, G., and J. Horwatich. Source Area and Regional Stormwater Treatment Practices: Options for Achiev- ing Phase II Retrofit Requirements in Wisconsin. Proc., of the National Conference on Urban Stormwater-Enhancing Pro- grams at the Local Level, Chicago, IL (February 17–20, 2003) pp. 12–19. Baxter, R. Modeling Tools for the Stormwater Manager: An Overview of EPA’s Widely Used Modeling and Assessment Tools. Stormwater (March–April 2002) www.forester.net/sw_ 0203_modeling.html Bennett, M. S. Dynamic Programming Model for Determining Optimal Sizes and Locations of Detention Storage Facilities. Proc., from the University of Kentucky Urban Hydrology,

Hydraulics and Sediment Control Symposium (July 1983) pp. 461–468. Bishop and Scheckenberger. HSP-F Simulation of a Constructed Wetland Stormwater BMP for Urban Highway Runoff. Proc., 1994 Stormwater and Water Quality Management Modeling Con- ference (March 1994). Boyd, M. J., Carr, R., and G. Mackintosh. Modeling On-site Deten- tion Storage in an Urban Catchment Using MOUSE. Proc., Inte- grated Urban Storm Runoff-7th European Junior Scientist Work- shop (June 1994) pp. 173–178. Debo, T. N. Computer Model for Infiltration System Design. Com- puting in Civil Engineering, American Society of Civil Engi- neers, New York, NY (1994) pp. 241–248. Gautier, A., Barraud, S., and J. P. Bardin. An Approach to the Char- acterization and Modeling of Clogging in Stormwater Infiltration Facilities. Proc., 8th International Conference on Urban Storm Drainage (August 1999) pp. 1007–1015. Goforth, G. An Advancement in Hydraulic Modeling of Porous Pavement Facilities. U.S. Environmental Protection Agency (Jan- uary 1983) pp. 237–254. Hayes, J. C., Barfield, B. K., Holbrook, F., Gillespie, J., Fersner, J., and B. Bates. A Model for Assessing the Impact of BMPs on Water Quality. Stormwater (September–October 2003). Heatwole, C. D., Bottcher, A. B., and L. B. Baldwin. A Model for Assessing the Cost-Effectiveness of Agricultural BMP Imple- mentation Programs on Two Florida Basins. U.S. Environmen- tal Protection Agency (January 1985) pp. 257–264. Huber, W. C. BMP Simulation Using the U.S. EPA Stormwater Management Model (SWMM). Proc., 7th International Confer- ence on Urban Storm Drainage (September 1996) pp. 1629–1634. James, W., and J. A. Ulan. Towards a Shallow Groundwater Routine for Modeling Infiltration BMPs in Urban Stormwater Models. Advances in Modeling the Management of Stormwater Impact, Vol. 6 (February 1997). Kurahashi and Associates, Inc. Port of Seattle-Stormwater Treat- ment BMP Evaluation. Port of Seattle, Pier 66 (1997). Lam, A. S., and R. M. Palmer. Modeling Retrofitted Extended- Detention Wet Ponds and Wetland Pockets. Proc., 1996 Storm- water and Water Quality Management Modeling Conference (February 1996) pp. 407–428. Loughreit, F., Barraud, S., Cres, F. N., and E. Alfakih. A Concep- tual Model for the Design and Simulation of Porous Pavements. Proc., 7th International Conference on Urban Storm Drainage (September 1996) pp. 533–538. Melancon, P. A., Maidment, D. R., and M. E. Barrett. Modeling Non-Point Source Pollution and the Impact of Best Management Practices Using a Geographic Information System. Proc., Water- shed 2000 Management Conference (July 2000). Morita, M., Nishikawa, R., and B. C. Yen. Application of Con- junctive Surface-Subsurface Flow Model to Infiltration Trench. Proc., 7th International Conference on Urban Storm Drainage (September 1996) pp. 527–532. Petterson, T. J. R., German, J., and G. Svensson. Modeling of Flow Pattern and Particle Removal in an Open Stormwater Detention Pond. Hydra Storm 1998 Proceedings, Combining 3rd Interna- tional Symposium on Stormwater Management and 6th Interna- tional Conference on Hydraulics in Civil Engineering (Septem- ber 1998) pp. 63–70. Sansalone, J. J. The Role of Water in Ecologically Sustainable Trans- portation. In Transportation in the New Millennium, Commit- 86 tee on Hydrology, Hydraulics, and Water Quality, TRB, National Research Council, Washington, DC (2000). Wong, K. S., Schaeffer K., and T. Tapley. A Dynamic Program- ming Approach to Stormwater Management Systems Design. Proc., Watershed 1996 (1996) pp. 435–439. Wu, J., and R. C. Ahlert. A Trajectory Model for Analyzing Sedi- ment Trapping Efficiencies in Stormwater Detention Basins. Proc., U.S. EPA Sanctioned Stormwater and Water Quality Man- agement Modeling Conference, Toronto, ON (December 1985) pp. 257–284. Xue, R. Z., Bechtel, T. J., and Z. Chen. Developing a User-Friendly Tool for BMP Assessment Model Using a Geographic Informa- tion System. Proc., Symposium on GIS and Water Resources, Fort Lauderdale, FL (September 1996). 3.2.11. Maintenance and Longevity Twelve state DOTs have conducted studies and prepared reports on the maintenance aspects of stormwater management measures during construction, as well as at DOT facilities. A review of the literature pertinent to the role of mainte- nance in BMP performance and longevity shows that numerous studies have attempted to link BMP performance and longevity to maintenance practices. All BMPs, both pro- prietary and nonproprietary, require regular and nonroutine maintenance in order to perform well. The frequency and extent of maintenance depend on pollutant loading and the availability of pretreatment. Regular maintenance activities include removing accumulated materials and cleaning inlets and outlets. Nonroutine maintenance may include structural repairs and revegetation (Livingston et al. 2000). The Watershed Management Institute, in cooperation with the EPA, published the document Operation, Maintenance, and Management of Stormwater Systems that includes guidance on BMP maintenance practices and costs as well as design information (Livingston et al., 1997). This document is one of the most comprehensive BMP maintenance guidance documents available. However, because of the large number of new and innovative BMPs, as well as variations to exist- ing BMPs, there have been several more case studies since its publication. Therefore, the topic area of BMP mainte- nance practices and costs is in need of further research. Areas of interest and possible knowledge gaps related to the effects of maintenance on BMP performance and longevity include • Recommended frequency of maintenance for various BMPs, • Determination of the most cost-effective maintenance activities, • Maintenance requirements for natural and constructed wetlands, • Cost and benefit analysis of BMP maintenance, • Sediment toxicity as a function of type and frequency of maintenance, and • Disposal of maintenance waste products such as sediment.

The current state of maintenance practices of various munic- ipalities have been the subject of a number of studies. An extensive survey of more than 800 stormwater structures was conducted in four North Carolina cities in an effort to evaluate stormwater maintenance practices and needs (Roenigk et al., 1992). Culverts, inlet devices, channels, detention ponds, wet ponds, and infiltration systems were evaluated. Stormwater officials in 88 North Carolina cities were interviewed on the phone about maintenance issues. The results of the study indicated that maintenance practices were adequate, with the exception of detention facility maintenance. Most of the surveyed systems were designed to operate primarily at flood control capacities; hence, maintenance requirements are expected to increase with the increasing prominence of water quality issues. The phone interviews revealed that the maintenance activ- ities performed, in order of frequency from highest to low- est, were mowing, inlet cleaning, facilities inspections, and then sediment removal at detention facilities. The final com- ponent of the study was conducting interviews with 25 storm- water management experts, who all agreed on the importance of inspections, mowing, and sediment removal, but who did not agree as to how frequently these maintenance activities needed to be performed. King County performed a survey of 17 wet ponds and 33 biofiltration swales to assess the state of these water qual- ity facilities (King County, 1995). Results of the survey indi- cated that because of poor design, construction problems, and inadequate maintenance practices, only 35% of the wet- ponds and 28% of the biofiltration swales were working prop- erly. The study attributed the unsatisfactory condition of the BMPs to the novelty of the stormwater facilities and to a lack of understanding about the effort required to sustain water quality facilities in decent working condition. The Caltrans BMP Retrofit Program evaluated the costs of acquisition, operation, and maintenance of 39 BMPs from 12 different BMP categories (Currier et al., 2001). The study estimated the annual maintenance requirements for sand fil- ters, extended detention basins, infiltration basins, biofil- tration strips, swales, and wet basins at 93 hours, 136 hours, 193 hours, more than 200 hours, and 570 hours, respectively. The private sector increasingly is adopting the use of BMPs for commercial, industrial, and residential development applications. Inspection and maintenance of BMPs are often the responsibility of the property owner. The City of Lacey in Washington State attests to the effectiveness of inspec- tions and education coupled with field activities as opposed to pure enforcement (Hielema, 2001). Infiltration facilities appear to be the most prone to failure because of inadequate maintenance practices. Consequently, many of the studies on BMP maintenance have examined infiltration facilities. The study by Nozi et al. (1999) exam- ined various infiltration facilities (infiltration inlet, infiltration trench, porous asphalt pavement, and an infiltration well) in Japan, evaluated the effects of maintenance, and demonstrated 87 that in most cases infiltration BMP performance can be improved significantly with maintenance. A comparison of infiltration facilities that had been in use for up to 10 years showed a decline of infiltration capacity with time. To evalu- ate the effects of washing and cleaning, infiltration facilities in four municipalities (Tokyo, Chiba, Nagoya, and Hamamatus) that were more than 10 years old were maintained and assessed. With the exception of one infiltration trench in Nagoya, all of the infiltration facilities showed a marked improvement in infiltration capacity. A study of four porous pavement systems, including pavers with infiltration joints, porous concrete pavers with filter lay- ers, greened porous pavers, and pavers with greened infiltra- tion joints was performed to determine the pollutant retention abilities of the various systems (Dierkes et al., 2002). All four systems demonstrated high pollutant retention capabilities, but the system with infiltration joints was relatively less effi- cient. Field evaluation of a 15-year-old piece of porous pave- ment revealed no impact to soil or groundwater. A device to alleviate clogging in porous pavement was tested successfully in a school yard. As a result of the cleaning, infiltration rates were increased from 1L/(s.ha) to 1500 L/(s.ha). Researchers concluded that porous pavements do get clogged and that the device developed in this study seemed suitable for mainte- nance of porous pavers. To evaluate the impacts of accumulated sediment on nutri- ent removal efficiencies in a pond, a field study was con- ducted on a submerged biofilter (Mothersill et al., 2000). Substantial removal of suspended solids (97%) from the influent stormwater resulted in a significant accumulation of sediment in the biofilter, which interfered with the system’s main treatment objective—removing soluble nutrients through bacterial assimilation. Removal efficiencies of total organic carbon and suspended orthophosphate were found to decrease with time; however, the removal efficiency for ammonium nitrate (64%) appeared independent of time or sediment accu- mulation. Sediment accumulation was attributed to infrequent backwashing of the filtration unit. Other studies have examined and compared the effects of maintenance on different classes of BMPs. A paper by Botts et al. (1996) presented maintenance requirements and lon- gevity estimates for four standard BMPs, including a water quality inlet, an infiltration trench, a wet detention pond, and a sand filter. Wet detention ponds and water quality inlets are shown to have long life spans, with well-designed detention ponds operating as designed for 20 years or more and 95% of water quality inlet installations operating as designed for up to 5 years. Infiltration trenches have short life spans with less than 50% of installation failing in fewer than 5 years. Proper design and regular maintenance can prolong the life of infiltration facilities to well over 5 years. A study by Galli (1992) in Prince George’s County, Mary- land, evaluated the performance and longevity of 11 types of BMPs. The BMPs studied included infiltration trenches and basins, dry wells, porous pavement, vegetated swales, extended

detention dry ponds, wet ponds, constructed marshes, pocket wetlands, oil and grit separators, and dry ponds. Assessment criteria used in the study included design strengths and weak- nesses, maintenance issues, and environmental considera- tions for each of the 156 sites included in the study. The results of the study suggested that infiltration basins, porous pavement, grass filters, swales, and “pocket” wetlands gen- erally required modifications or improvements in order to provide reliable pollutant removal. A King County study evaluated the effects of mowing on the performance of vegetated swales (Colwell et al., 2000). Two mowing regimes—mowing at both the beginning and at the end of the growing season and mowing only at the end of the growing season—were evaluated to determine impacts to swale treatment efficiencies. TSS and turbidity mitigation were significantly higher for the unmowed swale showing that mowing did not provide increased treatment. The two mowing strategies were found to be equivalent with respect to water quality benefits. The authors cautioned that the test systems may not be representative of all swales. The performance of BMPs commonly used in public works practices—such as water quality inlets, sedimentation man- holes, and catch basin inserts—also are dependent on the extent and frequency of maintenance. Maintenance-related information on public works practices are included under the Gross Pollutant Removal and the Drain Inlet/Gross Pollutant Studies sections of this report. Details of studies presented by England and Rushton (2003), Sedrak et al. (2001), Lipp- ner and Moeller (2000), and Dammel et al. (2001) are all rel- evant to maintenance of public works-related BMPs. 3.2.11.1. Identification of Research Gaps and Needs Based on the literature review, it is apparent that there is substantial information on maintenance practices of BMPs and how these practices affect performance. A compilation of the results of these studies in an updated BMP operations and maintenance manual, such as the document by Livingston et al. (1997), may be a potential research need. Another need is for guidance on estimating maintenance frequencies based on influent characteristics and site conditions. BMP maintenance costs frequently are not factored in dur- ing the initial planning and BMP selection phases of construc- tion projects. There appears to be a need for guidance on esti- mating the lifecycle costs that account for the maintenance required for continually functioning and efficient BMPs. A final need is for further development of methods to increase the longevity and to minimize maintenance requirements of infiltration BMPs—such as the use of presettling basins or the use of PAMs to maintain infiltration rates. Evaluations of sediment toxicity as a function of mainte- nance frequency and methods for disposing or reusing BMP maintenance-generated wastes would be helpful. 88 3.2.11.2. Primary References Botts, J., Allard, L., and J. Wheeler. Structural Best Management Practices for Storm Water Pollution Control at Industrial Facili- ties. Proc., Watershed 1996 (1996) pp. 216–219. Colwell, S. R., Horner, R. R., and D. B. Booth. Characterization of Performance Predictors and Evaluation of Mowing Practices in Biofiltration Swales. King County Land and Water Resources Division Report, Center for Urban Water Resources Manage- ment, Department of Civil and Environmental Engineering, Uni- versity of Washington, Seattle (2000). Currier, B., Taylor, S. M., Borroum, Y., Friedman, G., Robison, D., Barrett, M., Borroum S., and C. Beitia. California Department of Transportation BMP Retrofit Pilot Program. Presented at 80th Annual Meeting of the Transportation Research Board, Wash- ington, DC (January 7–11, 2001). Dierkes, C., Kuhlmann, L., Kandasamy, J., and G. Angelis. Pollu- tion Retention Capability and Maintenance of Permeable Pave- ments. Global Solutions for Urban Drainage, Proc., 9th Interna- tional Conference on Urban Drainage, Portland, OR (September 8–13, 2002) pp. 444–445. Galli, J. Analysis of Urban BMP Performance and Longevity in Prince George’s County, Maryland. Report, Prince George’s County Department of Environmental Resources, Watershed Protection Branch, MD (August, 1992) 203 pp. Hielema, E. J. Private Facility Inspection and Maintenance: ‘Deluxe With Bacon’ or ‘Maintenance Lite’? Stormwater (September– October 2001) www.forester.net/sw_0109_private.html#what King County Department of Public Works. Evaluation of Water Quality Ponds and Swales in the Issaquah/East Lake Sam- mamish Basins. Surface Water Management Division, Seattle, WA (1995) 75 pp. Livingston, E. H., Baldwin R., and B. Clevenger. Lessons Learned About Successfully Using Infiltration Practices. Proc., National Conference on Tools for Urban Water Resource Management and Protection, Chicago, IL (February 7–10, 2000) pp. 141–161. Livingston, E. H., Shaver, E., Skupien, J. J., and R. R. Horner. Operations, Maintenance, and Management of Stormwater Sys- tems. U.S. Environmental Protection Agency Report CS82361- 01-0, Watershed Management Institute, Washington, DC (1997). Mothersill, C. L., Anderson, B. C., Watt, W. E., and J. Marsalek. Biological Filtration of Stormwater: Field Operations and Main- tenance Experiences. Water Quality Research Journal of Canada, Vol. 35, No. 3 (2000) pp. 541–562. Nozi, T., Mase, T., and K. Murata. Maintenance and Management Aspects of Stormwater Infiltration Systems. Proc., 8th Interna- tional Conference on Urban Storm Drainage, Vol. 3, Sydney, Australia (August 30–September 3, 1999) pp. 1497–1504. Roenigk, D. J., Paterson, R. G., Heraty, M. A., Kaiser, E. J., and R. J. Burby. Evaluation of Urban Stormwater Maintenance in North Carolina. Report No. 267, Department of City and Regional Plan- ning, University of North Carolina, Chapel Hill (1992) 197 pp. 3.2.12. Use of Toxicity and Biological Indicators in Performance Evaluations Toxicity and other biological indicators—such as bioavail- ability, species diversity, and biomass—are underused meth- ods for evaluating BMP performance. This subject is not to

be confused with a full receiving waters biological impact analysis. Rather, use of indicators entails applying toxicity, bioavailability, and biological communities as metrics for BMP performance. Potential research questions include • How can biological indicators be used to assess BMP performance, and what are the limitations to their use and interpretation? • How are toxicity and bioavailability reductions related to the reduction or speciation of chemical constituents? • Which indicator organisms are most appropriate for evaluating BMP performance? Pitt et al. (1991) investigated the control of stormwater toxicants through conventional treatment processes. Twelve sheet flow samples were collected from the source areas that were found previously to produce the most toxic storm runoff waters. These areas were automobile service areas, industrial parking and loading dock areas, and automobile salvage yards. The samples were subjected to a variety of benchscale treata- bility tests, including settling columns, sieving screens, mem- brane filters, aeration, photo degradation, aeration and photo degradation combined, floatation, and alum addition. Toxicity changes were monitored using the Microtox bio- assay test. The benefits of the treatment processes varied for the different samples. However, some of the treatment processes consistently provided the greatest toxicity reduc- tion. The most beneficial treatment tests included settling for at least 24 hours (generally 40–90% reduction), screening through at least 40-micron screens (20–70% reduction), and aeration or photo degradation for at least 24 hours (up to 80% reduction). The floatation tests produced floating sample lay- ers that generally decreased in toxicity with time. However, the benefits were quite small (less than 30% reduction). Alum additions substantially reduced the turbidity of the samples but the changes in toxicity were highly irregular. The Port of Seattle tested four filtration media in con- trolled laboratory experiments to determine their effective- ness for concurrent metals removal and toxicity abatement in synthetic stormwater (Tobiason et al., 2003). Media tested included commercially available leaf compost (CSF®) media, a zeolite–perlite mix, and a polyamine sponge, as well as the recently developed citric acid modified soybean hull media. Toxicity was assessed using acute Ceriodaphnia dubia (48-hr) bioassays. Results indicated that the CSF® media removed up to 75% of the zinc and reduced toxicity signifi- cantly for influent concentrations of up to about 300 ppb zinc. The soybean hull material removed 80–99% of the zinc over all influent concentrations, though it reduced pH to toxic lev- els. After pH adjustment, the effluent from the soybean mate- rial was nontoxic over all concentrations tested (survival was 100% in pH-adjusted effluent samples). Augmenting the soy- bean material with leaf compost media or activated carbon effectively buffered effluent pH to circumneutral ranges. Other media tested removed modest amounts of zinc and 89 were capable of sufficiently reducing toxicity only in the lowest concentrations tested; some media appeared to gener- ate toxicity (which may have been due to reductions in pH or hardness). By studying algal communities in two stormwater man- agement ponds, Rouge Pond and Harding Pond, Olding (2000) noted that impacts to aquatic biota decreased as stormwater passed through ponds. The greatest disturbances to biologi- cal communities were observed in the sediment forebay area for both ponds. The author attributed the absence of blue- green algae populations in both ponds to the hydraulics of the ponds, despite nutrient-rich conditions, and suggested that stormwater ponds can be engineered to limit nuisance algal communities. The reduction of impacts to biological com- munities observed in the pond translated to a reduction of impacts to receiving water biological communities. The use of biological indicators for evaluating BMP per- formance is a relatively new method that is gaining popular- ity among stormwater regulators. Biological indicators such as those used in toxicity bioassays have been used exten- sively for evaluating potential impacts to and the contamina- tion of receiving water systems (see EPA’s Biological Indi- cators of Watershed Health: www.epa.gov/bioindicators). It is not difficult to extend this knowledge base to the evalua- tion of BMP performance, but some issues with traditional toxicity testing methods should be considered. Burton et al. (2000) points out that traditional toxicity tests may not pro- duce reliable conclusions when used to detect the adverse effects of fluctuating stressor exposures, nutrients, suspended solids, temperature, ultraviolet light, flow, mutagenicity, car- cinogenicity, teratogenicity, endocrine disruption, or other important subcellular responses. This inability to predict effects is largely a result of the complex biological response patterns that result from various combinations of stressor magnitudes, duration, and frequency between exposures, as well as from the interactions of stressor mixtures, such as synergistic effects of certain pesticides, metals, and temperature. In watersheds receiving multiple sources of stressors, accurate assessments should define spatial–temporal profiles of exposure and effects using a range of laboratory (such as WET tests) and novel in situ toxicity and bioaccumulation assays, with simultaneous characterizations of physicochemical conditions and indige- nous communities. Beginning in May 2003, the Louisiana DOT embarked on the research project Transport, Treatability, and Toxicity of Highway Stormwater Discharged into Receiving Waters across Louisiana (http://rip.trb.org). The primary objectives of this research are (1) characterization of highway storm- water based on hydrology, pollutant loadings, toxicity, and rainfall quality; (2) comparison of standard tests for storm- water characterization; (3) quantification of pollutant load- ings as a function of hydrologic parameters and traffic char- acteristics; and (4) assessment of treatment alternatives. Part of this research also will include documenting toxicity reduc- tions at three experimental sites: a site near Shreveport at the

I-220 bridge over Cross Lake, a site in Baton Rouge at the I-10 bridge over City Park Lake, and a site in New Orleans at the I-10/I-610 junction over the 17th Street canal. 3.2.12.1. Identification of Research Needs Based on the scarcity of studies that use biological indicators for BMP performance assessment (as well as for general high- way runoff characterization), it appears that this entire topic area is a research need. In fact, the top two research needs iden- tified by GKY and Associates in the original NCHRP Project 25-20 report were to (1) identify and develop regional aquatic biological indicators to assess impacts of highway runoff and (2) research methods and develop protocols for assessing the toxicity of highway runoff. This current effort demonstrates agreement that research in this area is still needed and should be expanded to include the use of biological indicators to assess BMP performance in terms of toxicity reduction. 3.2.12.2. Primary References Burton, G. A. Pitt, R., and S. Clark. The Role of Traditional and Novel Toxicity Test Methods in Assessing Stormwater and Sed- iment Contamination. Critical Reviews in Environmental Science and Technology, Vol. 30, No. 4 (2000) pp. 413–447. Olding, D. D. Algal Communities as a Biological Indicator of Stormwater Management Pond Performance and Function. Water Quality Research Journal of Canada, Vol. 35, No. 3 (2000) pp. 489–503. Pitt, R. E., Ayyoubi, A., Barron, P. F., and R. Field. The Treatabil- ity of Urban Stormwater Toxicants. New Technologies in Urban Drainage, London, Elsevier Applied Science (1991). Tobiason, S. A., Login, L. R. J., and C. Nickerson. Stormwater Fil- tration Media Testing for Metals Removal and Toxicity Reduc- tion. Port of Seattle, Parametrix, Inc., Taylor Associates, WA. 3.2.13. Public Perception and Aesthetics Few studies have attempted specifically to evaluate the importance of public perception and aesthetics in BMP selec- tion, design, implementation, and performance. However, a literature review pertaining to this subject shows that a fair number of BMP evaluation studies make mention of aesthet- ics or public perception in one way or another. Areas of inter- est and possible knowledge gaps in this area include • Public perception of BMPs in terms of impacts on private-property values; • Public perception of BMPs in terms of impacts on pub- lic safety; and, • The role of aesthetics in the design, selection, imple- mentation, and public acceptance of BMPs. According to Frederick et al. (1996), residential areas with open water areas are sometimes avoided by parents with young 90 children for fear of exposing their children to the risk of drowning. Poorly maintained ponds can become unsightly and odorous and provide a breeding ground for mosquitoes and other parasites. On<the contrary, well-designed, aesthet- ically-pleasing properties can cause property values to increase, vacancy rates to lower, and tenant turnover to decrease in rental properties. According to Baxter and Mulamoutti (1985), 49% of the residents in a neighborhood where retention ponds were constructed believed that the lakes had a positive impact on property values. The benefits of the ponds were perceived to be aesthetics and attraction of potential residents and recreational opportunities. A functional, aesthetically pleasing wetland designed near a golf course is discussed by White and Meyers (1997). The nitrogen, phosphorus, and suspended sediment removal effi- ciencies of the wetland were estimated at up to 60%. Accord- ing to the authors, proper vegetation selection can result in a beautiful wetland design that can provide bird habitat and educational opportunities. Well-designed and implemented public education programs can improve significantly public perception of stormwater pollution control programs and BMPs. An ongoing Caltrans study (PERS) is attempting to implement and to evaluate the effectiveness of a public education program targeted at reduc- ing litter (Caltrans, 2002). The details of this study are pre- sented in the Gross Pollutant Removal Section of this report. 3.2.13.1. Identification of Research Needs It is seemingly evident from a literature review that not very much work has been done on the use of aesthetics and public perception as a benchmark of BMP performance. Existing research may be less applicable to highway environments. Studies that rank BMPs in order of public performance and provide insight as to how to improve public perception of various types of BMPs would be a valuable addition to the collection of existing BMP evaluation studies. Research that quantifies the impacts of various types of BMPs on public property values could provide a useful tool for public educa- tion programs. 3.2.13.2. Primary References Baxter, E., and G. Mulamoutti. A Study of Residential Stormwater Impoundment—Perceptions and Policy Implications. American Water Resources Association Publication (1985). Caltrans., Caltrans Public Education Litter Monitoring Study 2001–2002. Preliminary Report CTSW-RT-02021 (2002). Frederick, R., Goo, R., Corrigan, M. B., Bartow, S., and M. Bil- lingsley. Economic Benefits of Urban Runoff Controls. Proc., Watershed 1996 (1996) pp. 389–392. White, K. D., and A. L. Meyers. Stormwater Management and Aes- thetics Using Wetlands. Proc., 24th Annual Water Resources Planning and Management Conference (April 1997).

3.2.14. Economic Analysis and Assessment Evaluating BMPs on the basis of cost is an integral part of the BMP selection process. Initial capital costs and some- times operation and maintenance costs can become the key controlling factors that dictate which BMPs are selected and whether projects get constructed. The tools and methods used in cost estimates therefore play a prominent role in stormwater management. According to Heaney et al. (2002), most of the cost-estimation methods are based on regression equations. Cost-estimation techniques can be improved through addi- tional funding and research and through the use of available technology. Potential research needs and knowledge gaps in the economic analysis and assessment of BMPs, some of which were suggested by Heaney et al. (2002), include • Quantification of the benefits of urban storm systems; • Consideration of receiving water impacts in cost–benefit analysis; • Development of cost–benefit evaluation methodologies for nonstructural BMPs; • The availability and application of flow and water qual- ity data to BMP cost–benefit analyses; • Inclusion of land use data into cost optimization analy- ses; and • Cost–benefit analysis of BMP treatment trains. Another important consideration is the assessment of cost differentials. Costs for similar needs (e.g., landscaping and maintenance) for many BMPs may have been incurred already or may be avoided (e.g., reduction of pipes and inlets via the use of biofilters). Cost studies may be misleading if these potential cost offsets are not assessed. A review of the literature pertaining to BMP cost estima- tion reveals a number of documents that provide BMP cost information. FHWA (2000) provided a table of relative BMP costs. Structural BMPs that were assigned relatively high capital costs included underground sand filters and organic media filters; detention tanks, underground sand filters, organic media filters and oil-grit separators were assigned high rela- tive operation and maintenance costs. Low-cost structural BMPs included treatment systems like bioretention, deten- tion ponds, vegetated swales, vegetated filter strips, and porous pavement. Relative costs for new innovative BMPs range from moderate to high for systems such as alum injection, MCTT, biofilters, and vegetated rock filters. Heaney et al. (2002) present a comprehensive collection of tables and equations for estimating the cost of drainage struc- tures, including BMPs. Also, the document contains a review of literature pertinent to drainage system cost evaluation. Cost information for BMPs includes costs for porous park- ing, swales, and cost estimates for streets with swales and porous pavement. Yu and Stopinski (2001) evaluated four ultra-urban BMPs consisting of a bioretention basin and three proprietary treatment systems (Isoilater, Stormceptor, and 91 Vortechs treatment systems). A comprehensive cost analysis was presented for each system. In order to compare the cap- ital costs of each treatment system, researchers evaluated the cost-per-volume served, the cost-per-volume served per year, and the cost-per-percentage of TSS removal. The bioreten- tion basin was found to be the most economical system of the four tested. The Vortechs system was not installed properly and hence provided unreliable results. Scott et al. (1999) evaluated and compared two flood con- trol mitigation systems: on-site detention (OSD) and on-site retention (OSR). Both flood control systems provided flood protection by attenuating peak flows of in-coming storms and discharging the effluent at lower flow rates. OSR outper- formed OSD in terms of cost efficiency and environmental benefits. Cost efficiency was based on the volume of site storage required to attain peak flow attenuation. Caltrans has assessed and approved more than 110 BMPs for use since 1996 and has included cost–benefit analysis as part of their assessment. Caltrans assessed the cost- effectiveness for each BMP in terms of its EUAC relative to a detention basin (Caltrans, 2003). A four-quadrant system was used as a tool to rate each BMP. The cost estimates were defined first by calculating the typical range of costs for con- structing or operating a BMP on a per-acre basis. The acre represented the drainage area served by the BMP. Operation and maintenance costs then were added, based on the design life of the BMPs. The EUAC for a particular BMP was esti- mated and compared qualitatively to that of a detention basin. If the EUAC was higher for the BMP than for a deten- tion basin, it was marked as a higher cost using the quadrant rating key. The benefit of the BMP was evaluated relative to the performance of a typical detention basin. If the con- stituent removal was greater than that of a detention basin, the BMP was marked as having a greater benefit. Using published literature and cost estimation guides, Sample et al. (2003) synthesized methods for estimating costs for BMPs such as detention, retention and infiltration basins, infiltration trenches, sand filters, and vegetated swales. According to the study authors, cost–benefit analysis method- ologies can be improved by considering additional parame- ters such as flow monitoring data, receiving water impacts, and the effects of streets and parking lots. Sear et al. (1996) explained the development of equations used to estimate BMP cost as a function of pollutant removal. The functions were used to evaluate nine alternatives for five stormwater treatment technologies in Lakeland, Florida. Pro- duction cost functions, with respect to TSS removal percent- ages, were developed for street sweeping, infiltration<sys- tems, wet ponds, dry ponds, and wetlands. The cost functions do not include property acquisition costs or operation and maintenance costs. The authors concluded that wet detention ponds and wetlands are more economical than dry detention bonds and curb-cut swales, if the cost of property acquisition is taken into account.

A number of simplifying assumptions are made and some external variables are overlooked in a lot of the available cost-estimation methods. Maintenance costs and longevity of BMPs are not considered in some BMP evaluation literature. According to England (1998), maintenance costs for retrofit projects often are neglected or underestimated. An evalua- tion of maintenance costs of BMPs—such as wet ponds, dry ponds, exfiltration and infiltration trenches, porous pave- ment, baffle boxes, inlet baskets, and sediment sumps—is presented by the authors, who recommended that mainte- nance needs be considered in the design and construction of retrofit projects in order to ensure that retrofits provide long- term pollutant removal. After evaluating baffle boxes and inlet devices, England (1998b) concluded that the tradeoff for the low initial cost of the evaluated BMPs is the perpetual maintenance expense. Baffle boxes are recommended for small to medium-sized drainage basins, while inlet devices are recommended for small flows and small drainage basins. BMPs can affect the market values of neighboring proper- ties, so another way to evaluate BMPs is to assess the eco- nomic benefits of implementing BMPs in residential and commercial areas. Frederick et al. (1996) presented a discus- sion about the potential increase in property value that can be gained through the construction of detention-type BMPs. Prices of homes situated close to a body of water tend to be significantly higher than comparable properties that are not near a body of water. In addition to environmental benefits, aesthetically pleasing BMPs can improve property values, lower vacancy rates of rental properties, and make properties easier to sell. As described in section 3.2.10., BMP Modeling, Bannerman et al. (2003) were able to find the most cost-effective combi- nation of high-efficiency street sweeping and treatment con- trol practices by using the SLAMM model to meet the TSS reduction goal of 40%. The annual cost of the source area practices was estimated (assuming a 20-year life span) to range from $573,000 to $1,504,000, while the range for deten- tion ponds was $963,000 to $1,840,000. The least expensive combination of source area practices would only increase the annual stormwater utility bill for the Madison taxpayers by about $6, while the most likely detention pond alternative would increase the utility bills by about $18. 3.2.14.1. Identification of Research Needs Based on the literature review addressing the economic analyses and assessment of BMPs, it is evident that there is cost estimation information for nearly all proprietary and most of the common nonproprietary structural BMPs. Cost regression equations have been developed for a number of BMP types that are based primarily on imperviousness, land use, and flow rates and volumes. However, lifecycle costs, opportunity costs, and externalities often are neglected in cost 92 estimation. As mentioned under the section Maintenance and Longevity, lifecycle costs account for the operations and maintenance requirements necessary to maintain lifetime BMP functionality and efficiency. Opportunity costs are the costs of land taken out of other uses and the costs of an alter- native conveyance system (which actually may be a net sav- ings in some cases). Externalities are the effects of produc- tion and consumption activities not directly reflected in the market, such as receiving waters protection and aesthetics (Willis and Finney, 1999). There is a need to develop BMP cost-estimation tools that account for land value, site con- straints, construction, operations, and maintenance, as well as receiving waters protection, aesthetics, and infrastructure savings on conventional drainage structures. Quantification of receiving waters protection requires the use of existing water quality, habitat, and bioassessment monitoring data for both the runoff and the receiving waters. With the possible exception of street sweeping, nonstruc- tural BMPs have been primarily overlooked. Costs associ- ated with public education, catch basin maintenance, and road side vegetation control activities would be helpful for the optimization and adequate allocation of stormwater man- agement funds. Finally, although there is an abundance of cost-evaluation methodologies for individual standard structural BMPs, BMP treatment trains and distributed BMP systems appear to have been neglected. There is a need for cost evaluations and com- parisons of BMP treatment trains, distributed BMPs, and large centralized regional BMP systems. 3.2.14.2. Primary References Bannerman, R., Fries, G., and J. Horwatich. Source Area and Regional Stormwater Treatment Practices: Options for Achiev- ing Phase II Retrofit Requirements in Wisconsin. Proc., National Conference on Urban Storm Water: Enhancing Programs at the Local Level, Chicago, IL (February 17–20, 2003) pp. 12–19. Caltrans. Caltrans New Technology Report. Report CTSW-RT- 03-010 (February 2003) 130 pp. England, G. Maintenance of Stormwater Retrofit Projects. Water Resources and the Urban Environment, Proc., 25th ASCE Water Resources Planning and Management Conference, Chicago, IL (1998). England, G. Baffle Boxes and Inlet Devices for Stormwater BMPs. Water Resources and the Urban Environment, Proc., 25th ASCE Water Resources Planning and Management Conference, Chi- cago, IL (1998b). Federal Highway Administration. Stormwater Best Management Practices in an Ultra-Urban Setting: Selection and Monitoring. Report No. FHWA-EP-00-002, U.S. Department of Transporta- tion, Washington, DC (May 2000). Frederick, R., Goo, R., Corrigan, M. B., Bartow, S., and M. Bil- lingsley, M. Economic Benefits of Urban Runoff Controls. Proc., Watershed 1996 (1996) pp. 389–392. Heaney, J. P., Sample, D., and L. T. Wright. Costs of Urban Stormwater Control. U.S. EPA Report No. 600/R-02 (January 2002) 105 pp.

Sample, D. J., Heaney, J. P., Wright, L. T., Fan, C. Y., Lai, F. H., and R. Field. Costs of Best Management Practices and Associated Land for Urban Stormwater Control. Journal of Water Resources Planning and Management, Vol. 129, No. 1 (2003) pp. 59–68. Scott, P., Santos, R., and J. R. Argue. Performance, Environmental and Cost Comparisons of On-Site Detention (OSD) and On-Site Retention (OSR) in Re-Developed Residential Catchments. Water Science and Technology, Vol. 39, No. 2 (1999) pp. 33–41. Sear, T. R., Bays, J. S., and W. G. Medley. Development of Cost- Effective Stormwater Treatment Alternatives. Proc., Watershed 1996 (1996) pp. 870–873. Yu, S. L., and M. D. Stopinski. Testing of Ultra-Urban Stormwater Best Management Practices. Final Report VTRC 01-R7, Vir- ginia Department of Transportation, Washington, DC (January 2001) 48 pp. 3.2.15. Vector Control The potential for structural BMPs to harbor and breed nuisance- and disease-causing organisms had received little attention until recently. A vector, as used in this section, refers to any organism that can transmit an infectious disease- causing organism to another living thing (Metzger et al., 2003). Because mosquitoes are ubiquitous and their life-cycle depends on humans and other warm-blooded animals, their potential to transmit infectious disease is high. Therefore many of the studies related to vectors in BMPs focus on mosquitoes (Metzger et al., 2003; VBDS, 2001). With the ever-increasing demand for BMPs, vector issues associated with BMPs could result in exponential increases in vector populations if not addressed. A review of the literature pertinent to the incidence of vectors in BMPs shows that wetlands appear to be the most targeted BMPs, and mosquitoes are the most targeted vector in studies. Caltrans established a comprehensive vector surveillance and monitoring study in cooperation with the Vector-Borne Disease Section (VBDS) of the California Department of Health Services. The objectives of the 2-year study were to develop vector abatement protocols and to recommend mod- ifications to Caltrans BMPs that would minimize their poten- tial to harbor vectors. VBDS monitored 37 structural BMPs at 31 sites with emphasis on mosquitoes. The study showed that BMP technologies that maintained permanent pools of standing water (i.e., multichambered treatment trains, contin- uous deflector separators, and wet basins) were more likely to support a large mosquito population. BMP technologies that drained completely (i.e., biofiltration swales, biofiltration strips, sand media filters, infiltration basins, infiltration trenches, drain inlet inserts, extended detention basins, and oil–water separators) were less likely to harbor vectors. Fac- tors that contributed to the incidence of vectors in BMPs include BMP design, BMP location, immediate and large- scale surroundings, nonstormwater discharges (such as irri- gation), and site maintenance. BMP design features to be avoided include the use of sumps, catch basins, or troughs that do not completely drain; loose riprap; automatic pumps or motors; and orifices that are prone to clogging. Recom- 93 mendations for pond-type BMPs included stocking permanent pools with Mosquito Fish (Gambusia affinis) and providing steep sideslopes to create a less desirable habitat for vectors. The combination of vegetation and permanent pools of stagnant water in wetlands makes wetlands prone to vector infestations. As a result numerous vector-related studies have targeted wetland locations. Studies that have evaluated the incidence and implication of vectors in wetlands include studies by Russell (1999a) and Russell et al. (1999b). Walton et al. (1999) examined the dispersal, survival, and host-seeking behavior of mosquitoes from a constructed wet- land in Southern California. The study showed that the lim- ited dispersal and the long survival of Culex erythrothorax were important factors in the development of large popula- tions at constructed wetlands. A study in Adelaide, Australia, evaluated 12 constructed wetlands in an attempt to under- stand the breeding habits of mosquitoes, especially those in urban constructed wetlands (Sarneckis, 2002). The study showed that wetlands with standing water, steep edges, and little emergent vegetation typically had fewer or no larval mosquitoes. Wetlands that supported large mosquito popula- tions typically had sheltered shallow water, isolated pools that limited predator access, poor water quality, and low mar- croinvertebrate diversities. The study concluded that well- designed wetlands were less likely to produce mosquitoes. MacLean (1995) presented mosquito management strate- gies for wetlands. The author suggested mosquito control strategies that included the use of bacteria, chemical larvi- cides, insect eating fish, copepods, and other animals. The efficiency and availability of selected controls are presented. The author inferred that wetlands designed to optimize sur- face area and plant growth without excessive mosquito pro- duction result in cost savings. 3.2.15.1. Identification of Research Needs The potential for vectors, particularly mosquitoes, to inhabit and breed in stormwater control facilities is of increasing concern to stormwater management practitioners. The evi- dent scarcity of studies and literature pertaining to the inci- dence of vectors in stormwater BMPs makes this whole cat- egory a research need. Details on the kind of research that is needed include the development and evaluation of mainte- nance and design practices that deter vectors. Poor water quality also has been linked to the mosquito proliferation in wetlands. It is thought that nutrients provide food for the bac- teria and algae on which mosquitoes feed. A better knowl- edge of the relationship between vectors and water quality is a necessary addition to the existing literature on vectors. 3.2.15.2. Primary References McClean, J. Mosquitoes in Constructed Wetlands-A Management Bugaboo? Watershed Protection Techniques, Vol. 1, No. 4 (1995) pp. 203–207.

Metzger, M. E., Messer, D. F., Beitia, C. L., Myers, C. M., and V. L. Kramer. The Dark Side of Stormwater Runoff Management: Disease Vectors Associated with Structural BMPs. Stormwater, Vol. 4, No. 7 (November–December 2003). Russell, R. C. Constructed Wetlands and Mosquitoes: Health Haz- ards and Management Options-an Australian Perspective. Eco- logical Engineering, Vol. 12 (1999a) pp. 107–124. Russell, R. C., Hunter, G., and G. Sainty. Wetlands for Stormwater Management: Water, Vegetation and Mosquitos—A Recipe for Concern. Proc., Comprehensive Stormwater and Aquatic Eco- system Management Conference, Vol. 2 (February 1999b) pp. 137–144. Sarneckis, K. Mosquitoes in Constructed Wetlands. Environment Protection Authority (December 2002) 27 pp. VBDS, California Department of Health Services. An Initial Assess- ment of Vector Production in Structural Best Management Prac- tices in Southern California. Vector-Borne Disease Section, Department of Health Services, CA (June 2001) 40 pp. Walton, W. E., Workman, P. D., and C. H. Tempelis. Dispersal, Survivorship, and Host Selection of Culex Erythrothorax (Diptera: Culicidae) Associated with a Constructed Wetland in Southern California. Journal of Medical Entemology, Vol. 36, No. 1 (Jan- uary 1999) pp. 30–40. 3.3. WATERSHED-BASED APPROACHES As previously mentioned, using watershed-based ap- proaches to stormwater planning and management in- volves coordinating and integrating human activities to implement watershed recovery efforts and to prevent fur- ther degradation of natural resources within the basin. Part- nerships and negotiations among various jurisdictions and levels of government often are required to fulfill multifac- eted social, economic, and environmental goals within the watershed. Only eight states have conducted studies or prepared reports on the retrofitting of existing stormwater manage- ment measures at DOT facilities where a watershed-based approach is employed to address fish passage or other issues pertaining to receiving waters. Five states have developed research or resources in the area of programmatic or other alternatives to project-specific mitigation, including means for establishing critical needs and priority mitigation on a watershed scale. Below, the topical areas of watershed-based approaches are divided into planning, which may involve the implemen- tation of regional and distributed stormwater management controls and practices throughout an entire basin, and into market-driven approaches, which may involve placing mon- etary value on stormwater quality that can be traded on the open market. 3.3.1. Watershed Planning In 1996, FHWA came out with Transportation Plan- ning—The Watershed Connection, which provided a national 94 focus for an envisioned relationship between transportation and watershed planning (Bank, 1996). A case study was used to illustrate how the relationship can work to maximize coor- dination and cooperation between watershed and transporta- tion stakeholders. Until TMDLs became a consideration, water-quality con- cerns were not a driving factor in DOTs’ consideration of watershed approaches. In a survey conducted in 1997, Clean Water Section 404 permitting for wetland impacts was the primary driver in state DOT efforts to consider or incorpo- rate a watershed approach (Venner, 1998). Concerns about endangered species were drivers in only a few states, includ- ing Idaho, Montana, Washington, and Maine; in those states, DOTs identified watershed boundaries on projects primarily to indicate red-flag potential impacts. The study also exam- ined success stories, barriers, and lessons learned in DOTs’ implementation of watershed approaches. Washington State DOT and North Carolina DOT remain the leaders among state transportation agencies in integrating a watershed-based approach into this work, primarily as it pertains to mitigation siting. Both Washington State DOT and North Carolina DOT target mitigation funds to sites offering the greatest ecological benefits. In North Carolina, such needs are identified through a formal watershed plan- ning process conducted by the State Department of Environ- ment and Natural Resources and are partially funded by North Carolina DOT. The watershed-based approaches of Wash- ington State DOT and North Carolina DOT follow: • Washington State DOT Endangered salmonids drive watershed planning and attention to watershed impacts in Washington State. Washington State DOT’s watershed-based approach is characterized by a community-based environmental decision-making process that coordinates and integrates human activities to implement watershed recovery efforts and to prevent further degradation of natural resources within large drainage basins. In 1996, Washington State DOT shifted from mitigating impacts on a project-by- project basis, irrespective of the top watershed needs, to analyzing mitigation opportunities based on watersheds. Now Washington State DOT’s approach makes links between watershed issues and creates partnerships with public, private, and nonprofit organizations that affect and are affected by the issues. Initiatives directly contributing to the watershed-based approach at Washington State’s DOT include the depart- ment’s Wetlands Strategic Plan, the Fish Passage Bar- rier Removal Grant Program, the Advanced Environ- mental Mitigation Revolving Account, Stormwater Retrofit Grants, Flood Management Strategy, and Cap- ital Budget Coordination. A common theme in each of these initiatives is the establishment of incentives for targeting mitigation investments to sites that protect,

preserve, or restore key components of the watershed, yielding substantial benefits for the state as a whole. Over the past year, Washington State DOT has under- taken a broad analysis of mitigation siting potential, aimed at making a tangible contribution to watershed restoration. To assist this effort, the DOT is developing landscape- based approaches and tools to systematically examine ecosystem function and identify core problems leading to degradation of water quality, increased peak flows, declining base flows, and the loss of anadromous fish habitat. These tools are pointing to more cost-effective and environmentally beneficial options when the depart- ment reaches technical limits for onsite mitigation. The approach lays the groundwork for a more flexible and less prescriptive process for achieving multiple natural resource goals, resulting in a more predictable permit- ting process with measurable transportation and envi- ronmental benefits. For each project, Washington State DOT inventories aquatic and terrestrial resources on site, identifies poten- tial impacts, and assesses the potential and sustain- ability of mitigating on site. On a watershed scale, the department determines offsite mitigation needs; charac- teristics of the predevelopment landscape; current land use and future build-out; and the condition, location, and extent of aquatic and terrestrial resources and sup- porting ecological processes. The DOT then identifies target areas for mitigation at multiple spatial scales. Within each spatial scale, Washington State DOT iden- tifies the ecological processes necessary for and capable of mitigating project impacts. To qualify as environmentally desirable offsite miti- gation, the potential mitigation site and local ecosystem processes must meet targeted threshold criteria, indicat- ing high potential to maintain ecological functions over the long term. The process identifies priority recovery areas for each targeted resource (fish and wildlife, water quality, riparian, and wetland) and opportunities and priority areas for multi-objective mitigation. Land uses that alter or decrease the success of ecological processes that the mitigation would seek to restore or enhance are a primary screen. Before candidate sites and restoration projects are chosen, a comparative assessment of eco- logical functions is performed along with social, eco- nomic, and environmental cost–benefit analyses for the candidate sites. From these assessments and analyses, Washington State DOT is able to develop a defensible priority list of sites capable of mitigating project impacts and maximizing environmental investment. Washington State DOT’s watershed-based approach is leading to the identification of mitigation sites on a watershed basis and the cost–benefit analysis of miti- gation options. A new state law has created a goal of achieving a 50% increase in environmental benefit from mitigation at a 25% reduction in cost. To direct trans- 95 portation mitigation dollars toward high-priority water- shed recovery projects in the basin, the DOT is working cooperatively with other agencies to look for ways to reduce transaction costs, increase environmental benefits, and obtain a more streamlined consensus that mitigation efforts happen in priority areas within the watershed. Washington State DOT’s Snohomish Basin Demon- stration Project has focused on developing methods to identify candidate transportation projects from the agency’s 2-year and 6-year programs, which have miti- gation needs that could be linked to watershed improve- ment activities. This project also provides an example of how a literature review was loaded into a GIS to collate environmental recommendations for the watershed. • North Carolina DOT The North Carolina DOT and the Department of Envi- ronment and Natural Resources (DENR) have designed the Ecosystem Enhancement Program (EEP) to deal with a rapidly expanding transportation program that has a high volume of new alignments, impacting an estimated 6,000 acres of wetlands and a million feet of streams over the next 7 years in a state with notable nutrient- loading concerns. The state has decided to tackle these issues through a strategic progress of riparian buffer and wetland restoration. EEP is intended to protect the state’s natural resources through the assessment, restoration, enhancement, and preservation of ecosystem functions and through identifying and implementing compensatory mitigation programmatically, at the watershed level. In particular, the program will – Enable multiple project impacts (wetlands, stream corridor, water quality, species, and habitat) to be addressed in a comprehensive manner. – Target mitigation resources to better protect the nat- ural resources of the state by assessing, restoring, enhancing, and preserving ecosystem functions and compensating for impacts at the watershed level. The program will address watershed concerns, including preservation of threatened high-quality sites and restoration of wetlands and riparian buffers along impaired streams. – Exceed the state and the Federal Highway Adminis- tration’s “no net loss” objectives for wetlands. – Allow implementation of mitigation years earlier than the current project-letting schedule, expediting proj- ects and eliminating temporal loss of wetland and riparian areas. – Reduce permit staff workload, rework, and duplica- tion of effort, thereby saving time and money. – Reduce project controversy and improve communi- cation, planning, and environmental stewardship. – Serve as a model for positive interagency relationships. – Dramatically increase the ecological effectiveness of the investments of public dollars in compensatory

mitigation, illustrating better stewardship of public resources, and setting a nationwide standard for miti- gation at the ecosystem level for unavoidable impacts resulting from transportation improvements. The EEP evolved from a multiyear effort by North Carolina DOT, DENR, FHWA, the U.S. Army Corps of Engineers, North Carolina Wildlife Resources Commis- sion, the EPA, and the U.S. Fish and Wildlife Service to streamline the project delivery process for transportation improvement projects, to reduce environmental impacts in concert with avoidance and minimization, and to pro- duce the most environmentally beneficial mitigation possible. A year of multi-agency process improvement workshops determined that compensatory mitigation should be “de-coupled” from individual permits and project reviews and performed on a watershed basis, with mitigation projects constructed in advance of per- mitted impacts. The program has been endorsed at the highest levels of participating agencies. Mitigation strategies under EEP embrace the concept of functional replacement for unavoidable impacts. Mit- igation needs and replacement opportunities are being developed through a collaborative process that includes all interested parties with the goal of restoring and protecting watersheds throughout North Carolina. The approach evaluates cumulative impacts of all projects within a watershed and implements mitigation focused on achieving a net increase in wetland and riparian func- tions in the watershed and across the state. To ensure that program goals are met, a ledger of implemented projects and actual impacts will be produced for each watershed. On an annual basis these ledgers will be compared to determine if a “no net loss” of wetland and riparian func- tions has been achieved. Any shortfall is programmed for correction in the next annual cycle, and excess miti- gation is reserved for future use. In the first year of the program, mitigation requirements were satisfied for 82 transportation projects by focusing on addressing the greatest environmental needs on a watershed basis. An interagency team led by the state’s DENR is charged with developing a watershed assessment method- ology to facilitate full replacement of functions. The team recently has compared, contrasted, and evaluated existing watershed assessment methods, including the methods utilized to develop watershed restoration plans and local watershed plans. The method will assess eco- system functions of importance and the appropriate scale and assessment methodology for each function of inter- est as determined by all the agencies involved. The team will oversee the adoption of standard protocols that will be used to establish goals and objectives for each water- shed. These protocols also will provide the framework for identifying traditional restoration and enhancement opportunities and other actions such as preservation and 96 BMPs that are consistent with the goals and objectives developed for each watershed. Anticipated deliverables will include – A watershed needs assessment methodology accepted by applicable resource management agencies; – The scale of watershed assessment for each ecosys- tem function of interest; – A guidance manual outlining the watershed needs assessment process; – Standard protocols that will be used to establish goals and objectives for each watershed; – Protocols for the selection, evaluation, and prioritiza- tion of projects, including compensatory mitigation; – Recommendations concerning the frequency of review and revision of watershed plans; – Recommendations for integrating the assessment out- comes and conclusions into a statewide GIS layer; – Criteria to measure the ecological-effectiveness and cost-effectiveness of identified projects; and – Resources (staff and funds) necessary to implement the recommended watershed assessment procedures throughout North Carolina. The compensatory mitigation strategy will include a sufficient amount of restoration and enhancement to ensure no net loss of wetland and riparian acres and functions, including water quality effects. The project’s preservation component is preserving the highest qual- ity and most biologically diverse wetland and riparian sites throughout North Carolina. 3.3.1.1. Identification of Research Needs Although much literature exists to support watershed man- agement, there is still a need for development and evaluation of techniques to integrate transportation-related runoff analy- sis into overall watershed management. Stream channels respond to changes in flow volume and sediment loading. Watershed change is known to have a corresponding effect on channels leading to bank erosion and head cutting. These processes are well understood and descriptions of channel morphology are well developed, but effective predictive models of channel geomorphic response are lacking, espe- cially in response to the episodic nature of runoff. Indices and indicators specific to transportation-related runoff are lack- ing as well. 3.3.1.2. Primary References Venner, M. Integrating Planning for Transportation and Water- shed Management. (January 1998). Venner, M. Personal Interviews: Dick Gersib, Watershed Program Manager, Washington State DOT; Bill Gilmore, EEP Manager, North Carolina DOT (2002–2003).

3.3.2. Market-Driven Approaches: BMP Asset Management and Pollutant Trading One of the major barriers to water quality planning and management in urban watersheds is the imbalance between economic development and environmental protection. Envi- ronmental economists often contend that the best approach to overcoming this barrier is to devise a mechanism for plac- ing monetary value on the quality of the environment, thus creating an economic incentive for environmental protection. Applying this reasoning to highway runoff management, BMPs may be treated as an asset that must be maintained, along with roads. Asset management is a business process and a decision-making framework that covers an extended time horizon, draws from economics as well as engineering, and considers a broad range of assets. The asset management approach incorporates the economic assessment of trade-offs among alternative investment options and uses the informa- tion to help make cost-effective investment decisions. Little asset management information exists for BMPs. Pollutant trading is a fairly new watershed-based, market- driven approach to improving receiving water quality while minimizing the costs associated with mitigation and restora- tion. Pollutant trading provides watershed managers and the regulated community more options for managing point and nonpoint source discharges. In 1996, the EPA issued Draft Framework for Watershed-Based Trading, which provides guidelines for establishing a market-based system of pollu- tant trading (U.S. EPA, 1996). Specifically, the document provides • Background on what effluent trading is and the benefits it offers; • A series of conditions that are necessary for trading, including those that ensure protection of water quality comparable to the protection that would be provided without trading; • A template of regulatory, economic, data, technical, sci- entific, institutional, administrative, accountability, and enforcement issues that facilitates identification and eval- uation of trading opportunities; and • Worksheets and checklists for evaluating whether poten- tial trades meet threshold conditions. Pollutant trading has been successfully implemented in sev- eral communities. The U.S. EPA (1999) compiled a document that summarizes 37 effluent trading and offset activities that are occurring or have occurred around the country. Half of the activities are still in the early stages of development and nearly all are trades based on point source discharges, mainly by pub- licly owned treatment works. The majority of the pollutants traded to date are nutrients (nitrogen and phosphorus). How- ever TSS, ammonia, temperature, pH, BOD, DO, and metals also have been traded at various implementation levels, including individual, watershed, and statewide trades. 97 On January 13, 2003, EPA’s Office of Water released the final Water Quality Trading Policy (U.S. EPA, 2003). The purpose of the policy is to encourage and provide guidance to states, interstate agencies, and tribes to develop and imple- ment water quality trading programs for nutrients, sediments, and other pollutants where opportunities exist to achieve water quality improvements at reduced costs. More specifi- cally, the policy is intended to encourage voluntary trading programs that facilitate implementation of TMDLs, reduce the costs of compliance with CWA regulations, establish incentives for voluntary reductions, and promote watershed- based initiatives. Based on the policy, EPA supports trading that involves nutrients or sediment loads, as well as cross- pollutant trading for oxygen-related pollutants where ade- quate information exists to establish and correlate impacts on water quality. EPA recognizes the potential value of trading other pollutants but believes such trades pose a higher level of risk and should receive a higher level of scrutiny. EPA currently does not support trading of pollutants considered (by EPA) to be persistent bioaccumulative toxics (PBTs). EPA would consider a limited number of pilot projects over the next 2 to 3 years to obtain more information on the trad- ing of PBTs. Another potential approach to stormwater management at the watershed level is using tradable allowances for excess stormwater runoff. Thurston et al. (2003) proposed a trad- able runoff allowance system that would create economic incentives for landowners to employ low-cost runoff man- agement practices so that excess stormwater flow to more ecologically sound levels could be reduced. The trading mech- anism requires detailed mapping information on individual properties, including size, percent imperviousness, and soil type to predict runoff using sophisticated hydrological mod- els. Each property owner is allocated a specific quantity of annual runoff based on an assessment of predeveloped con- ditions and receiving water sensitivity (this is assuming a stormwater management authority is in place). Anything above this allowance is considered excess runoff and must be mitigated or traded. The costs include the cost of mitigation, which would be the cost of BMP construction, operation and maintenance, and the opportunity cost of land taken out of other uses. Based on the theory of economic systems, excess runoff could be traded within a watershed-based open market. The Water Environment Research Foundation (WERF) currently is funding a 1-year, $100,000 research project— Common Currency for TMDL Commodities: Trading Infra- structure (RFP No. 02-WSM-1). The primary objective of the project is to generate practical tools that support the implementation of watershed-based trading efforts for use by point and nonpoint dischargers. The tools developed in this research will help trading participants to better delineate their watersheds and trading markets, allow participants and poten- tial participants and their commodities to identify each other, and help develop transferable marketplace infrastructure to enable creation of functional markets.

Building on lessons learned in past trading efforts (WERF- funded and others) and using EPA draft policy and other trad- ing documents as general guidelines, proposed research should advance watershed-based trading by improving the tools to accomplish trading and by providing guidance for their use. Research will develop generic infrastructure tools to assist with trading implementation. Even in places where trading has been identified as a potential solution, setting up successful trading systems is a challenging process subject to various pit- falls. Successful projects are likely to develop and improve the tools to allow trades to occur and probably will address the fol- lowing questions related to trading infrastructure: 1. Participatory Tools: How are appropriate stakehold- ers and participants identified? 2. Marketplace Tools: What is necessary for trading markets to work? How are potential traders and their commodities brought together in a functional market- place? How will transactions be made? How can trades and water quality improvements be tracked and man- aged? What tools can help meet oversight needs such as accountability and liability for trades? 3. Regulatory Tools: How might regulatory issues such as NPDES permitting, permit conditions, and discharge reporting, affect the potential market infrastructure? How could potential regulatory constraints be avoided? 4. Context: Do the tools differ? Should they differ if the trading programs are established in a pre-TMDL envi- ronment or to help meet an established TMDL? 5. Scale: What are appropriate scales for successful trad- ing programs? At what point is a system too large to handle trading programs to improve water quality? For simplicity, the research may focus on tools that work for single-pollutant programs, although ultimate extrapolation to multicredit or multimedia markets is desirable. Consideration and integration of the environmental and economic parameters necessary to accomplish trading—parameters such as admin- istration and transaction costs—are critical in this research. 3.3.2.1. Identification of Research Gaps and Needs Of the examples of market-driven trading programs reviewed in the literature, none were found that were specific to highways. However, the EPA does support trading for both point and nonpoint source load allocations, so the poten- tial exists for trading of highway runoff pollutants. As high- way runoff is typically higher in many concentrations than other urban and nonurban runoff, the potential exists to overtreat highway runoff relative to offsite other nonpoint sources (with some appropriate compensation). As the imper- viousness of highway facilities can be estimated readily, the excess stormwater runoff tradable allowances system pro- posed by Thurston et al. (2003) is a potentially feasible alter- 98 native to highway stormwater management. Since market- driven watershed-based stormwater management approaches are relatively new, there exists a need for further research into the practicality of such approaches, particularly for application to highway runoff management. Once completed, the current study funded by WERF should provide informa- tion and guidance on how a market-driven, watershed man- agement system could be applied to the highway environment. 3.3.2.2. Primary References Thurston, H. W., Goddard, H. C., Szlag, D., and B. Lemberg. Con- trolling Stormwater Runoff with Tradeable Allowances for Impervious Surfaces. Journal of Water Resources Planning and Management, Vol. 129, No. 5 (2003) pp. 409–418. U.S. Environmental Protection Agency. Water Quality Trading Policy. Office of Water (January 13, 2003). U.S. Environmental Protection Agency. Environomics: A Summary of U.S. Effluent Trading and Offset Projects. Office of Water (1999). U.S. Environmental Protection Agency. Draft Framework for Water- shed Based Trading. Report No. EPA 800-R-96-001, Office of Water (1996). 3.4. HIGHWAY RUNOFF CHARACTERIZATION AND ASSESSMENT This category refers to the hydrologic and water quality characterization of highway runoff before entering water quality control facilities or receiving streams. The knowl- edge gained from characterization monitoring helps DOT planners and managers understand how to establish storm- water management priorities. The research category of highway runoff characterization and assessment has been well studied and several general characterization studies are readily available in the literature. However, this category is broad and the level of detail for any one study may range from the microscale, such as unit chem- ical processes, to the macroscale, such as gross pollutant transport. Furthermore, because of the variability and com- plexity of environmental systems, the association or correla- tion between different physical, biological, and chemical parameters will require several more decades, if not longer, to fully understand at a satisfactory level. Due to this fact, this assessment does not attempt to identify all research gaps, but rather identifies the gaps where there is a current need for a better understanding for the purposes of improved highway runoff management. The following paragraphs are organized according to subcategories based on perceived needs of the transportation and water resources community. 3.4.1. General Constituent Characterization During the last three decades, researchers throughout the developed world have conducted highway runoff quality

characterization studies. The primary driver for runoff qual- ity characterization today is regulatory compliance; how- ever, characterization also is conducted to identify pollutants of concern for the development or refinement of stormwater quality management programs. In the United States, federal and state agencies have conducted several studies to identify the type and range of concentrations of water quality con- stituents typically found in highway runoff. During the 1970s and 1980s, FHWA funded several stud- ies pertaining to highway runoff quality characterization. The following excerpt from Bank et al. (1995) summarizes the multivolume research reports concerning highway runoff quality developed by FHWA: 1. Constituents of Highway Runoff—This six-volume report, completed in 1981, developed a predictive pro- cedure for determining the pollutant characteristics of stormwater draining from roadway surfaces. The pro- cedure is composed of equations that predict the run- off volumes and pollutant wash-off coefficients of 17 water quality parameters for three types of highways. 2. Sources and Migration of Highway Runoff Pollutants— This four-volume report, completed in 1984, identifies the sources of potential water pollutants, their deposi- tion and accumulation within the highway facility, and their subsequent removal to the surrounding environ- ment. The purpose of this research was to develop methods for controlling pollutant sources and mitiga- tion measures to lessen pollutant levels entering receiv- ing waters. 3. Effect of Highway Runoff on Receiving Waters—Com- pleted in 1985, this five-volume report analyzes the effects of highway stormwater runoff on receiving waters. Included in the effort were 1-year field studies at three sites and preparation of three user-oriented manu- als that provide guidelines for collecting information to use in highway project environmental assessments. 4. Highway Maintenance Impacts to Water Quality—This four-volume series of reports, completed in 1985, sum- marizes a research project involving impacts from high- way maintenance practices on water quality. Research efforts included (1) evaluating the impact potential of routine practices; (2) developing assessment methods for specific practices; (3) identifying measures to miti- gate impacts; and (4) conducting field studies to better define impacts from two common practices, herbicide application and surface treatment (seal-coating). 5. Retention, Detention, and Overland Flow for Pollu- tant Removal from Highway Stormwater Runoff— This report, completed in 1987, provides interim guide- lines for the removal of pollutants from highway stormwater runoff. Three general types of management measures have been determined, through previous FHWA studies, to be effective in treating highway runoff: vegetative controls (overland flow and grassed channels), detention basins (wet detention basins and 99 wetlands), and retention measures (retention basins, trenches, and wells). These interim design guidelines were developed through the project team’s experience and through a thorough review of available literature. 6. Pollutant Loadings and Impacts from Highway Storm- water Runoff—Published in 1990, this is a culmination of analytical effort using the results of previous water quality studies in concert with hydraulic, environmen- tal, and related concerns. The results of this study include a probabilistic design procedure for estimating impacts to waters receiving highway stormwater runoff. The procedure used and expanded on the predictive model developed in the first series of reports. Addi- tional runoff water quality data collected by this and other studies subsequent to the original work in the first phase were used to refine the regression analyses sup- porting the predictive procedure. The 1990 effort was undertaken to improve the highway practitioner’s ability to address highway stormwater runoff issues. Formulated by Driscoll and others, the model used factors such as storm event statistics and probability distri- bution of site event mean concentration to estimate runoff vol- umes, concentrations, and loads, and probability distribution of streamflow to estimate potential dilution in receiving waters. This statistical model, useful for planners and high- way practitioners, uses readily available rainfall statistics and water quality data to produce a frequency distribution of con- centration, loads, and effects for receiving waters. The existing FHWA model was formulated using data from 993 separate highway runoff events at 31 sites in 11 states dur- ing the period from 1975 to 1985. In the past 20 years, auto- mobile construction materials, technology, and fuel additives have changed, and these factors have affected the loads from highway surfaces. Research has indicated that lead may be substantially lower than in the 1970s and 1980s because of improvements in fuel formulations, emissions controls, and tire wear (even though total vehicle miles traveled has increased). FHWA has a study underway with USGS to incorporate the existing model in a new user-friendly software platform; update the existing model with new streamflow and rainfall data; and expand the model to include the availability of dis- solved concentrations, sediment size information, sediment- chemistry information, and a data quality advisory system. The final version will be an updated version of the existing 1990 model, along with a new version of the model that is designed to incorporate new highway runoff data as it becomes available. This information will benefit the municipalities and state and federal agencies charged with estimating high- way runoff pollutant loads. In addition to the FHWA-funded research, several other highway runoff characterization studies have been conducted by state DOTs and independent researchers. Stormwater sam- ples collected by the Michigan DOT at three major highways indicated that concentrations of conventional constituents—

such as biochemical oxygen demand (BOD), TSS, and phos- phorus—are comparable to the concentrations collected in the FHWA studies of the 1970s and 1980s (CH2MHill, 1998). However, concentrations of metals, lead in particular, were lower for the Michigan DOT sampling than they were for the FHWA studies. This finding can be attributed to the discon- tinuation of leaded gasoline and improvements in sampling and analytical techniques. The earlier FHWA data contains only limited information on the dissolved form of metals, a critical consideration regarding effects of metals on aquatic biota. Therefore, this study fills a significant gap in previous FHWA highway runoff characterization studies. Organic compounds were, for the most part, not detected in Michigan DOT runoff samples. In a study funded by the Texas DOT, water quality of high- way runoff in the Austin area was determined by monitoring runoff at three locations on the MoPac expressway, which represented different daily traffic volumes, surrounding land uses, and highway drainage system types (Barrett et al., 1996). The highest concentrations of all constituents were measured at the high-traffic site [average daily traffic (ADT) >30,000 vehicles]. The concentrations at all sites were simi- lar to median values for similar sites compiled in FHWA’s nationwide studies of highway runoff quality. Caltrans has ongoing runoff characterization studies at several different types of transportation facilities throughout the state, including highways (congested and free-flowing), construction sites, maintenance yards, and park-and-ride sta- tions. Caltrans’ Preliminary Report of Discharge Characteri- zation Studies summarizes runoff quality data collected since the 2000–2001 wet season from more than 50 sites (Caltrans, 2003). With the exception of total dissolved solids, TSS, dis- solved lead, total lead, and dissolved arsenic, all of the mean concentrations from the congested highways were greater than the mean concentrations from free-flowing highways; however, none of the differences were statistically significant. For the most part, the data agree with the ranges reported in FHWA’s studies as well as with studies by Texas DOT and Michigan DOT. As with Michigan DOT’s study, lead con- centrations were generally less in the Caltrans data as com- pared to the FHWA data. 3.4.1.1. Constituents Specific to the Highway Environment Volatile Organic Carbons. Many studies to characterize con- centrations of semivolatile organic compounds (SVOCs) in highway runoff and urban stormwater have been conducted since 1970 (Lopez and Dionne, 1998). To a lesser extent, studies also have characterized concentrations of volatile organic compounds (VOCs), estimated loads of SVOCs, and assessed potential impacts of these contaminants on receiv- ing streams. This review evaluates the quality of existing data on SVOCs and VOCs in highway runoff and urban stormwater 100 and summarizes significant findings. Studies related to high- ways are emphasized when possible. The review included 44 articles and reports that focused primarily on SVOCs and VOCs. Only 17 of these publications are related to highways, and 5 of these 17 are review papers. SVOCs in urban storm- water and sediments from the late 1970s to the mid-1980s were the subject of most studies. Criteria used to evaluate data quality included documenta- tion of sampling protocols, analytical methods, minimum reporting limit or method detection limit, quality-assurance protocols, and quality-control samples. The largest deficiency in documenting data quality was that only 10% of the studies described where the water samples had been collected in the stream cross section. About 80% of SVOCs in runoff are in the suspended solids. Because suspended solids can vary sig- nificantly even in narrow channels, concentrations from dis- crete point samples and contaminant loads estimated from those samples are questionable without information on sam- ple location or how well samplers control the quality of sam- ples. Comparison of results of different studies and evaluation of the quality of environmental data, especially for samples with low concentrations, is difficult without this information. The most significant factor affecting SVOC concentra- tions in water is suspended solids concentration. In sediment, the most significant factors affecting SVOC concentrations are organic carbon content and distance from sources such as highways and power plants. Petroleum hydrocarbons, oil and grease, and polycyclic aromatic hydrocarbons (PAHs) in crankcase oil and vehicle emissions are the major SVOCs detected in highway runoff and urban stormwater. The few loading factors and regression equations that were developed in the 1970s and 1980s have limited use in esti- mating current loads of SVOCs on a national scale. These factors and equations are based on few data and use incon- sistent units, and some are independent of rainfall. Also, more cars on the road today have catalytic converters, and fuels that were used in 2003 are cleaner than when loading factors and regression equations were developed. Comparisons to water-quality and sediment-quality criteria and guidelines indicate that PAHs, phenolic compounds, and phthalates in runoff and sediment exceeded EPA drinking- water and aquatic-life standards and guidelines. PAHs in stream sediments adjacent to highways have the highest potential for adverse effects on receiving streams. Few data exist on VOCs in highway runoff. VOCs were detected in precipitation adjacent to a highway in England, and chloromethane, toluene, xylenes, 1 2 4-trimethylbenzene, and 1 2 3-trichloropropane were detected in runoff from a highway in Texas. In urban stormwater, gasoline-related compounds were detected in as many as 23% of the samples. Land use could be the most significant factor affecting the occurrence of VOCs, with the highest concentrations of VOCs found in industrial areas; temperature is another factor. Urban land surfaces are the primary nonpoint source of VOCs in stormwater. However, the atmosphere is a potential source of

hydrophilic VOCs in stormwater, especially during cold sea- sons when partitioning of VOCs from air into water is great- est. Tetrachloroethene, dichloromethane, and benzene were the only VOCs detected in stormwater that exceeded EPA drinking-water standards. Polyaromatic Hydrocarbons. PAHs are a group of toxic and carcinogenic compounds rarely included in characterization studies, but often present in highway runoff due to traffic- related sources. PAHs represent more than 2000 PAH com- pounds; only 16 have been placed on the EPA list of priority pollutants (Pawluk et al., 2002). PAHs are ubiquitous and are emitted from practically every combustion source. Follow- ing combustion, PAHs enter the atmosphere, rivers, and lakes through wet deposition or through dry deposition, where they are washed away by stormwater runoff. Specific traffic-related sources of PAHs include tire wear, asphalt and asphalt coat- ings, vehicle exhausts, and lubricating oils and grease (Pawluk et al., 2002). Other sources include industrial effluents and spills or intentional dumping. Some PAHs will evaporate from water and soil, but the majority of PAHs in stormwater usually are found in partic- ulate form. A stormwater runoff study done by Marslek et al. (1997) found that the dissolved phase PAHs represented less than 11% of the total concentrations. In another study, Shinya et al. (2000) found that the higher molecular weight PAHs were more associated with suspended solids in the runoff and the predominant PAHs (phenanthrene, fluoran- thene and pyrene) comprised about 50% of 15 quantified PAH constituents in each sample. In results from Ames’ assay, mutagenicity was associated appreciably with PAHs in the particulate fraction of runoff water. The dissolved frac- tion also showed positive mutagenic response by unknown soluble aromatic compounds. Smith et al. (2000) collected 42 stormwater runoff samples from four sampling sites (a high- way off-ramp, a gas station, and a low- and high-traffic- volume parking lot). For each sample, the suspended-sediment and water phases were separated and analyzed for 16 PAHs. The gas station site produced the highest total PAH loading (2.24 g/yr/m2), followed by the high-traffic-volume parking lot (5.56 × 10-2 g/yr/m2), the highway off-ramp (5.20 × 10−2 g/yr/m2), and the low-traffic-volume parking lot (3.23 × 10−2 g/yr/m2). In several samples, one or more PAHs were detected in the aqueous phase at concentrations above aqueous solu- bility. This result suggests the presence of colloidal-size par- ticles capable of sorbing PAHs to an appreciable extent, or the presence of an oil-and-grease microemulsion. The effects of PAHs on aquatic systems are not well known. Platinum Group Metals. Another group of elements that have been found more recently in urban runoff are the platinum group metals (PGMs): palladium, platinum, and rhodium. PGMs are used in catalytic converters to abate the emission of aromatic hydrocarbons, CO and NOx. Due to the thermal and mechanical conditions under which autocatalysts work 101 (including abrasion effects and hot-temperature chemical reac- tions with oil fumes), significant release of the PGMs to the environment in the form of fine particles can occur (Carolia et al., 2000). This raises concern, because platinum is a known cytotoxin and tends to bioaccumulate. Because air quality regulations require catalytic converters in all new cars, the amount of PGMs released into the environment each year is expected to continue to increase. A study conducted by Rauch et al. (1999) investigated the concentrations of PGMs in road sediment samples in 1984, 1991, and 1999 and found a clearly increasing trend, especially as related to particles smaller than 63 µm. Gasoline Oxygenates. Methyl tert-butyl ether (MTBE), a gasoline oxygenate, disperses rapidly in water and is less bio- degradable than common gasoline compounds, such as ben- zene, toluene, ethylbenzene, and total xylene (BTEX). USGS sampled stormwater in 16 cities and metropolitan areas that are required to obtain permits to discharge stormwater from their municipal storm-sewer system into surface water (Delzer et al., 1996). Concentrations of 62 VOCs, including MTBE and BTEX compounds, were measured in 592 stormwater samples collected in these cities and metropolitan areas from 1991 through 1995. MTBE was the seventh most frequently detected VOC in urban stormwater, following toluene, total xylene, chloro- form, total trimethylbenzene, tetrachloroethene, and naph- thalene. MTBE was detected in 6.9% (41 of 592) of storm- water samples collected. When detected, concentrations of MTBE ranged from 0.2 to 8.7 µg/L, with a median of 1.5 µg/L. All detections of MTBE were less than the lower limit of the EPA draft lifetime health advisory (20 µg/L) for drinking water. Eighty-three percent of all detections of MTBE in storm- water were in samples collected from October through March 1991–1995, a timeline that corresponds with the expected seasonal use of oxygenated gasoline in areas where carbon monoxide exceeds established air-quality standards. The median concentration of MTBE and benzene for all samples was statistically different and higher in samples collected during the October–March season than in samples collected during the April–September season. Sixty-six percent of all MTBE detections occurred with BTEX compounds, and a pro- portionate increase in concentrations was found when these compounds occurred together. The proportionate increase could indicate a common source of MTBE and BTEX for those samples. Toluene and total xylene were the most fre- quently detected BTEX compounds and the most frequently detected VOCs in these investigations. Detected concentra- tions of toluene and total xylene ranged from 0.2 to 6.6 µg/L and from 0.2 to 15 µg/L with median concentrations of 0.3 and 0.4 µg/L, respectively. These data raise questions that remain to be answered because these stormwater investigations were not designed specifically to characterize the occurrence, sources, and behav-

ior of oxygenated gasoline components in stormwater. Ques- tions include • What are the ranges and seasonal distributions of con- centrations of MTBE in stormwater, including munici- pal separate-storm-sewer systems and combined sewer overflows, in other urban areas of the United States? • What is the persistence of MTBE in streams or rivers that receive stormwater runoff? Are the concentrations in the receiving stream a cause for concern because of potential effects on aquatic life? Similarly, what effect, if any, does MTBE have on public water supplies from surface-water sources? • What proportion of MTBE detected in urban storm- water is contributed by precipitation and what proportion is contributed by overland runoff? How much MTBE is contributed to surface water by precipitation that falls directly on larger bodies of water such as reservoirs and lakes? • Do other oxygenates react similarly to MTBE in the hydrologic cycle and occur in stormwater? • Is land use an important factor in the occurrence of MTBE or BTEX compounds in urban stormwater? • Is stormwater recharge or precipitation that contains VOCs an important source of MTBE to groundwater in urban environments? 3.4.1.2. Identification of Research Needs Many state DOTs have studied highway runoff, so there are several studies available that generally characterize highway runoff quality. The constituents sampled and the concentra- tions detected do not appear to vary significantly between the studies; therefore, the research topic of general highway runoff characterization does not represent a primary research need. However, there is clearly a need for better highway runoff characterization (including monitoring techniques) of petro- leum hydrocarbons, including PAHs, as well as other trace ele- ments not normally included in characterization studies, such as BTEX, MTBE, and PGMs. More advanced highway runoff characterization studies, such as those that investigate first- flush phenomena, the correlation of water quality parameters to independent variables, or the fate and transport of highway runoff pollutants, will be discussed in following subsections. 3.4.1.3. Primary References Bank, F., Kerri, K. D., Young, G. K., and S. Stein. National Evalu- ation of Water Quality Issues for Highway Planning. In Trans- portation Research Record 1483, TRB, National Research Coun- cil, Washington, DC (1995) pp. 89–91. Barrett, M. E., Malina, J. F. Jr., and R. J. Charbeneau. Characteri- zation of Highway Runoff in the Austin, Texas, Area. Report No. 102 CRWR-263, Center for Research in Water Resources, Bureau of Engineering Research, University of Texas, Austin (1996). Caltrans. Preliminary Report of Discharge Characterization. Report No. CTSW-RT-03-023, (2003). Carolia, S., Alimontia, A., Petruccia, F., Boccaa, B., Krachler, M., Forastierec, F., Sacerdotec, M.T., and S. Mallonec. Assessment of Exposure to Platinum-Group Metals in Urban Children. Spec- trochimica Acta, Part B, Vol. 56 (2000) pp.1241–1248. CH2M Hill. Highway Stormwater Runoff Study. Report, State of Michigan, Department of Transportation (1998). Lopes, T. J., and S. G. Dionne. A Review of Semi-Volatile and Volatile Organic Compounds in Highway Runoff and Urban Stormwater. Open File Report 98-409, U.S. Geological Survey (1998) 67 pp. Marsalek, J., Brownlee, B., Mayer, T., Lawal, S., and G. A. Larkin. Heavy Metals and PAHs in Stormwater Runoff from the Skyway Bridge, Burlington, Ontario. Water Quality Research Journal of Canada, Vol. 32, No. 4 (1997) pp. 815–827. Pawluk, D. W., Yonge, D., and M. Barber. Polycyclic Aromatic Hydrocarbons and Stormwater Management: Past, Present, and Future. Washington State University, Department of Civil and Environmental Engineering, Seattle (2002). Rauch, S., Morrison, G. M., Motelica-Heino, M., and O. Donard. Evaluation of Speciation, Transport and Ecological Risk of Pal- ladium, Platinum and Rhodium in Urban Stormwater Systems. Proc., 8th International Conference on Urban Storm Drainage (1999) pp. 202–209. Shinya, M., Tsuchinaga, T., Kitano, M., Yamada, Y., and M. Ishikawa. Characterization of Heavy Metals and Polycyclic Aromatic Hydrocarbons in Urban Highway Runoff. Water Science and Technology, Vol. 42, No.7-8 (2000) pp. 201–208. Smith, J. A., Sievers, M., Huang, S., and S. L. Yu. Occurrence and Phase Distribution of Polycyclic Aromatic Hydrocarbons in Urban Stormwater Runoff. Water Science and Technology, Vol. 42, No. 3 (2000) pp. 383–388. 3.4.2. Runoff Characterization with Independent Variable Correlation This broad category refers to water quality parameters whose presence or magnitudes, or both, are associated with other parameters or external factors. Correlation often is used to account for some of the variability observed in hydrologic and water quality data and to build regression equations for predicting difficult-to-measure parameters. Some of the common research questions posed with regard to runoff water quality correlation include • How do suspended solids or particle-size distribution relate to constituent concentrations? • What are the effects of ADT or vehicles during a storm (VDS) on stormwater quality? • How do hydrological factors such as the antecedent dry period (ADP), rainfall volume, intensity, or duration affect runoff quality? • Is there a discernable difference in runoff water quality from different land uses?

3.4.2.1. Suspended Solids and Particle-Size Distribution Several researchers have attempted to correlate various water quality parameters to suspended solids (the residue retained on a 1.2 µm filter, also referred to as filterable resi- due). Kerri et al. (1985) conducted an intensive runoff mon- itoring effort at urban highway sites in Redondo Beach, Wal- nut Creek, and Sacramento, California, as well as from a rural site near Placerville, California, in an effort to develop regression equations for estimating pollutant loads from high- ways. Rainfall and runoff were monitored continuously. Bub- bler flow meters were used with automatic sequential samplers so that stormwater samples could be collected to characterize entire storm events. The constituents that were analyzed were boron, total lead, total zinc, nitrate (nitrogen), ammonia (nitrogen), total Kjeldahl nitrogen (TKN), total phosphorus, dissolved orthophosphate, oil and grease, nonfilterable resi- due, filterable residue, total cadmium, and chemical oxygen demand (COD). Based on regression analysis, the total resi- due (TSS) was evaluated and accepted as a satisfactory inde- pendent variable for estimating total zinc, nonfilterable resi- due, and COD in runoff from highways with average daily traffic (ADT) of at least 30,000 vehicles. In another study, Zhao et al. (1999) correlated suspended solids and COD for stormwater runoff from an urban highway in Xi’an, China, indicating that COD is strongly associated with particulate matter. Sansalone and Buchberger (1997) studied the association of metal elements as a function of particle size for both rainfall runoff and snowmelt. Solids ranged from smaller than 1 µm to greater than 10,000 µm. Flow rate and duration controlled the yield and size of transported solids. Metal element analy- sis of particle size distribution (PSD) from snow and rainfall indicate that zinc, copper, and lead mass increase with decreas- ing particle size [i.e., increasing specific surface area (SSA)]. No clear trends, as a function of increasing SSA or between snow and rainfall runoff solids, are apparent for cadmium, which is a very mobile metal and mainly is dissolved in high- way runoff. Based on this study, it is apparent that zinc, lead, and copper concentrations on solids may vary significantly from one event to another with the tendency for higher con- centrations to be associated with the smaller particle sizes. Cadmium concentrations, however, do not tend to vary with storm events or particle sizes. Sediment particle size characterization and its relationship with nutrient content (especially phosphorus) is an important element in the treatability evaluation of stormwater runoff. Studies from the Lake Tahoe Basin suggest that movement of total phosphorus in the tributaries to Lake Tahoe correlates with the sediment transport, supporting the contention that erosion and nutrient loading are related. However, there appear to be conflicting observations suggesting that all sedi- ment is not the same with regard to the phosphorus content. A study by Reuter and Miller (2000) indicates that fine-grained 103 sediment (<63 µm, thus finer than sand) correlates better with nutrients than does coarse-grained sediment (≥63 µm), per- haps implying that nutrient transport is more sensitive to the movement of fine-grained materials. Another study reported by Hydro Science (1999) indicates that 46% of the particulate phosphorus is associated with par- ticles that are the size of sand or larger. Phosphorus, with similar tendencies as heavy metals, also may be bound dis- proportionately to larger particles (Glen and Sansalone, 2001). Based on this literature review, it is apparent that a broad and clear relationship between particle size and total phosphorus is lacking. Thus, there is a need to characterize suspended sediment loads into fine (clay and silt) versus coarse and to characterize the nutrients associated with them. 3.4.2.2. Average Daily Traffic, Vehicles during a Storm, and Antecedent Traffic Count Several researchers have attempted to correlate ADT to loads and concentrations in urban runoff with variable suc- cess. Driscoll et al. (1990) found that there was no definitive relationship between traffic density and pollutant levels. A strong positive correlation for zinc (r2 = 0.7) and a weak pos- itive correlation for VSS, DOC, and TOC were observed. However, based on the fact that the other metals often asso- ciated with highway runoff (e.g., copper and lead) did not appear to be correlated with ADT, the authors concluded that ADT should not be used as a surrogate measure of pollutant levels, and surrounding land use appeared to be more corre- lated to pollutant levels than to ADT. In a study conducted by Washington State DOT and sum- marized by Thomson et al. (1997), the researchers demon- strated that storm event loads of copper, lead, zinc, and TKN could be correlated with ADT and TSS loads. Correlation coefficients were all above 0.8, indicating that greater than 80% of the variability of these constituent loads could be explained by the variation in ADT and TSS loads. However, single variate correlation of ADT alone was not conducted, so it is not possible to statistically assess the amount of vari- ability in constituent loads associated with ADT. Table 3-7 is matrix of independent variables that affect var- ious constituent concentrations in highway runoff (Thomson et al., 1997). Kerri et al. (1985) found that the number of VDS was eval- uated and accepted as a satisfactory independent variable at the 5% confidence level for estimating the loads of total lead, total zinc, filterable residue (TSS), chemical oxygen demand, and TKN. The authors stress the use of these equations should be limited to highways with ADT of at least 30,000 vehicles. The numbers of antecedent dry days was found not to be a satis- factory independent variable for constituent correlation. One of the most profound highway runoff correlation stud- ies found in the literature review was conducted by Irish et al. (1995) in Austin, Texas. During this study, 35 storm events

were simulated using a full-scale rainfall simulator, and 23 natural storm events were sampled at the same location; both events occurred with active traffic. Twenty-one variables were identified for each storm event, and multiple regression analysis was used to determine the relationship of each vari- able to the quality of the highway runoff. The highway runoff constituents significantly affected by VDS were BOD, lead, copper, and oil and grease. The highway runoff constituents significantly affected by antecedent dry period traffic count (ATC) were COD, BOD, phosphorus, nitrate, and zinc. 3.4.2.3. Hydrological Factors Some of the common hydrological factors believed to affect the quality of highway runoff are ADP and the runoff volume, intensity, and duration during a storm. As with ADT, researchers have had variable success with the correlation between ADP and stormwater quality. Both Kerri et al. (1985) and Reinertsen (1981) found that the number of antecedent dry days was not a satisfactory independent variable for con- stituent correlation. Drapper et al. (2000) and Thomson et al. (1997) found that interevent duration can be a statistically significant factor for pollutant concentrations. Thomson et al. (1997) showed that iron, TSS, and COD could be positively correlated to ADP. However, other independent variables such as storm volume and storm intensity were shown to account for a greater amount of the variability in concentra- tions of these constituents. With regard to runoff volume, Reinertsen (1981) found that the discharge level alone did not influence the runoff quality as much as might be expected, and in fact no significant cor- relation was observed between the concentrations and the dis- charge either within or between rain events. Driscoll et al. (1990), in their analysis of 184 paired data sets representing 104 eight different pollutants at each of 23 highway sites, con- cluded that pollutant event mean concentrations (EMCs) are independent and unrelated to either rainfall or runoff volume. However, in another study, Colwill (1985) found that the amounts of particulate material and associated contaminant removed from the road surface by individual events are dic- tated largely by the intensity of rainfall and the total volume discharged. Stenstrom et al. (1982) found a strong correlation between total rainfall and total mass of oil–grease pollution. Therefore, it appears that there is not a significant correlation between runoff volume and concentrations, but there is a sig- nificant correlation between runoff volume and loads, as expected. Very few studies were found that investigated the correlation between storm intensity with concentrations or loads. However, it is expected that particulate-bound con- stituents are influenced greatly by rainfall intensity. 3.4.2.3 Land Use Effects FHWA and USGS are cooperating on research to deter- mine the various components of impervious surfaces to the overall stormwater runoff issue using existing land use, land cover, and impervious surface data. There are numerous studies on impervious surfaces, but some have differentiated between rooftops and transportation systems and some have identified buildings and roads as the only contributor for all the impervious surfaces. If the components of impervious surfaces are broken down into more detailed components with a watershed, methodologies can be developed and eval- uated on how to control and mitigate these impacts. In order to improve understanding of how much each impervious sur- face is contributing to the total imperviousness for each water- shed, this research specifically examines the percentage of transportation infrastructure as well as the percentage of con- TABLE 3-7 Identification of independent variables affecting constituent concentrations in highway runoff during multiple regression analysis (adapted from Thomson et al., 1997)

tributions that state transportation agencies maintained high- way systems contribute to the total imperviousness of an urban watershed. The correlation of land use to pollutant loads has been investigated by several other researchers over the last few decades. Probably the largest land use-based runoff charac- terization study to date is the EPA Nationwide Urban Runoff Program (U.S. EPA, 1983), in which runoff samples were collected from 28 major metropolitan areas across the United States and analyzed over a 5-year period. One of the most significant findings of this research was that runoff concen- trations from the various land uses (residential, mixed, com- mercial, industrial, and open/nonurban) were not statistically significantly different from one another, with the exception of total phosphorus from open/non-urban land use areas. Regardless of these findings, the characterization of runoff according to land use continues to be a topic of interest for many researchers because of the implications for predicting impacts of development. Since 1994, the County of Los Angeles has been collecting stormwater samples from various land uses throughout the county as part of their Phase I NPDES Permit requirements (Los Angeles County Department of Public Works, 2001). Land uses monitored include retail and commercial, vacant, high-density single-family residential, transportation, light industrial, education, multifamily residential, and mixed res- idential. Results suggest that there are some distinct differ- ences in the average runoff concentrations of monitored pol- lutants between land uses; however, the study did not evaluate the statistical significance of those differences. Metals, nutri- ents, and oil–grease concentrations were highest from trans- portation and light industrial and commercial sites, although the open space sites had the highest TSS concentrations. The Oregon Clean Water Agencies published a compila- tion of land use data collected in Oregon (Willamette Valley) for the Phase I Part II Municipal Stormwater Permit Applica- tions that showed that different urban land uses could be char- acterized as having different water quality for a number of constituents, including heavy metals (Strecker, 1995). Phos- phorus appeared to be more related to surrounding soil types. In a study conducted by Stenstrom et al. (1982), five field sampling stations were selected in a stormwater basin in Richmond, California, to determine oil–grease pollution by land-use type. Samples were taken from the mouth of the watershed, a parking lot, a commercial street, a residential area, and a light industrial facility. Results of the investiga- tion indicated that land use strongly affects oil–grease in stormwater with the major contributing factor being motor vehicles. Areas with the most auto traffic had the highest concentration of oil–grease in stormwater and the highest hydrocarbon load factor. Mean oil–grease concentration in runoff flow ranged from 4.13 mg/l in an upstream residential area to 15.25 mg/l in a parking lot. Highways usually are considered an individual land use type but are sometimes further divided according to urban 105 and rural, congested and free-flowing, total area, percent impervious area, the number of lanes, on- and off-ramps, and ADT, which has already been discussed. Only a few studies could be found that discuss differences among some of these more detailed highway land use classification levels. Based on their analysis, Driscoll et al. (1990) recom- mended using urban versus nonurban for classifying high- way sites rather than ADT. Caltrans (2003) classified high- way sites according to congested and free-flowing, but they have yet to statistically analyze their data. However, after briefly perusing the means and standard deviations of the monitored constituents, it appears that there will not be sta- tistically significant differences between the two classifica- tion types. Thomson et al. (1997) analyzed runoff quality data from several different highway classification types including the total area, total impervious area, and total num- ber of lanes. The analysis revealed that with the addition of other regression parameters such as TSS, TDS, and TOC, percent impervious area may be useful for predicting COD, and the total number of lanes may be useful for predicting chloride, sodium, and COD on a site-specific basis. Drapper et al. (2000) found that sites incorporating exit lanes have recorded higher concentrations of acid-extractable copper and zinc, supporting the hypothesis that brake pad and tire wear caused by rapid deceleration contributes to the concen- trations of these metals in road runoff. However, these data were discussed qualitatively only. 3.4.2.4. Identification of Research Needs With regard to suspended sediment and particle size dis- tribution, the association of typically observed highway runoff pollutants with suspended sediment seems to be well charac- terized. However, there still appears to be a need for better characterization of constituents associated with different sized particles in highway runoff, particularly heavy metals, nutrients, and hydrocarbons. The association of traffic volume with runoff concentra- tions is well documented in the literature reviewed but still lacks a statistically valid amount of data to support signifi- cant conclusions. ADT does not appear to be a consistently good predictor of pollutant concentrations and loads. How- ever, VDS may hold promise for estimating concentrations for some metals and nutrients, as well as TSS, COD, BOD, and oil and grease. Thus, there appears to be a need for more research in the area of runoff characterization with correla- tion to VDS. Only one study was found that investigated the effects of ATC, so this may also represent a research gap. The hydrological factors such as runoff volume, rainfall volume, intensity, and duration are independent variables that have been shown by a few researchers to affect runoff con- stituent levels. Total storm volume affects loads of some water quality parameters such as TSS and oil and grease but does not appear to significantly affect concentrations. Corre-

lations between intensity and duration with constituent levels are sparse in the literature reviewed, indicating this may be a research gap. Land use appears to affect average stormwater runoff con- centrations, yet no studies have been found that show statis- tically significant differences in concentrations based on land use type alone. Land use classification and separation prob- ably are major factors influencing the variability in land use- based runoff concentrations. Land uses often are mixed (which frequently is considered a net benefit to stormwater quality) making it difficult to classify and to separate stormwater flows. Differences in classification levels also make it difficult to compare studies. Runoff quality characterization according to the various highway classifications, especially urban ver- sus rural, on-ramps and off-ramps, and percent impervious area, appears to be an area needing further research. Staff at the Center for Watershed Protection, in collabora- tion with Dr. Robert Pitt, are compiling and summarizing the available national data on urban runoff water quality and are conducting data explorations to ascertain potential explaining factors. 3.4.2.5. Primary References Colwill, D. M. Water Quality of Motorway Runoff. Supplemen- tary Report 823, Transport and Road Research Laboratory (1985) 26 pp. Drapper, D., Tomlinson, R., and P. Williams. Pollutant Concen- trations in Road Runoff: Southeast Queensland Case Study. Journal of Environmental Engineering, Vol. 126, No. 4 (2000) pp. 313–320. Driscoll, E. D., Shelly, P. E., and E. W. Strecker. Pollutant Load- ings and Impacts from Highway Stormwater Runoff—Volume III. Analytical Investigation and Research Report FHWA-RD-88- 008, FHWA (1990) 150 pp. Glenn III, G. W., and J. J. Sansalone. Accretion and Partitioning of Heavy Metals Associated with Snow Exposed to Urban Traffic and Winter Storm Maintenance Activities—Part II. Journal of Environmental Engineering, Vol. 128, No. 2 (2001) pp. 167–185. Hydro Science. Bioavailable Nutrient Loading into Lake Tahoe and Control Opportunities with an Emphasis on Utilizing SEZs to Treat Urban Runoff. Tahoe Regional Planning Agency (1999). Irish, L. B., Lesso, W. G., Barrett, M. E., Malina, J. F., and R. J. Charbeneau. An Evaluation of the Factors Affecting the Quality of Highway Runoff in the Austin, Texas, Area. Report No. CRWR 264, Bureau of Engineering Research, Center for Research in Water Resources, University of Texas, Austin (1995). Kerri, K. D., Racin, J. A., and R. B. Howell. Forecasting Pollutant Loads from Highway Runoff. In Transportation Research Record 1017, TRB, National Research Council, Washington, DC (1985) pp. 39–46. Los Angeles County Department of Public Works. Los Angeles County 1994–2000 Integrated Receiving Water Impacts Report. Watershed Management Division (2001). Reinertsen, T. R. Quality of Stormwater Runoff from Streets. Proc., 2nd International Conference on Urban Storm Drainage (1981) pp. 107–115. 106 Reuter, J. E., and W. W. Miller. Aquatic Resources, Water Quality, and Limnology of Lake Tahoe and its Upland Watershed. Chap- ter 4 of The Lake Tahoe Watershed Assessment, Pacific South- west Research Station, USDA Forest Service (2000). Sansalone, J. J., and S. G. Buchberger. Characterization of Solid and Metal Element Distributions in Urban Highway Stormwater. Water Science and Technology, Vol. 36, No. 7-8 (1997) pp. 155–160. Stenstrom, M. N., Silverman, G., and T. A. Bursztynsky. Oil and Grease in Stormwater Runoff. University of California-Los Ange- les’ System’s Report, Association of Bay Area Governments, Oakland, CA (February 1982) 241 pp. Strecker, E., Iannelli, M., and B. Wu. Analysis of Oregon Urban Runoff Water Quality Monitoring Data Collected from 1990 to 1996. Woodward-Clyde (for the Association of Clean Water Agencies) (1997). Thomson, N. R., McBean, E. A., and I. B. Mostrenko. Prediction and Characterization of Highway Stormwater Runoff Quality. Report MAT-94-09, Ontario Ministry of Transportation, Research and Development Branch (1997). U.S. Environmental Protection Agency. Results of the Nationwide Urban Runoff Program Executive Summary. NTIS Access number PB 84-185545, Water Planning Division, Washington, DC (1983). Zhao, J. Q., Yu, Y., and W. N. Yuan. Water Quality of Storm Runoff from an Urban Highway. Proc., 8th International Con- ference on Urban Storm Drainage (1999). 3.4.3. Atmospheric Deposition Pollutants in the atmosphere consisting primarily of metals and nitrogen can be returned to the earth through processes of wet and dry atmospheric deposition. Atmospheric pollutants are generated from natural and anthropogenic sources. Natural sources include volcanic activity, windblown dust, forest fires, and vegetation. Anthropogenic sources include smelting of ores, fugitive dust from emission controls, and automobile emissions (Osmond et al., 1995). Wet deposition occurs when rain, snow, or fog bring down gaseous or particulate pollutants into the atmosphere. Dry deposition occurs when atmospheric pollutants find their way to the earth in the absence of precip- itation (Nilles, 2000). Mercury, lead, and other metals are con- trolled at the source as required by the provisions of the Clean Air Act (CAA), and emissions have been reduced signifi- cantly; however, pollutants deposited previously and atmos- pheric nitrogen emitted from various sources—some unregu- lated by the CAA—still pose a significant threat to the environment (Osmond et al., 1995). Potential knowledge gaps in the area of atmospheric deposition include • Atmospheric pollutant monitoring and sampling meth- ods and technologies, • Modeling and estimation of atmospheric deposition and prediction of dispersion patterns, • Characterization of impacts of atmospheric deposition on receiving water systems and biota, and • Specific contribution of highways to atmospheric depo- sition.

Atmospheric deposition can be a significant source of trace metals. Research performed by Atasi et al. (1998 and 2000) attributed cadmium and mercury levels in stormwater runoff to atmospheric deposition. The objective of the 1998 study was to investigate the impact of atmospheric deposition on surface runoff, the combined sewer system, and the publicly owned treatment works. Phase I of the study sought to quan- tify and characterize atmospheric deposition in relation to stormwater loadings. Pollutants of interest were mercury, cadmium, and polychlorinated biphenyl (PCB). A Phase II follow-on study may determine the effects of atmospheric loadings on the Detroit Waste Water Treatment Plant. Monitoring and sampling activities were performed using state-of-the-art air and deposition equipment. Sampling was performed at sites located in three distinct geographical areas and land use types. For each of the three sites, researchers monitored precipitation, wind speed and direction, and tem- perature. The study concluded that almost 100% of the cad- mium and 36–90% of the PCBs in runoff could be attributed to deposition. Higher-than-expected mercury and cadmium levels were observed in runoff from one site; this observation was hypothesized to be linked to direct deposition from vehicular traffic. The study showed that atmospheric deposi- tion was the main source of the study pollutants. The Atasi et al. 2000 study investigated the concentrations of 12 trace metals in the atmosphere, in precipitation, and in runoff. The metals sampled included mercury, cadmium, anti- mony, aluminum, arsenic, chromium, copper, manganese, nickel, lead, vanadium, and zinc. Researchers used specialized equipment and ultraclean analytical methods to monitor mete- orological parameters as well as pollutant concentrations and observed that pollutant concentrations were related to spatial variations and dependent on land use. The conclusions indi- cated that atmospheric deposition is a significant source for trace metals within an urban watershed. Other studies also have established atmospheric deposition as a significant source of pollutants. Tsai et al. (2001) investi- gated the contribution of atmospheric deposition to loadings of selected pollutants in the San Francisco estuary. Pollutants studied included copper, nickel, cadmium, and chromium. Monitoring was performed from August 1999 through August 2000, at three different sites chosen to represent the various segments of the estuary. The study observed dry deposition fluxes of copper, nickel, cadmium, and chromium at concen- trations of 1100, 600, 22, and 1300 µg/m2/year respectively and at concentrations of 1200, 420, 110, and 230 ng/L respec- tively in precipitation. Researchers concluded that atmo- spheric deposition contributed less than 30% of the loadings for copper and nickel in stormwater runoff. Contributions for cadmium and chromium in stormwater runoff approximate contributions from effluent discharges. By combining direct loads to the estuary and indirect loads through stormwater runoff, atmospheric deposition may contribute up to three times as much loading as effluent discharges to the estuary. 107 Researchers are looking constantly for ways to harness technology for the solution of environmental problems. The use of computer models and the development of new moni- toring techniques and equipment is a growing area of atmo- spheric deposition research. Davies (1976) discusses the appli- cation of remote sensing to highway environmental problems. Remote sensing can be used to verify results from computer and mathematical models. It also can reveal the nature and concentrations of pollutant gases and can track the mass flow and transport of pollutants. Davies discusses instrumentation and computer models, such as Cospec and Gaspec, and com- puter models, including a grid-point model, a fixed-source sulphur dioxide model, and a carbon monoxide model. The pollutants contained in precipitation are acquired from the atmosphere either through rain-out which occurs within clouds or washout which occurs as precipitation leaves the cloud. Shiba et al. (1999) used numerical simulations and a mathematical model to investigate the origins of atmospheric pollution found in stormwater runoff. Researchers provide chemical and mathematical descriptions for the cloud drop- let acidification process, and they conclude that pollutants acquired during cloud formation constitute a significant part of the pollution process. The MAGICWAND model is used to simulate soil and water acidification attributable to atmospheric deposition. Bobba et al. (2000) successfully applied the MAGICWAND model to the Turkey Lake Watershed in central Ontario, Canada, to evaluate the effects of atmospheric change and deposition. Shivalingaiah and William (1983) discuss the use of a multiregression model ATMDST, NEWBLD, and SWMM3 to model the Chedoke Creek catchment in Hamilton. ATMDST was developed to simulate the dust fall and provide input data for NEWBLD to calculate pollutant accumulation. Researchers compared pollutographs from this approach to pollutographs generated from the unmodified SWMM3 algo- rithms. Pollutants modeled in this study are suspended sedi- ments, BOD, total nitrogen, and total phosphorous. Sharma and McBean (2002) developed an atmospheric dis- persion model for the transport of PAHs using two indepen- dent data sets from Ontario, Canada. The object of the inves- tigation was to simulate PAH transport and accumulation in an urban snow pack. Researchers concluded that dry weather deposition is a dominant process in the urban environment. Estimates of deposition velocities and washout ratios were comparable to values obtained in previous investigations. Researchers have attempted to link atmospheric deposi- tion to external variables such as land use and surface type. Halverson et al. (1982) observed higher concentrations of metals runoff from highly used areas. Runoff sources used in the study included through-fall and stream flow from an urban tree, a suburban residential roof and street, a moderately used shopping center, and a heavily used highway. The authors found that the shopping center and the highway were the pri- mary sources of metals and sulphates. Cadmium, manganese, and copper were observed in only a few samples at very low

concentrations. Garnaud et al. (1999) selected four sites in the Paris metropolitan area to investigate dry and wet weather deposition in an attempt to better understand metal transport and metal distributions between dissolved and particulate fractions. Researchers provide a comparison of both dis- solved and particulate atmospheric deposits from four roofs, three yards, six gullies, and one catchment outlet. The authors observed medium-range transport of atmospheric pollutants. 3.4.3.1. Identification of Research Needs There appears to be a paucity of studies that directly relate highways and transportation systems to atmospheric deposi- tion. Filling this gap would provide a better basis to under- stand how highway-specific atmospheric deposition and dis- persion of highway-related pollutants affect receiving water quality, receiving water biota, and roadside ecosystems. There is also a need to quantify the contribution of atmos- pheric deposition to pollutant concentrations found in high- way runoff. Stormwater runoff data collected as part of DOT NPDES monitoring programs have revealed that surface runoff from highways and other DOT facilities contains pol- lutants known to be unrelated to transportation activities (except as spilled during transport). Most of these pollutants are organic and include chemicals with a wide range of volatil- ity. The contribution of organic and inorganic pollutants from atmospheric deposition likely differs between urban- ized and nonurbanized areas. The fraction of pollutants con- tributed by atmospheric deposition is not known for different land uses or classes of contaminants. Research Objectives. Further research in atmospheric depo- sition would enable DOTs to • Work with other dischargers to reduce pollutants on a watershed basis, • Show that they are not responsible for everything appear- ing in their surface runoff, and • Implement the best management practice on a regional basis to better control the organic and inorganic pollu- tants of concern. The following areas were ranked relatively low by state DOTs but could be considered potential research tasks on a second tier list of research priorities based on gaps in the literature: • Create a GIS database of regional atmospheric deposi- tion using existing Air Resources Board ambient moni- toring data and published deposition velocities (for dry deposition) and washout ratios (for precipitation inputs) to predict mass inputs to varied watersheds. • Measure atmospheric deposition (both dry deposition and precipitation inputs) of major pollutants in major 108 regions under various land uses. These measurements will be used to validate and calibrate the estimates pro- duced in task 1. • Identify the pollutants contributed in significant amounts by atmospheric deposition on a regional- and land-use basis using the results of the first two tasks. • Identify sources of organic and inorganic pollutants using established “fingerprinting” techniques, including atmo- spheric tracers and elemental ratios. • Refine the initial GIS model to predict the atmospheric pollutant contributions and their relative loads on regional and land use bases. 3.4.3.2. Primary References Atasi, K. Z., Chen, T., Hufnagel, C., Kaunelis, V., and G. Keeler. Atmospheric Deposition and Runoff of Mercury and Trace Metal in an Urban Watershed. Proc., WEFTEC 2000 Conference, Ana- heim, CA (2000) 13 pp. Atasi, K. Z., Fujita, G., Geoffrey, L. P., Hufnagel, C., Keeler, G., Graney, J., and T. P. Chen. Impact of Atmospheric Deposition on Surface Water Runoff of Mercury, Cadmium and PCBs. Proc., 71st Water Environment Federation Annual Conference, Orlando, FL (October 3–7, 1998). Bobba, G., Jeffries, D. S., and R. G. Semkin. Application of MAGIC- WAND Model to Turkey Lakes Watershed, Canada, to Predict Changes in Stream Water Quality due to Atmospheric Deposition and Climate Change. 2000 Annual Meeting and International Con- ference of the American Institute of Hydrology (November, 2000). Davies, J. H. Remote Sensing of Atmospheric Pollutants to Assess Environmental Impact of Highway Projects (Abridgment). In Transportation Research Record 594, TRB, National Research Council, Washington, DC (May 1976) pp. 6–9. Garnaud, S., Mouchel, J. M., Chebbo, G., and D. R. Thevenot. Heavy Metal Concentrations in Dry and Wet Atmospheric Deposits in Paris District: Comparison with Urban Runoff. Sci- ence of the Total Environment, Vol. 235, No. 1-3 (1999) pp. 235–245. Halverson, H. G., DeWalle, D. R., Sharpe, W. E., and D. L. Wirries. Runoff Contaminants from Natural and Manmade Surfaces in a Nonindustrial Urban Area. University of Kentucky Urban Hydrol- ogy, Hydraulics and Sediment Control Symposia (July 1982) pp. 233–238. Nilles, M. A. Atmospheric Deposition Program of the U.S. Geo- logical Survey. Fact Sheet FS-112-00, U.S. Geological Survey, Washington, DC (December 2000). Osmond, D. L., Line, D. E., Gale, J. A., Gannon, R. W., Knott, C. B., Bartenhagen, K. A., Turner, M. H., Coffey, S. W., Spooner, J., Wells, J., Walker, J. C., Hargrove, L. L., Foster, M. A., Robillard, P. D., and D. W. Lehning. WATERSHEDS: Water, Soil and Hydro- Environmental Decision Support System. (1995). Sharma, M., and E. A. McBean. Atmospheric PAH Deposition: Deposition Velocities and Washout Ratios. Journal of Environ- mental Engineering, Vol. 128, No. 2 (February 2002) pp. 186–195. Shiba, S., Hirata, Y., and S. Yagi. Acid Cloud Droplet Formed By Condensation of Atmospheric Water Vapor as Pollution Source of Urban Runoff. Proc., 8th International Conference on Urban Storm Drainage (August 1999) pp. 1528–1535.

Shivalingaiah, B., and J. William. Atmospheric and Land-Based Loadings for Stormwater Runoff Quality Models. University of Kentucky Urban Hydrology, Hydraulics and Sediment Control Symposia (July 1983) pp. 115–122. Tsai, P., Hansen, E., Lee, K., Yee, D., Tucker, D., and R. Hoenicke. Atmospheric Deposition of Trace Metals in the San Francisco Bay Area. Proc., Water Environment Federation Technical Exhi- bition and Conference 2001 (2001). 3.4.4. Highway Construction Materials The contribution of highway construction materials as a source of runoff pollutants cannot be overlooked. The end- less rehabilitation and maintenance of the system of high- ways and the move to use new materials and recycled prod- ucts in highway construction has increased significantly the chances of runoff contamination from highway construction materials. Research needs and possible knowledge gaps per- taining to highway materials as a source of runoff materials suggested by Nelson et al. (2001) include • Expansion of available material data, • Soil sorption and desorption research, • Role of aluminum, • Temperature effects, • Leaching mechanisms, and • BMPs for mitigating impacts from leached chemicals from highway construction impacts. A comprehensive NCHRP study (Project 25-09) pre- sented by Nelson et al. (2001) investigated the potential environmental impacts of highway construction and repair (C&R) materials on surface water runoff and groundwater quality. The study’s main objective was to develop and eval- uate methodologies for identification of possible surface and groundwater impacts from construction materials. Materials evaluated include asphaltic materials (such as asphalt cement), cementitious materials (such as Portland cement), industrial– manufacturing by-products (such as coal combustion fly ash, aggregate dust palliatives, and wood preservatives), and other miscellaneous materials (such as reflective glass spheres and ground tire rubber—“crumb rubber”). Amendment of most test materials with asphalt or aggre- gate resulted in a reduction or elimination of aquatic impacts. Soil sorption was identified as the most effective pollutant removal mechanism. The mathematical model IMPACT was developed for performing fate and transport analysis for sur- face and subsurface pathways. The result of this study was the successful development and evaluation of a complete methodology for screening and evaluation of potential envi- ronmental impacts of highway C&R materials. Toxicity tests using algae and Daphnia were conducted to determine the toxicity level in water elutriates prepared from selected road construction and maintenance materials that emulate stormwater runoff (Eldin, 2002). Many of the tested 109 construction materials proved to be toxic to the test organ- isms. Heavy metals such as aluminum, arsenic, lead, mer- cury, and some hydrocarbon compounds present in the test elutriates appeared to be major causes of toxicity. However, measured toxicity was reduced greatly or eliminated when elutriates were allowed to be in contact with selected soils. Under actual field conditions, mechanisms other than soil sorption—such as volatilization, photolysis, and biodegrada- tion—are believed to further reduce the toxicity of storm- water runoff. There appears to be a significant amount of research on finding either new materials for highway construction or ways to reuse existing materials and by-products. Schroeder (1994) presents a synthesis of information about various new and existing materials that are being used in highway construc- tion. He also cites examples of DOTs and other public orga- nizations that have used and evaluated alternative materials for highway construction or repair. Completed studies related to highway construction and repair materials as sources of pollutants are relatively scarce compared to the number of ongoing studies. EPA, in conjunction with the Vanderbilt University in Nashville, Tennessee, has an ongoing study to develop test- ing and interpretation protocols for evaluating leaching from granular alternate aggregate replacement materials. Mathe- matical models and interpretation protocols will be used to evaluate the environmental impacts of leaching that occurs as a result of intermittent infiltration of precipitation into aggregate. Other related research includes an FHWA-sponsored study being performed by the National Academy of Sciences; the study is called “Impacts of Significant Waste Materials Uti- lized in Highway Construction.” For more information about this project, see the Transportation Research Board’s (TRB’s) Research-in-Progress (RIP) website. An ongoing joint effort between the EPA and New Jersey DOT aims to evaluate the use of recycled materials in high- way construction. The goals of this study include evaluation of long-term pavement durability, evaluation of environ- mental concerns, and cost-effectiveness of using recycled materials. Details of this effort are available on TRB’s RIP website. 3.4.4.1. Identification of Research Needs A literature review on the subject of highway construction and maintenance materials as a source of runoff contaminants reveals a limited number of studies on the subject (NCHRP Project 25-09 is the most comprehensive to date) and a sig- nificant amount of research in progress. Currently, potential knowledge gaps include the availability of materials proper- ties data, knowledge of soil leaching and the sorption and de- sorption processes, a better understanding of the chemistry of aluminum in complex mixtures of chemical leachate and in soils, the effects and impacts of temperature, and a better

understanding of the capabilities of existing BMPs to mitigate impacts from highway construction material contamination. 3.4.4.2. Primary References Eldin, N. N. Road Construction: Materials and Methods. Journal of Environmental Engineering, Vol. 128, No. 5 (2002) pp. 423–430. Nelson, P. O., Huber, W. C., Eldin, N. N., Williamson, K. J., Azizian, M. F., Thayumanavan, P., Quigley, M. M., Hesse, E. T., Lundy, J. R., Frey, K. M., and R. B. Leahy. NCHRP Report 448: Environ- mental Impact of Construction and Repair Materials on Surface and Ground Waters: Summary of Methodology, Laboratory Re- sults, and Model Development. TRB, National Research Council, Washington, DC (2001) 134 pp. Schroeder, R. L. The Use of Recycled Materials in Highway Con- struction. Public Roads, Vol. 58, No. 2 (Autumn 1994) pp. 32–41. 3.4.5. Fate and Transport The topic of fate and transport is very broad with whole university courses and several environmental engineering and science textbooks dedicated to the subject. This subsec- tion touches on the surface of this vast topic with regard to highway runoff. The topic areas identified as the most impor- tant to highway characterization and assessment are sedi- ment transport, speciation, and sorption processes. Other fate and transport topics such as advection, dispersion, diffusion, and volatilization are considered a lower priority in relation to the research needs of highway stormwater management. 3.4.5.1. Sediment Transport Very little attention was given to urban sediment prob- lems in the past. However, recent booms in housing devel- opment, road construction activities, and other large-scale earth-moving activities have drawn more attention to the urban sediment problems that previously received little attention. Urban sediment studies have now expanded in scope to include sediment quality as well as sediment quantity con- cerns. Even though much progress has been made, certain areas require more work; Brush (1981) suggested the fol- lowing knowledge gaps and areas of interest: • Characterization of particle sizes and settling velocities; • Sediment transport mechanics such as transport and blockage potential of partly full and full conduits in noncircular cross-sections, local storage of sediment at various inlets and through various types of transitions in drainage systems; and • Sediment yields in relation to various soils, topography, land use, and different kinds of storm hydrographs. Sediment yield is the amount of sediment removed from a watershed at a given time. Fusillo et al. (1997) studied sedi- 110 ment yields for a watershed in central New Jersey. The authors discovered that construction sites contributed about 80% of the sediment for the basin, which is almost 50 times as high as yields for other land use areas. They observed that sedi- mentation basins installed at construction sites may reduce significantly sediment loads to streams. Several researchers have performed studies and experi- ments with the aim of extending existing sediment transport computer models or creating new ones. Ziegler et al. (2001) replaced the step function in the KINEROS2 model with an exponential decay function. They observed that the method improved the continuous sediment transport time series esti- mate but underestimated peak sediment output, just like the original version of the model did. They concluded that peak output estimates may be improved through optimization using rainfall simulation data. They also recommended that the method be validated at the hill slope scale before its use- fulness for simulating road erosion can be determined. The Queen’s University Urban Runoff Model evolved from the integration of a sediment transport model into an urban runoff model. The sediment transport model is based on the equivalent solids reservoir concept and requires only simple quantity and quality inputs. The model is capable of simulat- ing sediment transport surfaces, gutters, pipes, and detention ponds. Schroeter and Watt (1983) tested the model with data from a stormwater sampling program in Kingston, Ontario, and the results of an independent study in Burlington, Ontario. The Hydrologic Simulation Program-Fortran (HSPF) was used to model stream flow and TSS within Contentnea Creek in North Carolina. Input data for the model included land use data from EarthSat and historical meteorological data such as precipitation, pan evaporation, and temperature. The model was calibrated on historical time series of observed flow obtained from USGS. The study concluded that the simu- lated flows were a good fit to observed flows while TSS con- centrations were replicated less accurately. The parameter values used in the study were well within the range of values used in other HSPF studies (Cyterski, 2000). HYPOCRAS is a French-made model that is used to sim- ulate the transport of solids in sewers. Ashley and Bertrand- Krajewski (1993) present a discussion about their work to extend and test the HYPOCRAS model. Other models in this category include SWMM, MOSQITO, and FLUPOL. The authors used data from a substantial data collection program in Dundee, Scotland, supported by the United Kingdom’s Water Research Center for extending and testing the model. Two deterministic suspended solids models, one based on SWMM3 algorithms and the other based on fundamental erosion mechanics and sediment transport processes, were applied to two urban catchments. The models were tested with historical data collected on catchments in Pinetown and Alexandria; 16 events and 12 events were tested, respec- tively. To assess the performance of the models, Coleman (1993) compared the ratio of predicted load to observed load for each event and compared pollutographs from the models

to observed pollutographs using the coefficient of efficiency for each event. Computer models are useful for processing large quanti- ties of data and solving complex problems; however, in situ- ations in which quick estimates of quantities are needed, equations and formulas suffice. Younkin and Connelly (1981) developed an equation based on regression analysis of data from nine stream gages and five watersheds in Pennsyl- vania. The equation can be used to estimate the increase in sus- pended sediment yield in a stream due to highway construc- tion. The equation relates factors such as soil erodibility, rainfall, construction phases, and site proximity to stream as well as increases in quantity of transported sediment. The equation may find applications in highway location studies, highway development impact assessment, and the design of sediment control devices. Other researchers have investigated the significance of sed- iment particle size in sediment transport processes. Ota et al. (1999) investigated the effects of particle size on sediment transport in sewers. They tested three uniform materials of varying sizes and observed that test results were very sensi- tive to particle size. Transport rates were observed to be high for finer sand. Coarse material was harder to transport. The graded material was studied further using two different mate- rials. Modified Meyer-Peter and Muller bed load functions were used to fit sediment transport rate for uniform materials. 3.4.5.2. Speciation of Constituents Speciation of heavy metals in aquatic systems plays a key role in their transport, chemical reactions, and bioavailabil- ity. Physical and chemical forms that may cause significant consequences, known as consequential species, must be iden- tified before the potential environmental impact of the metal can be assessed adequately, since biotoxicity is dependent on the available species and not the total metal concentration. Yousef et al. (1985) investigated heavy metal speciation in rainfall and highway runoff at two sites in central Florida: at the intersections of Maitland Interchange and I-4 and at US- 17-92 and Shingle Creek. Total and dissolved fractions of cadmium, copper, lead, and zinc were determined in the study. The dissolved metals fractions were further divided according to behavioral differences and bioavailability. Dis- solved metals were first divided according to labile (reactive) and nonlabile (nonreactive) compounds, then according to organic and inorganic, and finally by soluble and colloidal. As dissolved metals do not exist as labile-organic-soluble, there are a total of seven possible speciation classifications. Results indicate that the labile, organic, and colloidal fractions aver- age 82.0%, 5.3%, and 3.2% for cadmium; 92.9%, 0.3%, and 42.7% for zinc; 60.9%, 22.1%, and 55.6% for lead; and 63.7%, 48.9%, and 69.8% for copper in all water samples tested. Therefore, the authors conclude that zinc and cadmium are more reactive, may exist in ionic forms, and are more readily available to biota in natural environments than copper 111 and lead. Other significant conclusions of the study include the following: (1) acidic rainfall generally is neutralized on contact with paved surfaces, and (2) carbonates and fulvates have a substantial effect on dissolved metal speciation, with the tendency to form complexes that are not as bioavailable. In another speciation study by Morrison et al. (1984) zinc, cadmium, lead, and copper stormwater samples collected from selected urban catchments in England and Sweden were analyzed. The study found that zinc and cadmium exhibited a preference for the dissolved phase, whereas lead predomi- nated in the suspended solid phase. Copper was distributed equally between both phases. Furthermore, the potentially toxic forms of the metals in the dissolved phase (electro- chemically available) and in the particulate phase (exchange- able) accounted for 63% of the total zinc, 77% of the total cadmium, 66% of the total lead, and 32% of the total copper. The biogeochemical processes affecting metal speciation in a gullypot system and at stormwater outfalls were investi- gated by Morrison et al. (1989). Ionic lead and copper species released from road sediments by acid rain are scavenged by dissolved organic material and suspended solids as a result of a rise in pH through the road–gullypot system. Cadmium tends to remain in the dissolved phase. Bacterial activity and acid dissolution produce increases in dissolved metal in the gullypot liquor, and it is these metals that contribute to the early storm profile. Metals in basal gullypot sediments are mobilized readily during high-volume, high-intensity storms. The resulting stormwater contains dissolved ionic forms of cadmium and zinc, and lead is adsorbed mainly to suspended solid surfaces. Copper also binds to solids, although approxi- mately 50% is transported by dissolved organic material (mo- lecular weight ∼ 1000–5000). For the separation of directly toxic metal species, anodic stripping voltammetry at polymer- coated electrodes is preferred. Lead and copper are present respectively as iron/humic colloids and organic complexes, which are not directly toxic to algae. Cadmium is predom- inantly ionic and inorganically complexed and therefore directly toxic to algae. Glenn et al. (2002) examined heavy metal (cadmium, cop- per, lead, and zinc) partitioning results for a series of rainfall runoff events and found that aqueous chemistry, such as low alkalinity and hardness, and short pavement residence time (less than 30 minutes) can result in a majority of the heavy metal mass remaining in solution at the edge of the pave- ment. Metals partitioning approaches equilibrium conditions only toward the end of the event as heavy metals partition to entrained solids. 3.4.5.3. Sorption Processes Sorption refers to the removal of a solute (sorbate) from the solution phase by the solid phase (sorbent). The two basic categories of sorption are adsorption (when the sorbate inter- acts with the surface of the sorbent) and absorption (when the sorbate penetrates the surface of the sorbent).

As a result of sorption of heavy metals onto particulate mat- ter such as iron and manganese oxyhydroxides or organic mat- ter, the concentrations of metals in natural waters are com- monly far lower than would be predicted from simple min- eral solubility calculations (Bricker, 1999). As such, sorption processes often are used in stormwater BMP technologies. One important process responsible for the sorption of cations is ion exchange. The negative charge on soil colloids, clay, and organic matter on soil surfaces makes ion exchange one of the most important reactions influencing transport of cations in soils (Bricker, 1999). Ion exchange involves the sorption of one or more species of ions accompanied by the desorption of the previously sorbed species equivalent in total ionic charge. Soils often have surfaces with a net nega- tive charge because of, for example, isomorphic substitution of ions in a clay lattice structure. An electrostatic double- layer is formed when the negative surface charge is counter- balanced by cations, which accumulate on the surface of the particle forming an electrostatic double-layer. This double- layer provides the ability of the matrix to attract ions and eventually to attenuate them. Sorption processes usually are thought of as beneficial to stormwater quality due to the tendency for pollutants to adsorb and settle out with sediments. However, Davies and Bavor (2000) found that the adsorption of thermotolerant coliforms to fine clay particles (<2 µm) contributed to their survival in stormwater treatment systems. Other studies iden- tified by Bricker (1999) have found that metals and trace organic chemicals also tend to adsorb to fine particulates, with metal concentrations on particulates tending to increase with decreasing particle size (increasing surface area), and that the suspended particulates in highway runoff contained higher overall metal concentrations than road surface dusts. 3.4.5.4. Identification of Research Gaps and Needs Uncertainty exists in identifying sediment sources and defining transport rates and residence time of sediment in receiving waters. With respect to sediment transport, there appears to be an abundance of information on sediment trans- port models. However, there may be a need for more detailed studies of sediment transport mechanics in relation to block- age of full and partly full conduits in various cross-sections. Comprehensive studies on the effects of soils, topography, land use, and various storm hydrographs on sediment yield appear to be limited in number. In addition, the behavior of sediment at inlets, junctions, and transitions in the drainage system may require further study. Good predictive models that consider runoff–storm relationships, particularly storm scour and redeposition, are unavailable. Research on the speciation of pollutants has been primar- ily on the common metals found in highway runoff, cad- mium, copper, lead, and zinc. It is well documented that the dissolved phase of the metals are more bioavailable and 112 therefore more toxic to aquatic biota than the particulate phase. In fact, in 1993 the EPA’s Office of Water revised its policy to use dissolved metals concentrations rather than total recoverable metals concentrations to set and measure compliance with water quality standards. The amended National Toxics Rule now includes dissolved metals aquatic life criteria (40 CFR 131). Using dissolved metals instead of total recoverable metals for the purposes of assessing impacts to aquatic life is a step in the right direction, but using dissolved metals alone still does not appear to be an adequate measure of aquatic toxicity. As discussed in the lit- erature above, the reactive and ionic portions of dissolved metals concentrations are more available to aquatic biota than the nonreactive and nonionic portions. Therefore, there appears to be a need for better characterization of the bioavailable fraction of dissolved metals, as well as trace organics, in highway runoff. Sorption plays an important role in the speciation and bio- availability of pollutants. However, the factors controlling sorption, such as cation exchange capacity and specific sur- face area, are not investigated often. Highway agencies could benefit from information on the sorption capacity of roadside soils for the purposes of prioritizing retrofits and installations of treatment control practices. Native soils may have the capacity to retain pollutants, which would circumvent the need for additional treatment controls. Based on this fact, there appears to be a general need for more research on the sorp- tion of pollutants to sediment in highway runoff. 3.4.5.5. Primary References Ashley, R. M., and J. L. Bertrand-Krajewski. Sewer Sediment Ori- gins and Transport in Small Catchments. Proc., 6th International Conference on Urban Storm Drainage (July 1993) pp. 772–777. Bricker, O. P. An Overview of the Factors Involved in Evaluating the Geochemical Effects of Highway Runoff on the Environment. Open File Report 98-630, U.S. Geological Survey, Washington, DC (1999) 38 pp. Brush, L. M. Urban Sediment Problems. Proc., 2nd International Conference on Urban Storm Drainage (June 1981) pp. 518–524. Coleman, T. J. A Comparison of the Modeling of Suspended Solids Using SWMM3 Quality Prediction Algorithms with a Model Based on Sediment Transport Theory. Proc., 6th International Conference on Urban Storm Drainage (July 1993) pp. 790–795. Cyterski, M. Hydrology and Sediment Modeling Using the BASINS Non-Point Source Model. 2000 Annual Meeting and International Conference of the American Institute of Hydrology, Research Tri- angle Park, NC (November 2000). Davies, C. M., and H. J. Bavor. The Fate of Stormwater-Associated Bacteria in Constructed Wetland and Water Pollution Control Pond Systems. Journal of Applied Microbiology, Vol. 89, No. 2 (2000) pp. 349–360. Fusillo, T. V., Nieswand, G. H., and T. B. Shelton. Sediment Yields in a Small Watershed Under Suburban Development. University of Kentucky Urban Hydrology, Hydraulics and Sediment Con- trol Symposia (July 1977) pp. 302–308.

Glenn, D., Liu, D., and J. Sansalone. Influence of Chemistry, Hydrology and Suspended Solids on Partitioning of Heavy Met- als to Particles—Considerations for In-Situ Control of Urban Stormwater Quality. Global Solutions for Urban Drainage, Proc., 9th International Conference on Urban Drainage (September 8–13, 2002). Morrison, G. M. P., Revitt, D. M., and J. B. Ellis. Metal Speciation in Separate Stormwater Systems. Proc., 3rd International. Sym- posium on Highway Pollution (1989) pp. 53–60. Morrison, G. M. P., Ellis, J. B., Revitt, D. M., Balmer, P., and G. Svensson. The Physico-Chemical Speciation of Zinc, Cadmium, Lead and Copper in Urban Stormwater. Proc., 3rd International Conference on Urban Storm Drainage, Göteborg (June 1984) pp. 989–1000. Ota, J. J., Nalluri, C., and G. Perrusquia. Graded Sediment Trans- port–The Influence of Particle Size on Sediment Transport over Deposited Loose Beds in Sewers. Proc., 8th International Con- ference on Urban Storm Drainage (August 1999) pp. 626–634. Schroeter, H. O., and W. E. Watt. Practical Simulation of Sediment Transport in Urban Watersheds. University of Kentucky Urban Hydrology, Hydraulics and Sediment Control Symposia (July 1983) pp. 411–420. Younkin, L. M., and G. B. Connelly. Prediction of Storm-Induced Sediment Yield from Highway Construction. In Transportation Research Record 832, TRB, National Research Council, Wash- ington, DC (June 1981). Yousef, Y. A., Harper, H. H., Wiseman, L., and M. Bateman. Con- sequential Species of Heavy Metals. Environmental Research Report FL-ER-29-85, State of Florida, Department of Trans- portation, Bureau of Environment (1985). Ziegler, A. D., Giambelluca, T. W., and R. A. Sutherland. Model- ing Road Sediment Transport with KINEROS2. Proc., Soil Ero- sion for the 21st Century—An International Symposium, Hon- olulu, HI (January 3–5, 2001). 3.4.6. First Flush Characterization The tendency for concentrations of stormwater runoff pol- lutants to increase rapidly at the onset of a storm and then to decline slowly is known as the first flush phenomena. First flush can be caused by the accumulation of surface pollutants during dry weather and the subsequent wash-off of those pol- lutants during wet weather. Thus, the first storm of the wet season usually results in the highest first flush concentrations due to the length of the preceding dry period. This is not always the case though; in fact, a discernable first flush period is not evident for some watersheds and pollutants. Another issue is that higher flushes of pollutants have been observed later in storm events when rainfall intensities increase or per- vious areas start contributing to runoff, or both. Furthermore, there is no clear agreement among stormwater professionals how the first flush should be defined. This leads to the fol- lowing research questions: • How can first flush be meaningfully defined? • What water quality parameters are observed commonly in the first flush of highway runoff? 113 • How do hydrological factors and watershed characteris- tics relate to first flush? Several researchers have provided definitions of the first flush. Barbosa and Hvitved-Jacobsen (1999) noted that a storm event exhibits a first flush if the first 50% of the runoff volume contains greater than 50% of the loads. Deletic (1998) defined the first flush as the pollutant load carried by the first 20% of the runoff volume. Several other definitions used by various researchers were summarized by Ma et al. (2002), such as first flush is when at least 80% of the pollutant load is emitted in the first 30% of the runoff or simply the first 25% of the runoff volume (assuming a mass first flush actually occurs). Another definition is provided, and actually recom- mended by Ma et al. (2002): first flush is when the slope, or mass first flush ratio, of the normalized cumulative mass emission versus the normalized volume is greater than 1. This is a useful definition, because unlike the other methods it does not depend directly on the size of the storm event or on the total loads discharged. Also, the definition provides metrics for first flush magnitude and timing, which can be used to size structural stormwater BMPs based on the fraction of loads desired to be captured and to determine when to take first flush grab samples. This method does require that storm event samples are analyzed before compositing, so the monitoring costs associated with first flush characterization are signifi- cantly greater than for general runoff characterization. A broad range of the pollutants found in stormwater runoff will exhibit a first flush depending on the drainage hydrology, pollutant mobility, and pollutant supply. With regard to typi- cal highway pollutants, Barbosa and Hvitved-Jacobsen (1999) observed a first flush effect for TSS, zinc, copper, and lead loads. Smith et al. (2000) noted that PAH concentrations were highest usually during the first flush of stormwater runoff and that they tapered off rapidly as time progressed. Lau et al. (2002) reported COD, oil and grease, dissolved organic car- bon, and particulate phase PAHs all exhibited a first flush. Wachter and Herrmann (2002) noted that trace organic pollu- tographs of the particle-bound fraction showed a first flush effect, while Deletic (1998) observed only slight first flush effects for suspended solids and conductivity and no first flush effect for pH or temperature. Therefore, it appears that solid-phase pollutants typically exhibit a first flush effect depending on whether or not the source is continuous or sub- ject to buildup and washoff. Hydrological factors such as rainfall intensity and spatial variability, and watershed characteristics such as watershed size, slope, stream order, and percent imperviousness are all factors that likely are associated with flush phenomena. Under- standing how these factors influence the flush of pollutants may circumvent the need for site-by-site first flush characteri- zation. Lee and Bang (2000) analyzed pollutographs from storm events in nine watersheds in the cities of Taejon and Chongju, Korea, and found that for watersheds less than 100 ha with a total imperviousness of 80%, the peak of pollutant concentration preceded that of the flowrate, but for watersheds

greater than 100 ha with a total imperviousness of less than 50%, the peak of pollutant concentration was followed by that of flowrate. Caltrans has completed several first flush studies and a final report. Preliminary findings or conclusions can be sum- marized as follows: 1. Preliminary results show the existence of a first flush for some parameters, especially for parameters such as oil and grease and COD. For medium-size storms, there is generally 40% of the total mass load in the first 20% of the runoff volume. In some cases metals show a first flush as well. Some parameters, such as the sulfate ion, show a last flush. 2. Strong correlations exist among many of the water qual- ity parameters and metals. Antecedent dry periods in some cases show trends (greater contaminant concen- trations with larger elapsed time between storms), but so far there are insufficient data to show statistically significant correlations. 3. An extensive database incorporating the results from all sites is being developed and analyzed. Various hypotheses are being tested, including correlations among parameters to determine if surrogate parameters are useful. 4. In most instances, a first flush phenomenon also was observed for the gross pollutant and litter concentra- tions. However, the gross pollutant and litter mass load- ing rates were not highest during the first flush but gen- erally appeared to correlate with the peak flow rate, which is similar to the water quality data. The total lit- ter volume generated appeared to be related to the rel- ative intensity of the storm event. The litter mass load- ing rates also did not generally decrease across the storm season. 5. A procedure and notation is developed for quantifying mass first flush ratios. The notation allows mass first flushes to be analyzed statistically. 6. The concentrations of PAHs were low, generally at or below detection limits in the dissolved phase. Particu- late phase PAHs are reported and show a first flush, although there were fewer monitored events. 7. The concept of collecting a grab sample at the best time to approximate EMC for oil and grease was investigated. Caltrans also initiated research on the first flush of the parti- cle size during the 2002–2003 monitoring season. See sec- tion 3.5.4, Toxicity and Bioassessment, for further discus- sion of toxicity studies. In Portland’s stormwater monitoring for the NPDES per- mit application efforts and beyond, interstorms were sam- pled to explore within storm variability (Strecker, 2003). Results generally showed that pollutants associated with particulates did show a tendency to wash off earlier in storms. Constituents such as dissolved metals either showed no change 114 or increased during the storm. Consequently, while pollutant loads decreased, toxicity potentially increased. Herricks found similar results in his urban runoff sampling work. 3.4.6.1. Identification of Research Gaps and Needs Several researchers have identified a first flush effect dur- ing runoff characterization studies, but nearly all use a dif- ferent definition. The mass first flush ratio used to define first flush by Ma et al. (2002) appears to be the most meaningful and does not depend on the time of concentration like other definitions that are based on time from beginning of the storm. There appears to be a need for the adoption of a stan- dardized method for defining and identifying first flush phe- nomena. Also, some parameters appear to exhibit a first flush, while others do not. Therefore, a comprehensive list of high- way runoff pollutants that tend to exhibit a first flush may be useful for evaluating receiving waters impacts and the feasi- bility of treating only the first flush part of a storm. First flush toxicity information in conjunction with other first flush characterization data can be used to design BMPs that can treat properly the early portion of runoff and bypass the rest for most small watersheds. The current research effort did not find any studies that specifically investigated how the first flush effect was related to hydrological and watershed characteristics, indicating a potential research gap with regard to first flush characterization and assessment. 3.4.6.2. Primary References Barbosa, A. E., and T. Hvitved-Jacobsen. Highway Runoff and Potential for Removal of Heavy Metals in an Infiltration Pond in Portugal. Science of Total Environment, Vol. 235, No. 1-3 (1999) pp. 151–159. Deletic, A. First Flush Load of Urban Surface Runoff. Water Research, Vol. 32, No. 8 (1998) pp. 2462–2470. Lau, S-L., Ma, J-S., Kayhanian, M., and M. K. Stenstrom. First Flush of Organics in Highway Runoff. Proc., 9th International Conference on Urban Drainage, Portland, OR (September 8–13, 2002). Lee, J. H., and K. W. Bang. Characterization of Urban Stormwater Runoff. Water Resources (G.B.), Vol. 34, No. 6 (2000) pp. 1772–1780. Ma, J-S., Khan, S., Li, Y-X., Kim, L-H., Ha, S., Lau, S-L., Kayhanian, M., and M. K. Stenstrom. First Flush Phenomena for Highways: How It Can Be Meaningfully Defined. Proc., 9th International Conference on Urban Drainage, Portland, OR (September 8–13, 2002). Smith, J. A., Sievers, M., Huang, S., and S. L. Yu. Occurrence and Phase Distribution of Polycyclic Aromatic Hydrocarbons in Urban Stormwater Runoff. Water Science and Technology, Vol. 42, No. 3 (2000) pp. 383–388. Strecker, E. W. City of Portland—NPDES Monitoring Results. Personal Communication (2003).

3.4.7. Water Quality Runoff Modeling Runoff models can be grouped into the following three main categories based on the method of analysis used in the model: regression-based models, simulation-based models, and stochastic models. Regression-based models are rela- tively simple and are sometimes no more complex than sim- ple equations. Simulation-based models are models that are capable of using historical data to account for temporal vari- ations in the variables of interest. Stochastic models are mod- els that are founded on the principles of statistics and proba- bility. Models also can be grouped by the kinds of processes that they simulate. Geochemical models are specialized mod- els that are used mostly to simulate unit processes. The areas of interest and possible knowledge gaps with respect to run- off water quality modeling are as follows: • The availability of data, data replacement and updating, and new data requirements; • Development of hybrid models to benefit from the advan- tages of each of the categories; • Extension of models to predict loads of a wider range of pollutants; and • Simulation of herbicide and pesticide concentrations and transport processes. The main advantages of regression models are in their rela- tive simplicity. Regression methods also tend to have minimal data requirements; however, these models are less capable of simulating temporal and spatial variations. The advantages of simulation models include the ability to simulate the effects of changes and abatement effects by a simple alteration of parameters and the provision of temporal and spatial distri- butions. Simulation models tend to have the most substantial data requirements, which can be problematic due to the high cost of data collection. Statistical models offer the ability to produce a distribution instead of the mean concentrations provided by regression-type analysis. Distributions can be used then to estimate the probabilities of exceedance of spec- ified concentrations. However, statistical models are not as capable as simulation models at simulating either the inter- actions of flow and concentrations or the effect of abatement actions (Barrett et al., 1995). A review of literature related to runoff water quality mod- eling shows that there are numerous studies that developed or applied, or both, regression models as tools for runoff characterization. A study to identify the variables that affect highway runoff in Austin, Texas, applied regression analysis techniques for predicting pollutant loads. The results of the study suggested that highway stormwater loading variations during a storm event depend on variables measured during previous events, the antecedent dry period, and the current storm event. TSS loads were found to depend on the build up of pollutants before storms, the characteristics of the storm, and the wash-off processes. Oil and grease were found to 115 depend on current storm conditions such as runoff volume and number of vehicles during the storm event (Irish et al. 1998). A study presented by Kerri et al. (1985) resulted in the development of pollutant load estimation regression equa- tions for urban highway sites in Redondo Beach, Walnut Creek, and Sacramento in California. The regression equa- tions were based on continuous rainfall monitoring data and sequential water quality sampling data. Contaminants ana- lyzed included boron, total lead, total zinc, nitrate-nitrogen, ammonia-nitrogen, and TKN. The authors cautioned against using the regression equations in situations where ADT is less than 30,000 vehicles. The Washington State study presented by Chui et al. (1982) resulted in the development of a pollutant load model for Washington State. The model was based on extensive moni- toring data consisting of 500 storms at nine locations. The model correlates TSS loads, traffic conditions, runoff coeffi- cients, and land use. Pollutant load estimates for individual storms are determined less accurately by the model, as com- pared to multiple storms over a period of time. Examples of simulation models include EPA SWMM, STORM, HSPF, and the FHWA urban Highway Storm Drain- age Model (Barrett et al., 1995). DeVries and Hromadka (1993) present a comprehensive discussion of runoff water quality models. Models discussed include SWMM, HPSF, QUAL2E, WASP4, AGNPS, and MIKE 11. For each model, the authors present a general overview that includes a descrip- tion of the model’s origins and the applications for which the model was developed. A description of hardware require- ments and directions on how to obtain the model also are included. In some cases, the authors also discuss model com- ponents and the kinds of problems for which the model has been applied. Guay and Smith (1988) discussed the applica- tion and evaluation of DR3M-II and DR3M-qual. The mod- els were applied to a multiple-dwelling residential catchment and a commercial catchment in Fresno, California. Calibra- tion and verification of errors for dissolved solids, dissolved nitrite plus nitrate, total recoverable lead, and suspended solids ranged from 11% to 54%. Statistical methods were applied in the analysis of runoff quality in the National Urban Runoff Program (NURP), EPA’s comprehensive 5-year runoff characterization study in which runoff samples from 28 metropolitan areas across the United States were collected and analyzed (U.S. EPA, 1983). The results of the study suggested that EMCs can be described by lognormal distributions. The statistical method- ology used in the NURP program entailed defining dilution ratios and calculating statistical properties of the resulting instream concentrations from the statistical properties of the flows and concentrations. Frequency of exceedance of any target concentration during wet weather was obtainable either through the use of formulas, standard plots of cumulative probability distributions, or calculations from statistical prop- erties of stream concentrations (Barrett et al., 1995).

An FHWA study presented by Driscoll et al. (1990) also applied statistical methods as the basis for the development of a procedure for predicting highway stormwater runoff pol- lutant loadings. The study also developed methods for esti- mating potential impacts on receiving waters, including guidance for evaluating the performance of mitigation mea- sures. A total of 993 individual storm events at 31 highway sites in 11 states were monitored. Barks (1996) discussed the development of statistical methods using site-specific data to adjust values obtained from the use of regional equa- tions so that more accurate values could be acquired. The regional regression equations were developed using data from a national database and are used to estimate runoff pollutant loads. The method consists of a single adjustment procedure: a regression of the observed data against the predicted values, a regression of the observed data against the predicted values and additional local independent variables, and a weighted combination of a local regression with regional prediction. Geochemical models are useful for evaluating the bioavailability and mobility of pollutants. Definitions of four categories of models provided by Bricker (1999) include the following: Speciation Models—Models used to calculate the parti- tioning of an element among different aqueous species and complexes. Examples of speciation models include WATEQF and WATEQ4F. Mass-Transfer Models—Models used to simulated changes in solution chemistry caused by mass-transfer processes. Examples of mass-transfer models are SOLMNEQ.88, MINEQL+, MINTEQ (4.00), and PHREEQC. Mass-Balance Models—Models used to simulate the net changes in the masses of reactants and products in waters along a flow path. An example of a mass-balance model is NETPATH. Geochemical Mass-Transport Models—Models used to simulate hydrodynamic advection and dispersion of dis- solved species in porous media as well as to speciate the aqueous solution and determine geochemical mass trans- fer. Examples of geochemical mass-transport models include CHMTRNS, PHREEQM-2d, and PHREEQC. The author also includes a discussion of the applications and limitations of each of the above categories. 3.4.7.1. Identification of Research Gaps and Needs As with BMP Modeling (section 3.2.10.), water quality modeling is heavily dependent on the availability of data. Therefore, there is a general need for accurate and represen- tative data for parameter estimation and model calibration and for stochastic models development. The NURP study resulted in the development of a large database of runoff characterization data; however, even with this large data set, 116 differences in runoff quality among different land use types could not be validated statistically. There is, though, still a need for forward-looking data collection efforts that focus more on modern parameters and less on parameters of dimin- ishing importance in highway runoff such as lead. There also is a need to develop hybrid models that take advantage of both stochastic and deterministic methods in order to pro- duce models that have the benefits of both statistical and simulation-based models. Adaptation of agricultural models for herbicide and pesticide modeling in the context of highway runoff management would provide insights into the transport processes of highway pesticides and herbicides. The most commonly modeled contaminants are heavy metals, nutri- ents, bacteria, dissolved oxygen, and solids. Existing models need to be extended and enhanced to simulate a wider range of contaminants. 3.4.7.2. Primary References Barks, C. S. Adjustment of Regional Regression Equations for Urban Storm-Runoff Quality Using At-Site Data. In Transporta- tion Research Record 1523, TRB, National Research Council, Washington, DC (1996) pp. 141–146. Barrett, M. E., Zuber, R. D., Collins III, E. R., Malina, J. F. Jr., Charbeneau, R. J., and G. H. Ward. A Review and Evaluation of Literature Pertaining to the Quantity and Control of Pollution from Highway Runoff and Construction. CRWR Online Report 95-5, (April 1995) 186 pp. Bricker, O. P. An Overview of the Factors Involved in Evaluating the Geochemical Effects of Highway Runoff on the Environment. Open-File Report 98-630, U.S. Geological Survey, Washington, DC (1999) 28 pp. Chui, T. W. D., Mar, B. W., and R. R. Horner. Pollutant Loading Model for Highway Runoff. ASCE Journal of Environmental Engineering, Vol. 108, No. EE6 (1982) pp. 1193–1210. DeVries, J. H., and T. V. Hromadka. Computer Models for Surface Water. In Handbook of Hydrology, New York, McGraw-Hill (1993) pp. 21–39. Driscoll, E. D., Shelley, P. E., and E. W. Strecker. Pollutant Load- ings and Impacts from Highway Stormwater Runoff—Volume I: Design Procedure. FHWA Report No. FHWA-RD-88-006, Fed- eral Highway Administration, Office of Research and Develop- ment (1990). Guay, J. R., and P. E. Smith. Simulation of Quantity and Quality of Storm Runoff for Urban Catchments in Fresno, California. Investigation Report 88-4125, U.S. Geological Survey, Water- Resources Investigations, Washington, DC (1988) 76 pp. Irish, L. B., Barrett, M. E., Malina, J. F., and R. J. Charbeneau. Use of Regression Models for Analyzing Highway Stormwater Loads. ASCE Journal of Environmental Engineering, Vol. 124, No. 10 (October 1998) pp. 987–993. Kerri, K. D., Racin, J. A., and R. B. Howell. Forecasting Pollutant Loads from Highway Runoff. In Transportation Research Report 1017, TRB, National Research Council, Washington, DC (1985) pp. 39–46. U.S. Environmental Protection Agency. Results of the Nationwide Urban Runoff Program—Volume 1. Final Report WH-554, Water Planning Division, Washington, DC (1983) 186 pp.

3.4.8. Highway Construction and Vegetation Maintenance Twelve state DOTs indicated that they have conducted stud- ies or prepared reports on the design or efficiency of storm- water management measures during construction, although 34 state DOTs have not conducted any studies in this area. Highway construction and maintenance activities have the potential to impact receiving water systems depending on numerous factors, such as land disturbance area, storm event timing, topographic and geological characteristics, and con- struction and maintenance BMPs. Construction activities that include grubbing, grading, and excavating may reduce slope stability and increase erosion, thereby increasing sediment loads and concentrations to receiving waters. Erosion and sed- iment controls were discussed in section 3.2.6, and impacts to receiving waters caused by sedimentation and turbidity are discussed in section 3.5.3. Vegetation maintenance activities such as roadside herbicide application have the potential to cause impacts to receiving streams. Several researchers have investigated conditions and activities that may contribute to poor runoff quality from highway construction and vegeta- tion maintenance sites. Common research questions posed include • How can suspended-sediment data be used to make ero- sion control and vegetation maintenance decisions at construction sites? • How does construction site runoff impact receiving waters? • How mobile are herbicides applied to highway shoulders? • How do roadside vegetation maintenance practices impact receiving streams? NCHRP Synthesis 20-5, Topic 33-04, is synthesizing road- side vegetation practices for erosion control and stormwater management, along with a variety of other purposes. This research will be available in early 2004. Barrett et al. (1995a) provides a thorough literature review of environmental effects of highway construction that includes more than 30 references of studies conducted in the 1970s and 1980s. A more recent investigation by Barrett et al. (1995b) involved monitoring the impacts to Danz Creek in south- western Travis County, Texas, during the construction of a new highway. Ten storms were monitored at sites upstream and downstream of the highway crossing. The results indi- cated that the concentration of suspended solids in Danz Creek increased at least fivefold during and immediately after storm events despite the presence of a system of tem- porary controls (primarily silt fences) and restrictions on the use of heavy equipment at the creek crossings. The only other monitored parameter that appeared to increase substantially was iron, due to high iron content in the site soils. Copper and zinc were shown to increase by 11% and 85%, respectively. Fifteen highway construction sites were monitored by Cal- trans to assess the water quality of stormwater runoff from 117 the sites (Kayhanian et al., 2001). Results obtained during the 2-year characterization study indicated that 1. Caltrans construction-site runoff constituent concentra- tions detected during this study were less than typical Caltrans and non-Caltrans highway runoff constituent concentrations, with the exception of total chromium, total nickel, total phosphorus, TSS, and turbidity. 2. The concentrations of TSS and turbidity likely are due to the disturbed soils present at most construction sites. 3. The origin of the high concentrations of total chromium, total nickel, and total phosphorus concentrations is unknown. Concentrations of these constituents varied between sites, so it is possible that site-specific soils and vegetative conditions contributed to the concentra- tions of these constituents. 4. A correlation (R-squared values greater than 0.5) was observed between TSS runoff concentrations and par- ticulate runoff concentrations of chromium, copper, and zinc, indicating that minimizing particulate matter may reduce total metals concentrations. In another Caltrans study (Caltrans, 2002), 120 storm events were monitored at 27 construction sites during four rainy sea- sons beginning in 1998–1999 and ending in 2001–2002. One of the primary purposes of the sampling study was to develop a baseline set of construction site stormwater quality con- centrations. Sites were selected to represent a wide range of typical Caltrans construction activities, geographic areas, and hydrometeorologic conditions, as well as other site-specific conditions. The results were reviewed to compare annual means of individual parameters for the four reporting years. Mean concentrations of total lead, nickel, and zinc varied over the 4-year period, while mean concentrations of total copper, cadmium, and arsenic were relatively consistent over the study period. All dissolved metals remained relatively consistent over the study period except for zinc, which had consistently higher concentrations during the later years. With the exception of TKN, nutrient concentrations were rel- atively consistent over the 4-year monitoring period, exclud- ing one abnormally high total phosphorus concentration in the second year. Measured hardness was relatively consistent over the 4-year monitoring period, while TSS concentrations were much higher during the second monitoring year com- pared to other monitoring years. Total and dissolved organic carbon concentrations were low compared to dissolved and suspended solids, suggesting that dissolved and suspended solids are composed primarily of inorganic particulate matter. Statistical comparison tests showed a statistically significant difference in measured runoff concentrations between new construction and modification facilities and existing facilities for dissolved copper, total coliform, dissolved lead, dissolved nickel, and dissolved zinc, with the concentration of each of these constituents being lower at new construction sites. Comparing water quality runoff from northern California

versus southern California sites, the statistical comparison test showed a significant difference for dissolved arsenic, dis- solved chromium, nitrate, nitrite, ammonia, TKN, dissolved lead, dissolved nickel, total nickel, TSS, TOC, and DOC, with the majority of these constituents showing higher concentra- tions in southern California. Statistical comparisons between seasons showed a significant difference for dissolved ortho- phosphate, nitrate, ammonia, oil and grease, diazinon, total coliform, dissolved zinc, TDS, TSS, pH, and specific conduc- tance for one or more seasons compared to other seasons. Still, no consistent pattern was observed. Construction site storm- water runoff data was compared to Caltrans highway runoff data. The statistical comparison showed significantly higher concentrations in highway runoff for total cadmium, dissolved copper, dissolved lead, total zinc, and dissolved zinc. TSS and hardness were significantly higher in construction site runoff than highway runoff, while oil and grease and COD were sig- nificantly higher in highway runoff. With regard to highway vegetation management a few stud- ies were found that investigated herbicide migration from highway rights-of-way. Powell et al. (1996) conducted a study in Glenn County, California, to assess the concentrations of simazine and diuron (herbicides often applied to highway rights-of-way) in runoff from the pavement shoulder at three highway sites during simulated rainfall events and at two sites during natural rainfall events. At the simulated rainfall sites, soil was sampled to a depth of 3 m at the site where no runoff occurred and to a depth of 1 m at the other sites. Herbicide was not found below a 0.3 m-depth at any of the three sites. Of the total 38 samples taken from the top 0.3 m of soil, 13 contained simazine (maximum concentration 694 µg/kg, found prior to herbicide application) and 17 contained diuron (maximum concentration 874 µg/kg, just after rainfall simulation). At one of the natural storm event sites, concentrations ranged from 29 to 337 µg/L simazine and from 46 to 2849 µg/L diuron. The largest amounts removed in any sampled period were 5.3% of the applied simazine and 8.4% of the diuron in one 28-hr period. At the other natural storm event site only simazine was applied. Samples were collected from a flume that discharged runoff into a drainage canal. The first runoff sample was taken after a total of 100 mm of rain had fallen, and simazine concentration averaged 105 µg/L in 52–66 m3 of runoff water collected. The greatest mass discharge in any sampled period was 155–200 m3 of runoff in 20 hr, with an average concen- tration of 83 µg/L simazine. In another study, Huang et al. (2002) investigated the trans- port of five different pesticides (glyphosate, oryzalin, isox- aben, transline, and diuron) in highway biofiltration strips at two geographically separated sites in northern California. Herbicides were detected in runoff water from both sites after all storm events. The EMC and loading percentage had large ranges among different herbicides at the two sampling loca- tions for the past three years (glyphosate: 0.1–21.5 µg/L, oryzalin: 0.1–42.4 µg/L, isoxaben: 0.1–14.4 µg/L, transline: 0.5–7.1 µg/L, diuron: 0.1–10.2 µg/L). Loadings as a percent- 118 age of the amount of pesticide applied also varied greatly (glyphosate: <1%, oryzalin: 0.1–5.4%, isoxaben: 0.1–15.0%, transline: 44%, and diuron: 0.6–4.4%). The high percent load- ing for transline undoubtedly was due to its relatively high solubility. The results of the study suggest that biofiltration strips along highway roadsides can significantly attenuate herbicides in runoff, particularly those with low solubility. A study by Wood (2001) investigated the potential of her- bicides applied to roadsides in the Willamette Valley near Colton, Oregon, to migrate to nearby surface waters. The study was divided into two phases. During the first phase (spring 1999), 0.3-inch-per-hour rainfall events were simulated and runoff was collected 1 day, 1 week, and 2 weeks after the application of an herbicide compound typical of Oregon DOT application rates and concentrations. The simulated rainfall was applied long enough to collect between 13 and 15 liters of runoff (between 0.5 and 1.9 hours). The EMC in the runoff of each of the herbicides (diuron, glyphosate, bromacil, and sulfometuron-methyl) declined by about 1.5 orders of mag- nitude between the first day after application and the second week after application. The results of the simulated rainfall experiments suggested that a heavy rainstorm occurring soon after herbicide application could generate concentrations in the runoff leaving the road’s shoulder of nearly 1 mg/L gly- phosate and diuron and concentrations on the order of a few hundred µg/L of sulfometuron-methyl. Bromacil was not measured in this phase. During the second phase (winter 1999–2000), concentrations were measured in the runoff occurring from natural rain events after a single herbicide application. Five rainfall events were chosen for the sampling. Runoff flowing directly from the shoulder remained in the 1–10 µg/L range for diuron for all events sampled with con- centrations decreasing with time. Based on the studies summarized above, it is clear that her- bicides have the potential to migrate to receiving waters. However, what is not clear is whether these herbicides pose a significant threat to receiving water biota. A study con- ducted by Johnson and Hall (2002) evaluated the impacts of Surflan™ (with the active ingredient oryzalin) on Japanese medaka (Oryzias latipes), a standard laboratory organism for testing impacts to fish. Results from three distinct assays sup- port the conclusion that Surflan and oryzalin are endocrine- disrupting compounds. But, this study was conducted at much higher concentrations than those observed by Huang et al. (2002). Since lowest-observed-adverse-effect-levels for repro- ductive effects of oryzalin and Surflan were not defined in this study, and there appears to be a nonlinear dose-response relationship, this study should be repeated at concentrations more typical of highway runoff concentrations. 3.4.8.1. Identification of Research Gaps and Needs The disturbance of land during highway construction activ- ities significantly increases the potential for sediment trans- port even with the implementation of erosion control prac-

tices. To evaluate the effectiveness of (or need for) erosion control practices, suspended sediment is the primary (and often the only) parameter monitored during highway con- struction runoff characterization studies. However, it often is not clear in the literature, particularly in abstract summaries, whether TSS or SSC are being reported, as these two terms often are used interchangeably. As discussed by Bent et al. (2001), these two measures of sediment concentration are not transferable because SSC is a measure of the total mass of sediment, while TSS is a measure of a subsample of the water- sediment mixture. Subsampling may inadvertently preclude larger-sized particles, resulting in an underrepresentation of the true sediment concentration. Due to the fact that TSS is reported and used often in the calculation of sediment loads, there appears to be a research gap in this area and a need to make stormwater practitioners aware of this potential issue. An equally important and related parameter that is not as frequently monitored as sediment concentration is the sedi- ment particle-size distribution. Particle size plays an impor- tant role in the transport and aquatic biota impacts of mobi- lized sediment. Particle size distribution also seems to play an important role in the transport of some metals, nutrients, and trace organics. Monitoring for particle size and these other parameters could increase significantly the costs of a construction project. It would be beneficial to have an initial screening method for assessing the quality of site soils on a grain-size basis to determine the level of monitoring as well as sediment and erosion controls necessary to prevent impacts to receiving waters. It is apparent from the literature reviewed that additional work is needed in the area of roadside vegetation manage- ment. The potential for herbicides to migrate from roadsides to receiving waters is strongly dependent on the type of chemical applied (i.e., depends primarily on solubility and hydrophobicity). Numerous herbicides are in use by DOTs throughout the country, of which only a small number have been tested for their mobility and potential toxicity to aquatic biota. Most studies have been conducted under highly controlled condi- tions in a laboratory or by using simulated rainfall. Further- more, toxicity studies have been conducted at higher con- centrations than likely to occur at the rates applied. More herbicide runoff characterization studies during storm condi- tions are needed in conjunction with toxicity studies at the concentrations found. Furthermore, an analysis of the adsorp- tion of herbicides to various grain sizes would assist in deter- mining the potential for migration. Once more information is available on potential impacts of herbicides, a detailed cost– benefit comparison of using herbicides as opposed to other vegetation control methods, such as manual clearing, should be considered. 3.4.8.2. Primary References Barrett, M. E., Zuber, R. D., Collins, E. R., Malina, J. F., Charbeneau, R. J., and G. H. Ward. A Review and Evaluation of Literature 119 Pertaining to the Quantity and Control of Pollution from High- way Runoff and Construction. Technical Report No. CRWR 95-5, Center for Research in Water Resources (1995a). Barrett, M. E., Malina, J. F. Jr., Charbeneau, R. J., and G. H. Ward. Water Quality and Quantity Impacts of Highway Construction and Operation: Summary and Conclusions. Technical Report CRWR-266, Center for Research in Water Resources (1995b). Bent, G. C., Gray, J. R., Smith, K. P., and G. D. Glysson. A Synop- sis of Technical Issues for Monitoring Sediment in Highway and Urban Runoff. Open File Report 00-497, U.S. Geological Survey in cooperation with the Federal Highway Administration (2001). Caltrans. Caltrans Construction Sites Runoff Characterization Study Monitoring Seasons 1998–2002. Report CTSW-RT-02-005, Stormwater Management Division (2002). Huang, X., Fischer, M., White, R., Lu, Y., and T. Young. Field Monitoring and Treatment of Herbicide Runoff from Highway Roadsides. Workshop on Storm Water Monitoring Techniques, California Department of Transportation (June 26, 2002). Johnson, M. L., and L. C. Hall. The Estrogenicity of Selected Her- bicides and Adjuvants Endocrine Disruption Capabilities of Sur- flan™ and Oryzalin. Caltrans Report, Division of Environmen- tal Analysis (2002). Kayhanian, M., Murphy, K., Regenmorter, L., and R. Haller. Char- acteristics of Stormwater Runoff from Highway Construction Sites in California. In Transportation Research Record: Journal of the Transportation Research Board, No. 1743, TRB, National Research Council, Washington, DC (2001) pp. 33–40. Powell, S., Neal, R., and J. Leyva. Runoff and Leaching of Simazine and Diuron Used on Highway Rights-of-Way. Environmental Hazards Assessment Report EH 96-03, California Environmen- tal Protection Agency, Department of Pesticide Regulations, Environmental Monitoring and Pest Management Branch (1996). Wood, T. M. Herbicide Use in the Management of Roadside Vege- tation, Western Oregon, 1999–2000: Effects on the Water Qual- ity of Nearby Streams. Investigation Report 01-4065, U.S. Geo- logical Survey, Water-Resources Investigation, Washington, DC (2001) 27 pp. 3.4.9. Drain Inlet and Gross Pollutant Studies The control of gross pollutants in highway runoff was addressed in section 3.2.2., Gross Pollutant Removal. This section focuses on the characterization of gross pollutants in highway runoff. Gross pollutants can be grouped loosely into three cate- gories: litter or trash, debris, and coarse sediments. Exam- ples of litter include unwanted anthropogenic waste materi- als such as paper, metal, glass, and plastic. Examples of debris include organic materials such as grass, leaves, and wood. Coarse sediments consist mainly of inorganic solids such as construction materials and soil particles (England and Rushton, 2003). Gross solids can cause odors, release pollutants, and become an unsightly mess; yet gross solids are monitored infrequently as are other pollutants in many of the studies available (England and Rushton, 2003). Improp- erly disposed gross solids can be carried by stormwater or wind to water bodies, causing environmental degradation (Sedrak et al., 2001). Factors that determine the mobility and persistence of gross solids include buoyancy, the ability to be

blown around by the wind, and degradability (Sedrak et al., 2001). Research questions and areas of interest with respect to gross solids management include • Sources of gross solids, • Gross solids monitoring measuring techniques, • Gross solids impacts to stormwater and receiving water systems, and • Gross solids modeling and estimation techniques. Several studies are available that define and characterize gross solids. Armitage and Rooseboom (2000) presented a discussion that defines urban litter, identifies sources of lit- ter, and suggests litter management strategies. Factors iden- tified as contributing to litter problems included antisocial behavior, excessive packaging, inadequate street sweeping services and disposal facilities, and lack of effective law enforcement. The authors noted that the rate of litter produc- tion is related to type of development, density of development or land use, income level of the community, types of indus- try, rainfall patterns, types of vegetation in the catchment, and the level of a community’s environmental awareness. The authors conclude by providing a discussion of litter load esti- mation with equations for evaluating litter quantities for design purposes. The results of a comprehensive study showed that plastics made up more than 40% (by weight) of the floatable litter found on the streets of New York City. Details of this study are provided by England et al. (2003) in addition to simple meth- ods for measuring and characterizing gross solids removed by various BMPs for both wet and dry weather. A study pre- sented by Sedrak et al. (2001) identified high trash generat- ing areas in Los Angeles and proposed both structural and nonstructural controls to manage trash. This study also found that plastics are the single largest component of trash. Trash enters receiving water bodies mainly by direct disposal by hikers or beachgoers, stormwater, and wind. The authors concluded that commercial, industrial, and residential land use areas produce the most trash. They suggested nonstruc- tural trash control measures including street sweeping, catch basin cleaning, antilittering statutes, abandoned trash hot- lines, trash cans, educational programs, and community clean- up programs. Structural trash or litter controls suggested include Continuous Deflective SeparationTM units, Netting Trash TrapTM, catch basin inserts, and catch basin opening covers. In a Caltrans study, Lippner et al. (2001) investigated the characteristics of litter in highway stormwater and evaluated the effectiveness of BMPs by conducting a 2-year pilot study in the Los Angeles area. The researchers found that plastic, paper, and Styrofoam constituted about 42% by weight and 57% by volume of total freeway litter. Securing a mesh bag on an outfall with VelcroTM worked well as a monitoring tech- nique, however the suggested monitoring technique is not rec- ommended for outfalls that directly connect to other subgrade 120 drainage systems. Two of the BMPs that the researchers tested increased litter pick-up, and the modified drain inlet substan- tially reduced litter. Street sweeping, the bicycle grate, and the Litter Inlet Deflector were ineffective in controlling litter. The importance of the effects of water velocity and depth in the transport of gross solids was investigated by Davies et al. (1998). The authors presented the results of a labora- tory study on solids advection with applications in solids trans- port modeling. Milne et al. (1996) collected and analyzed gross solids in an attempt to estimate the related impacts and interaction with sediment. They sampled wet and dry weather flows and monitored gully discharge. 3.4.9.1. Identification of Research Gaps and Needs Because of the variety of materials that make up gross pol- lutants in highway runoff, characterization and assessment is difficult. Areas that appear to be well covered in the litera- ture include the determination of sources of gross solids and the composition, characteristics, and transport of gross solids. Most researchers quantify gross pollutants by either weight or volume, and some segregate according to material type, such as plastic or metals. For the purposes of data transfer, there appears to be a need for the development of standard methods for quantifying gross pollutants. As mentioned in section 3.2.2, a potential research need may be to identify a uniform definition of gross solids (and its components) for purposes of standardizing the reporting of data. There also appears to be a need for more studies on receiving water impacts of gross solids, with a particular need for modeling and estimation techniques for gross solids especially in rela- tion to TMDLs. Another potential research gap may be the leaching or sorption capacity, or both, of pollutants captured in catch basins. For instance, cigarette butts, which can contribute as much as 10% by dry weight of all street litter (City of Los Angeles, 2001), contain several toxic and carcinogenic com- pounds that may leach to receiving waters during storm- water runoff events. On the other hand, bulk paper trash may aid in the sorption of oil and grease in stormwater runoff. 3.4.9.2. Primary References Armitage, N., and A. Rooseboom. The Removal of Urban Litter from Stormwater Conduits and Streams: Paper 1—The Quanti- ties Involved and Catchment Litter Management Options. Water SA, Vol. 26, No. 2 (April 2000) pp. 181–187. City of Los Angeles. High Trash-Generation Areas and Control Measures. Report, Department of Public Works, Bureau of San- itation (2001). Davies J. W., Sekuloski, V., and D. Butler. Inclusion of Gross Solids Advection and Deposition in Urban Drainage Models. Proc., 4th International Conference on Developments in Urban Drainage Modeling (September 1998) pp. 365–372.

England G., and B. Rushton. Monitoring Guidelines for Measuring Stormwater Gross Pollutants. Proc., STORMCON Conference, Session M-103, San Antonio, TX (July 2003). Lippner, G., Johnston, J., Combs, S., Walter, K., and D. Marx. Results of California Department of Transportation Litter Man- agement Pilot Study. California Water Environment Association (CWEA) 72nd Annual Conference, Sacramento (April 16–19, 2000). Milne, D. A., Jefferies C., and R. M. Ashley. Pollutional Aspects of Gross Solids and Their Interaction with Sewer Sediments. Water Science and Technology, Vol. 33, No. 9 (1996) pp. 31–37. Sedrak, M. F., Tam, W. K., Kaporis, K., and J. Pang. High Trash- Generation Areas and Control Measures. (August 2001) 36 pp. 3.4.10. Cold-Weather Studies Cold-weather highway runoff quality studies primarily characterize snowmelt runoff and evaluate winter mainte- nance activities such as highway sanding and deicing agent application. This field also encompasses studies that evalu- ate the effects of frozen soil on runoff and infiltration rates, as well as snowbank pollutant accumulation studies. Finally, a few studies have looked at the functioning of BMPs during cold weather. Compared to stormwater runoff, snow exposed to traffic and winter maintenance practices has a much greater capacity to accumulate and retain heavy metals, fine dusts, and other anthropogenic constituents. Traffic activities and winter storm management practices can have a significant impact on pollutant accretion in urban snow. Urban snow- packs accumulate large quantities of solids and contami- nants, which originate from such sources as airborne fallout, vehicular deposition, and applied and ground up grit and salt. Both contaminants and solids may be released quickly dur- ing the periods of snowmelt, and consequently melting con- taminated snow in urban areas in cold climates has the poten- tial to impact substantially the water quality of receiving waterbodies (Oberts, 2000; Smith et al., 2000). Cristina et al. (2001) and Sansalone and Glenn (2001), in their detailed studies involving physical and chemical characterization of snowmelt runoff, indicated that extended residence times of snow as a roadway snowbank exposed to traffic activities and winter maintenance practices lead to significant pollutant accretion and partitioning in the snow matrix. Several other studies discuss the accumulation of pollutants in the snow and eventual shock loading of pollutants during snowmelt or rain on snow events (Thorolfsson, 1999; Hatch et al., 1999). In another study, Sansalone and Buchberger (1997) pre- sented the effect of snowbank residence time on PSDs and particulate-bound metal element concentrations for two snow events. Results indicated that for each snow event, increas- ing residence time of the snowbank did not result in a clear increase in zinc, cadmium, or copper concentrations. Zinc concentrations on solids from rainfall events were signifi- cantly higher than for snow events. Snowbank lead concen- trations decreased over time for the finer fractions of solids for the first snow event with a similar trend for the second 121 snow event, except for the finest solid class, which showed a slight increase with time. In contrast to zinc, lead concentra- tions on rainfall runoff solids were generally lower than on snow solids. In their highway runoff monitoring study in Lake Tahoe, Caltrans (2002) characterized particles removed from the sand traps and filter boxes using the sieve and hydrometer method. Particle sizes ranged from less than 2 µm to more than 9500 µm, with the majority of particles falling in the range from 100 to 2000 µm. These mid-sized suspended particles are rela- tively large and over a relatively short period settle easily out of suspension because of gravity. Yet, the remaining colloidal (0.001 µm–1 µm) and smaller suspended particles tend to remain suspended in waters because of their low gravitational settling (less than 0.01 cm/sec) which could cause an increase in turbidity. It should be noted that in their study about the effectiveness of double barrel sand traps, removal of more than 90% of the total mass of TSS did not remove total phosphorus to the same degree (i.e., less than 20%). Particles associated with the snow ranged from 5000 µm to less than 25 µm and had a d50 of 1222 µm. Specific gravity ranged from 2.5 to 3.2 and tended to be lower for particles less than 100 µm. Metal analyses of the snow residuals indicated that 50% of the heavy metal mass of lead, copper, cadmium, and zinc was bound to particles greater than 250 µm. The treatment of snowmelt runoff is confounded by sev- eral factors including frozen conduits, ponds, soils and wet- lands, biological dormancy, and the addition of chemicals and grit to roadways (Oberts, 2000). Adaptation of com- monly used BMPs can be undertaken to accomplish some level of treatment, such as modifying outlet structures on detention ponds and using new subsurface “vault” treatment systems. Other measures include selective use of deicing chemicals (see section 3.5.7 for discussion on deicing impacts to receiving waters) and constructing road snow storage areas. However, there is still significant research needed in this area. The results of some of the more recent cold weather stud- ies, such as those discussed above, indicate the quality of snowmelt runoff from highways may be highly degraded and may be seriously impacting receiving streams. This realiza- tion, combined with the implementation of the NPDES Phase II stormwater regulations, is causing an increasing interest in nonpoint source control of cold climate runoff. In fact, a first- of-its-kind North American 3-day stormwater conference entitled Stormwater Management in Cold Climates: Planning, Design, and Implementation was held in Portland, Maine, in November 2003 to focus specifically on the challenge of man- aging stormwater in cold climates (http://www.cascobay. usm.maine.edu/coldsw.html). 3.4.10.1. Identification of Research Needs Based on the literature review, there is clearly a need for more monitoring and characterization of snowmelt runoff

from highways. The studies reviewed indicate that snowmelt runoff, especially during the first major snowmelt runoff events of the year, often have highly elevated pollutant con- centrations. Still, because of the hydrological complexity of snowmelt and freeze phenomena, it is difficult to monitor and characterize snowmelt runoff events. One alternative is to collect snow samples from roadsides, melt them, and then analyze them. This approach, however, is highly subjective and dependent on the age of the snow. It may overrepresent actual snowmelt concentrations, since it is likely that not all of the pollutants present in the roadside snow will become mobilized during melting periods. Based on these difficulties, there is an apparent need for guidance on monitoring roadside snow as well as snowmelt runoff. Developing models that can be used to predict the occurrence of a snowmelt runoff event would be helpful in determining when monitoring should take place. Also, the performance and feasibility of stormwater BMPs during cold weather need to be evaluated. Another issue faced by cold weather stormwater managers is the man- agement of removed snow from urban highways. 3.4.10.2. Primary References Caltrans. Tahoe Highway Runoff Characterization and Sand Trap Effectiveness Studies: 2001–2002 Monitoring Session. Report No. CTSW-RT-02-044, Stormwater Division (2002). Cristina, C., Tramonte, J., and J. J. Sansalone. A Granulometry- Based Selection Methodology for Separation of Traffic-Generated Particles in Urban Highway Snow-melt Runoff. Water, Air and Soil Pollution, Vol. 136, No. 1-4 (2001) pp. 33–53. Hatch, L. K., Reuter, J. E., and C. R. Goldman. Daily Phosphorus Variation in a Mountain Stream. Water Resources Research, Vol. 35, No. 12 (1999) pp. 3783–3791. Oberts, G. L., Marsalek, J., and M. Viklander. Review of Water Quality Impacts of Winter Operation of Urban Drainage. Water Quality Research Journal of Canada, Vol. 35, No. 4 (2002) pp. 781–808. Oberts, G. L. The Search for Effective Cold Climate BMPs. Proc., American Water Resource Association Spring Specialty Confer- ence, Water Resources in Extreme Environments (2000) pp. 147–153. Sansalone, J. J., and D. W. Glenn III. Accretion of Pollution in Roadway Snow Exposed to Urban Traffic and Winter Storm Maintenance Activities—Part I. ASCE Journal of Environmental Engineering, Vol. 128, No. 2 (2001) pp. 151–166. Sansalone, J. J., and S. G. Buchberger. Characterization of Solid and Metal Element Distributions in Urban Highway Stormwater. Water Science and Technology, Vol. 36, No. 7-8 (1997) pp. 155–160. Smith, D. W., Facey, R. M., Novotny, V., and D. A. Kuemmel. Management of Winter Diffuse Pollution from Urban Areas: Effect of Drainage and Deicing Operations. Cold Regions Impact on Civil Works (2000). Thorolfsson, S. T. New Strategies in Stormwater-Meltwater Man- agement in the City of Bergen, Norway. Water Science and Tech- nology, Vol. 39, No. 2 (1999) pp. 169–176. 122 3.5. RECEIVING WATERS IMPACT ASSESSMENT This category refers to studies conducted in receiving waters, including mixing zones. Thirty-five state transporta- tion agencies have performed some research on impacts on receiving waters. This section presents research that seeks to address the impacts of beneficial uses on receiving waters. According to Pitt et al. (2002) beneficial uses of receiving waters can be summarized as • Stormwater conveyance, • Biological uses, • Noncontact recreation, and • Contact recreation. Other beneficial uses include drinking water, domestic ani- mal drinking water, crop irrigation, and fisheries. Urbaniza- tion often leads to changes in the physical, chemical, and bio- logical characteristics of receiving waters. These changes often result in habitat that is significantly different from the habitat to which aquatic life is accustomed (May, 1998). Increased impervious area and degradation of water quality are traits that accompany urbanization. These traits can have negative impacts on stream morphology, in-stream habitat, wetlands, and aquatic biota. Transportation development contributes to that increase in impervious area, in addition to contaminants generated from highway construction, main- tenance, and usage. Such contaminants include deicers, met- als, petroleum-related organic compounds, sediment, and agricultural chemicals (Buckler and Granato, 1999). Since 1879, USGS has played a vital role in monitoring and assessing surface and ground water. USGS activities, studies, and programs provide support for decision making at all lev- els of government. According to Gail and Pixie (2002), USGS contributions to receiving waters impact research include • Monitoring more than 40 watersheds from 1980 to 1996 as part of nutrient transport studies in the Mississippi River Basin, • Conducting studies in San Francisco Bay for more than 26 years to assess impacts to aquatic biota in the context of environmental and meteorological changes, and • Pioneering studies on the impacts of MTBEs. USGS also has pioneered the use of several techniques use- ful for assessing receiving waters impact, including ground- water age dating, and maintains a large database containing chemical data from more than 335,000 waterbodies (Gail and Pixie, 2002). USGS and the Delaware River Basin Commis- sion, funded by the New Jersey DEP, conducted a study to investigate the impacts of urbanization of five watersheds in New Jersey. The objective of the study was to assess the cur- rent state of water quality, habitat, and stream morphology; develop and evaluate watershed assessment methods; and develop goals and objectives for the watersheds. The study,

summarized in a report by Albert and Limbeck (2000), reviewed the effects of urban runoff on stream channel sta- bility, water quality, aquatic organism habitat, and macro- invertebrates. The authors observed that percent impervious- ness is a good indicator of receiving water impacts and impervious area mitigation using BMPs may be pivotal to successful watershed management strategies. Percent of con- nected impervious surface is still more precise, when such information is available. 3.5.1.1. Primary References Albert, R. C., and R. L. Limbeck. High Flow Management Objec- tives for New Jersey Non-Coastal Waters: A Study of the Dela- ware, Saddle, Whippany, and Musconetcong Rivers and Flat Brook. Delaware River Basin Commission, West Trenton, NJ (2000) 34 pp. Buckler, D. R., and G. E. Granato. Assessing Biological Effects from Highway-Runoff Constituents. Open-File Report 99-240, U.S. Geological Survey, Washington, DC (1999) 45 pp. Gail, E. M., and A. H. Pixie. Monitoring and Assessing Our Nation’s Water Quality. Fact Sheet 076-02, U.S. Geological Sur- vey, Washington, DC (August 2002) 6 pp. May, C. W. The Cumulative Effects of Urbanization on Small Stream Watersheds in the Puget Sound Lowland Ecoregion. Proc., Water Environment Federation, Watershed Management—Moving from Theory to Implementation, Denver, CO (May 3–6, 1998). Pitt, R., Hoffman, D., Rattner, B., Burton, G. A. Jr., and J. Cairns, Jr. Receiving Water Impacts Associated with Urban Runoff. Handbook of Ecotoxicology, 2nd Edition, CRC-Lewis, Boca Raton, FL (2002). 3.5.2. Stream Crossings Stream crossings are especially vulnerable to pollution from roads. Contaminants such as sediment have easy access to the underlying streams at stream crossings at every stage in the lifecycle of a road. Most of the available studies on roadway impacts on stream crossings are related to unpaved roads and forest road impacts and bridge construction and maintenance impacts. Areas of interest cited by Taylor et al. (1999) include • Short- and long-term impacts of stream crossing instal- lations; • Data from varying stream sizes, soil types, terrain, and climatological conditions; • Development of standard measuring methods and con- tinuous automated sampling technologies; • Evaluation of proportions of contaminant contribu- tions from the stream crossing structure and the road approaches; and • Stream crossing impacts of stream ecology. Another potential research area is the effect of scour, sed- imentation, and turbidity on aquatic biota. Since this topic is 123 beyond the realm of stream crossings, it is more appropri- ately addressed in the next section. The discussion here is limited to the potential impacts of roadway runoff at stream crossing on aquatic biota. Data and analytical methods are available to predict the runoff constituents and concentrations for highway and water- way scenarios. NCHRP Projects 25-13 and 25-13(02) devel- oped guidelines on how to use these data and methods to make comprehensive assessments of the impacts of bridge runoff on receiving waters and a guide to assist practitioners in making decisions on the need for, and the extent of, con- trol of bridge-deck runoff in both new and retrofit applica- tions (Dubois, 2002a and 2002b). These projects integrated known technology applicable to the quality of runoff water, the background quality of the receiving water, and the water quality criteria applicable to the receiving water and addressed reasonable treatments and proper disposal systems if and when warranted. The guidebook encompassed consideration of runoff con- stituents (e.g., metals, sediments, and nutrients), types of bridge runoff-management designs, impacts on receiving waters and aquatic biota, and other potential runoff impacts. Also included in the guidebook were a risk assessment for special potential problems, benefit and cost-effectiveness assessments, and other elements of a strong management process to streamline and normalize consideration of runoff concerns within the project development process. Where warranted, the process addressed a range of mitigation alter- natives from on-site control of bridge deck runoff to off-site watershed-based mitigation and pollution trade-off opportu- nities. Where on-site control is proposed, appropriate new bridge design parameters for runoff and opportunities for existing bridge retrofits were considered along with non- structural BMPs. Both the design and construction of stream crossing struc- tures can impact receiving waters and roadside ecosystems. A study of aquatic communities at three bridge sites in Florida was performed by Birkitt and Dougherty (1984). An analy- sis of aquatic communities including dominance, diversity, and evenness values revealed adverse impacts to aquatic biota at one bridge site; the authors attributed the adverse impacts to bridge design practices. Adverse impacts at the second bridge site were attributed to construction practices. The third site showed only minimal impacts to aquatic biota. The authors recommended locating bridges at sites that require minimum alteration to river channels or the flood plain, using design principles and construction methods that strive to maintain existing hydrological, sedimentary, and illumina- tion characteristics of the river system and result in minimum site disturbance. A stream relocation and culvert installation project presented by Kober and Kehler (1987) found that incorporating mitigative measures into the project resulted in cost savings. Furthermore, post-construction biological con- ditions in the two streams used in the study were similar to or better than preconstruction conditions.

Research has shown the presence of heavy metals in bridge runoff. An analysis of runoff from the Skyway Bridge in Ontario, Canada, for five heavy metals (zinc, lead, nickel, copper, and cadmium) found EMCs for zinc, copper, and lead to be 0.337 mg/l, 0.136 mg/l, and 0.072 mg/l, respec- tively, in whole water samples. Mean PAH EMCs ranged from 0.015 µg/l to 0.5 µg/l. Sediment analysis revealed mean con- centrations of zinc, copper, and lead to be 997 µg/g, 314 µg/g, and 402 µg/g. This study by Marsalek et al. (1997) concluded that discharging bridge runoff directly into receiving waters without prior treatment could cause significant impacts to receiving water bodies. The preservatives used to treat wood bridges (or compo- nents) often are slowly released into the environment over time and could potentially end up in receiving waters. An evaluation of six bridges—two bridges treated with creosote, two bridges treated with pentachlorophenol, and another two treated with chromated copper arsenate—was performed by Brooks (2000). Study results indicated an absence of PAHs in water near any of the bridges. However, low levels of PAHs were detected in sediment directly underneath the bridges and immediately downstream. An analysis of aquatic invertebrate communities did not reveal any adverse effects and neither did laboratory bioassays conducted on water and sediment. According to the authors, robust invertebrate com- munities found in slow-moving streams were not susceptible to PAH levels that would be expected to impact more sensi- tive organisms in faster-moving streams. Dilution of con- taminants in the faster-moving streams was found to attenuate contaminant concentrations to levels that were not biologi- cally significant. The authors recommended that even though timber bridges pose little environmental risk, BMPs should be developed and deployed for all bridge types. A cooperative effort by Auburn University’s Biosystems Engineering Department and the U.S. Department of Agri- culture’s Forest Service Southern Research Station and its engineering research work unit in Auburn, Alabama, is under- way to fill some of the knowledge gaps pertinent to the impacts of stream crossings. The objectives of the project include • Quantifying and comparing water quality impacts from different types of stream crossings, • Quantifying the amount of sediment produced by road approaches at stream crossing sites, and • Documenting lifecycle costs of various types of stream crossings. Another ongoing research effort by the Kentucky Trans- portation Cabinet will evaluate potential receiving water impacts from lead and other heavy metal contaminants gen- erated as a result of pressure washing operations. The study also will examine existing paint on the selected bridges to assess the potential risk to receiving waters. The study will culminate in the development of alternative practices for wastewater treatment and disposal. Details of both this study 124 and the Auburn University study are available on the TRB RIP website. 3.5.2.1. Identification of Research Needs Receiving waters impacts at stream crossings include impacts that are well beyond the topic of highway runoff, such as impacts to stream channel morphology, sonic impacts to aquatic species during pile-driving activities, and impacts to fish passage through culverts. Although further research in these areas may be needed, the focus herein is on highway runoff, and the discussion has been limited to impacts caused directly by stormwater runoff or by runoff generated during maintenance activities such as bridge deck cleaning. Other bridge maintenance activities such as painting, surface treat- ments, substructure repair, joint repair, drainage structures repair, and pavement repair or repaving also may cause impacts to receiving waters depending on storm event tim- ing, duration, and intensity. With regard to highway runoff, the potential impacts to receiving waters at stream crossings appear to have been assessed by only a limited number of studies. NCHRP Project 25-13 is likely the most extensive assessment to date on this topic. The recommended research topics suggest (1) examining the water quality effects of maintenance practices through field studies, (2) developing a bridge deck runoff quality con- stituents database, (3) applying laboratory bioassays appropri- ate for stormwater discharges and field biosurveys, (4) exam- ining the potential risks associated with hazardous material spills, and (5) identifying mitigation practices for controlling bridge runoff quality. Effects of bridge design and ADT on runoff quality present another research need. 3.5.2.2. Primary References Birkitt, B. F., and B. J. Dougherty. Effects of Highway Bridges on the Aquatic Biota of Three Florida Rivers. In Transportation Research Record 969, TRB, National Research Council, Washington, DC (1984) pp. 1–7. Brooks, K. M. Assessment of the Environmental Effects Associ- ated with Wooden Bridges Preserved with Creosote, Pentachlor- ophenol, or Chromated Copper Arsenate. Research Paper FPL– RP–587, U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI (2000) 100 pp. Dupuis, T. V. NCHRP Report 474: Assessing the Impacts of Bridge Deck Runoff Contaminants in Receiving Waters—Volume 1: Final Report, TRB, National Research Council, Washington, DC (2002a) 77 pp. Dupuis, T. V. NCHRP Report 474: Assessing the Impacts of Bridge Deck Runoff Contaminants in Receiving Waters—Volume 2: Practitioner’s Handbook. TRB, National Research Council, Washington, DC (2002b) 101 pp. Kober, W. W., and S. E. Kehler. An Analysis of Design Features in Mitigating Highway Construction Impacts on Streams. In Transportation Research Record 1127. TRB, National Research Council, Washington, DC (1987) pp. 50–60.

Marsalek, J., Brownlee, B., Mayer, T., Lawal, S., and G. A. Larkin. Heavy Metals and PAHs in Stormwater Runoff from the Skyway Bridge, Burlington, Ontario. Water Quality Research Journal of Canada, Vol. 32, No. 4 (1997) pp. 815–827. Taylor, S. E., Rummer, R. B., Yoo, K. H., Welch, R. A., and J. D. Thompson. What We Know and Don’t Know About Water Qual- ity Impacts at Forest Road Stream Crossings. Journal of Forestry, Vol. 97, No. 8 (1999) pp.12–17. 3.5.3. Sedimentation and Turbidity Increases in flow rates to receiving streams caused by increased impervious areas, in turn, increase the potential for streams to scour, particularly near outfalls without sufficient energy dissipation controls. Scour can cause increased down- stream turbidity. Poor erosion controls near surface waters, particularly during and shortly after grubbing and grading activities, can cause sedimentation and high turbidity in receiving waters. Methods to reduce erosion and turbidity were discussed in section 3.2.6. This section addresses the impact of sedimentation and turbidity on receiving waters. Research by Bash et al. (2001) evaluated the effects of sed- iments and turbidity on salmonids in Washington State. A lit- erature review found that excessive sediment in hatchery water may smother eggs by depriving them of oxygen and by reducing the ability of juveniles to capture prey. The litera- ture review also suggested that gill injuries increase as angu- larity and particle size of suspended solids increase. The authors concluded that a better understanding of sediment size, shape, and composition, as well as a better understand- ing of salmonid species and life history stages, cumulative and synergistic stressor effects, and overall habitat complex- ity and availability in a watershed is required. They also rec- ommend that for short-term construction projects, operators must measure background turbidities on a case-by-case basis to determine if they are exceeding regulations. Turbidity standards developed by several states and provinces in the region attempt to consider natural variability in turbidity by requiring the regulated community to measure “background turbidity” upstream of any proposed activity. Although, since the background turbidity measured in these situations repre- sents a measurement at one point in time, regulating turbid- ity levels based on this type of measurement may not protect salmonid health. 3.5.3.1. Identification of Research Needs With regard to sediment and turbidity impacts to fish in general and salmonids in particular, significant research needs identified by Bash et al. (2001) included (1) develop- ing new exposure metrics that account for sublethal effects (as opposed to direct mortality); (2) examining the effect of frequent short-term pulses of suspended sediment; (3) con- ducting additional research on correlations between particle size, shape, and composition of sediments to fish sensitivity; 125 (4) studying relationships between seasonal timing and effect of sediment load; and (5) determining whether knowledge of survival responses to turbid flows can be used to develop mixing zones, work windows, treatment systems, and buffers that will allow fish to perform their necessary life functions during project construction and operation. 3.5.3.2. Primary Reference Bash, J., Berman, C., and S. Bolton. Effects of Turbidity and Sus- pended Solids on Salmonids. Final Research Report No. T1803- Task 42, Effects of Turbidity on Salmon, Washington State Tran- spiration Commission (2001). 3.5.4. Toxicity and Bioassessment Toxicity testing and bioassessments can be used to char- acterize and assess the cumulative/synergistic impacts of stormwater pollutants on receiving waters and sediments. Bioassessment includes evaluating indicators of receiving water health, such as biomass and species diversity. Toxicity testing requires evaluation of a test species’ survivability in a water sample or sediment sample. The testing is included fre- quently in water quality management studies, because it can provide an indication of the potential impact of discharged contaminants on receiving waters and associated biota. Regulators are attempting to increase regulatory control of toxic contaminants relative to constituents that have low or zero toxicity. Knowing the potential toxicity of highway run- off is important, because if toxicity is high, one can expect greater regulatory control and the implementation of treat- ment programs to reduce the toxic pollutants. Alternatively, if runoff is demonstrated to be nontoxic, controls subse- quently will be reduced. Potential research questions include • What is the applicability and limitation of the various toxicity testing methods with regard to the assessment of receiving water impacts? • How can biomass and species diversity be used to eval- uate the health of receiving waters? As mentioned in section 3.2.12, information describing toxicity specific to road runoff in the open literature is scarce. Few toxicity studies have been conducted where the runoff was predominantly or exclusively from roadways or high- ways. One multiyear toxicity study was conducted in the Santa Clara Valley, California; samples were collected pre- dominantly from freeway runoff. These samples showed high incidences of toxicity to C. dubia (freshwater crustacean), but the toxic response was quantitatively different from that seen in samples deriving from other land use categories (BASMAA, 1996). The cause of toxicity for highway runoff in the BASMAA study was found to be nonpolar organics and metallo-organics. Two other highway runoff toxicity

studies reached similar conclusions on measuring a higher level of toxicity in highway runoff compared to the other land uses [Pitt et al. (1991) and Marsalek et al. (1999)]. The high level of toxicity in these runoff samples may have been due, in part, to the presence of deicing chemicals or to higher concentrations of bioavailable heavy metals. In addition to water toxicity testing, some researchers have evaluated the toxicity of sediments near urban stormwater outfalls to assess the effects of these discharges. Rochfort et al. (2000) assessed relationships among three separate aspects of the benthic environment: sediment chemistry (met- als, PAHs, and nutrients) and particle size, sediment toxicity (ten endpoints with four benthic taxa), and benthic inverte- brate community structure. Researchers found that while con- taminant (metals and PAHs) concentrations were relatively high in sediments, biological effects were not evident (i.e., toxicity of sediments was low, benthic communities appeared unaltered, and neither toxicity endpoints nor benthic com- munity descriptors could be related to sediment contaminant levels). In a similar investigation, the chemical characteristics of urban stormwater sediments in the rapidly growing Phoenix metropolitan area of Maricopa County, Arizona, were ana- lyzed (Parker et al., 2000). Results showed that the inorganic component of the sediments generally reflected geologic background values, but some metals concentrations (e.g., cadmium, copper, lead, and zinc) were above background values, indicating an anthropogenic contribution of these ele- ments. Organochlorine compounds and PCBs were ubiqui- tous in the sediment samples, even though many of these compounds have been banned from general use for as long as three decades. Sediment toxicity results seem to suggest that surficial sediments from stormwater-control basins, city streets, vacant lots, and unpaved parking areas are a significant environmental problem, but the temporal and spatial variabil- ity in the test results makes such a conclusion tentative. With respect to bioassays for the evaluation of potential toxic effects of highway runoff, Dubois (2002) noted that the toxicity test methods should be modified to account for the episodic nature of runoff; hence, test organisms for bioassays should be exposed to runoff for a length of time equal to the storm event length. Limitations of traditional toxicity testing methods also are discussed by Burton et al. (2000). Such time-variable bioassays were performed in the study for run- off from two distinct bridges. The I-85 bridge in North Car- olina, which crosses a small stream, had a medium-level of ADT—74,000 vehicles at the time of the study. The San Francisco–Oakland Bay Bridge, which crosses the San Fran- cisco Bay, had a high ADT of 274,000 vehicles at the time of the study. No toxicity was found in time-variable bioassays for I-85 runoff. Some toxicity was found in traditional chronic 7-day bioassays with 100% runoff (did not reflect runoff event duration). This demonstrates the importance of using the time-variable technique (described in this report) to assess 126 accurately potential toxicity. There was some toxicity with 100% runoff from the San Francisco–Oakland Bay Bridge using the time-variable technique. There was significant tox- icity with 100% runoff using the traditional 7-day chronic test (did not reflect runoff event length). Recently, Caltrans initiated a comprehensive toxicity study of runoff from their various facilities, including 24 highway sites on a statewide basis (Caltrans, 2002). The goal of this Statewide Toxicity Testing Research Project was to assess the toxicity associated with discharges from its storm drain system, determine the cause of the toxicity, and provide some understanding of the sources of these discharges. In most cases, a single discrete sample was obtained from various facilities and tested for toxicity based on the EPA’s standard three species test. These single discrete samples were col- lected at different points of the hydrograph with the majority being collected at the beginning of the storm events. Storm- water was captured by grab samples and shipped to the Aquatic Toxicology Laboratory at the University of Califor- nia, Davis. The results obtained for the past two monitoring seasons (2000–2001 and 2001–2002) are summarized below (Caltrans, 2002). • Pimephales—Of the 98 tests performed, 82 (83.7%) indicated significant toxicity for either survival or growth. Significant reductions in biomass were found in 52 samples, and significant mortality was found in 28 samples, indicating that most often, reductions in biomass were common, and acute toxicity was less common. No pattern in toxicity with respect to date of sampling was apparent, as significant toxicity was found at all dates from October to May. • Ceriodaphnia—Of the 98 tests performed, 72 (73.5%) indicated significant toxicity. These results included all tests for which acute toxicity occurred and chronic tests were not possible to perform. As with the Pimephales’ toxicity test results, there appeared to be no pattern with respect to date of sampling, as significant toxicity was found throughout the entire period of sampling. • Selenastrum—Of the 98 tests performed, 46 (46.9%) indicated significant toxicity. The Selenastrum test was never the sole positive test result for any site at any sam- ple date. Again, no patterns were evident in the positive results. • Toxicity Identification Evaluations (TIEs)—Thirty TIEs were performed on samples in which acute toxicity was observed. The TIEs indicated that no single source of toxicity was common among sites. However, nonpolar organic compounds were suggested as the putative source of toxicity in 5 of the TIEs, metals were suggested as the putative source in 11 TIEs, and surfactants were suggested as the putative cause in 7 cases. In one case, a metabolically active pesticide was implicated. The remainder had no discernable cause.

Overall, more than two-thirds of the discrete samples col- lected were found to be toxic. This method of sample collec- tion and toxicity testing may produce misleading results, as the single sample is not representative of the entire event. More importantly, the relative toxicity of samples from the beginning of the event (first flush) compared with the rest of the event will not be known. Toxicity measurement on a hydrographic scale is more appropriate as it would provide information describing the variability of toxicity on a time and flow scale. An investigation of the relationship of toxicity to flow and time was initiated as part of the Caltrans first flush study dur- ing the 2002–2003 monitoring season (Caltrans, 2003). Only three storm events were monitored during the 2002–2003 wet season. The results indicate the presence of a toxic first flush at some sites, which maybe useful for BMP selection. How- ever, results are insufficient to make conclusions regarding the cause of toxicity and the influence of site-specific or storm-specific factors. Caltrans would like to continue the first flush toxicity study for two more seasons to address hydrographic toxicity of highway runoff for BMP selection, but the availability of future funding is unsure. Nábe˘lková et al. (2002) studied two tributaries of the Vltava River in Prague in an effort to evaluate the ecological risk of pollutants. They analyzed potential impacts of individual pol- lutants to aquatic organisms with the aid of mathematical sim- ulations. For each pollutant they developed an ecological risk description in toxicological units. The researchers found that heavy metals did not pose an ecological risk in surface water, but chronic heavy metal loads were found in bottom sedi- ment, which did pose an ecological risk. Environmental indicators developed by the Center for Watershed Protection (CWP) can indicate the extent of impacts to receiving water and the effectiveness of storm- water management programs. The Santa Clara Valley Urban Runoff Pollution Prevention Program implemented and tested 20 of the CWP’s 26 Environmental Indicators to Assess Stormwater Programs and Practices (Cloak and Bicknell, 2001). The researchers found that the CWP indicators were useful for tracking and enhancing pollution prevention efforts and also for holistic evaluations of stream function for the purposes of watershed management and planning. Bioindicators have been used in studies to demonstrate impacts to the environment. Lemly and King (2000) used the occurrence of bacterial growth on aquatic insects as an indi- cator of nutrient impacts on wetlands. During field investiga- tions, nitrate and phosphate levels were linked to the growth of filamentous bacteria on insects. The authors concluded that the use of the insect bacteria bioindicator is a reliable metric for evaluating nutrients impacts on wetlands where Wetland Bioassessment Protocols were applicable. Biologi- cal assessment methods can be used to assess wetland con- dition, evaluate the performance of wetland protection and restoration activities, and track water quality conditions in wetlands (Danielson, 1998). 127 3.5.4.1. Identification of Research Needs As mentioned above in section 3.2.12, the top two research needs identified by GKY and Associates in the original NCHRP Project 25-20 report were (1) to identify and develop regional aquatic biological indicators to assess impacts of highway runoff and (2) to conduct research methods for assessing the toxicity of highway runoff. This recent litera- ture review effort supports the claim that there do not appear to be adequate bioassessment methods for assessing impacts of highway runoff on receiving water systems, particularly at the time-scales typical of stormwater-runoff events. Also, there are a wide variety of assessment methods currently used by the few highway water quality researchers conduct- ing toxicity and bioassessment studies, so it is difficult to quantitatively compare existing data or to make any general assessment of the impacts of highway runoff on receiving water biota. In addition, more within-storm toxicity testing needs to be conducted to ascertain what parts of storm events are most toxic. A comparison of drainage systems (e.g., veg- etated versus piped conveyance) with regard to toxicity is also a potential research gap. 3.5.4.2. Primary References BASMAA. Guidance for Monitoring the Effectiveness of Stormwater Treatment Best Management Practices. Report, EOA, Inc. (1996). Burton, G. A., Pitt, R., and S. Clark. The Role of Traditional and Novel Toxicity Test Methods in Assessing Stormwater and Sed- iment Contamination. Critical Reviews in Environmental Science & Technology, Vol. 30, No. 4 (2000) pp. 413–447. Caltrans. Storm Water Monitoring & Data Management: 2001–2002 Annual Data Summary Report. Report No. CTSW-RT-02-048 (2002). Cloak, D., and J. C. Bicknell. Use of Environmental Indicators for Assessing Stormwater Program Effectiveness (2001) 11 pp. Danielson, T. J. Wetland Bioassessment Fact Sheets. Report No. EPA 843-F-98-001, U.S. Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds, Wetlands Division, Washington, DC (1998). Dupuis, T. V. NCHRP Report 474: Assessing the Impacts of Bridge Deck Runoff Contaminants in Receiving Waters—Volume 1: Final Report. TRB, National Research Council, Washington, DC (2002) 77 pp. Lemly, A. D., and R. S. King. An Insect-Bacteria Bioindicator for Assessing Detrimental Nutrient Enrichment in Wetlands. Wet- lands, Vol. 20, No. 1 (March 2000) pp. 91–100. Marsalek, J., Rochfort, Q., Brownlee, B., Mayer, T., and M. Servos. Exploratory Study of Urban Runoff Toxicity. Water Science and Technology, Vol. 39, No. 12 (1999) pp. 33–39. Nábe˘lková, J., St’astná, G., Komínková, D., and Z. Handová. Eco- logical Risks Assessment of Small Urban Streams. Proc., 9th International Conference on Urban Drainage (2002) 10 pp. Parker, J. T. C., Fossum, K. D., and T. L. Ingersoll. Chemical Char- acteristics of Urban Stormwater Sediments and Implications for Environmental Management, Maricopa County, Arizona. Envi- ronmental Management, Vol. 26, No. 1 (2000) pp. 99–115.

Pitt, R. E., Lalor, M., Field, R., and M. Brown. Investigation of Source Area Controls for the Treatment of Urban Stormwater Toxicants. Water Science and Technology, Vol. 28, No. 3-5 (1991) pp. 271–282. Rochfort, Q., Grapentine, L., Marsalek, J., Brownlee, B., Reynold- son, T., Thompson, S., Milani, D., and C. Logan. Using Benthic Assessment Techniques to Determine Combined Sewer Overflow and Stormwater Impacts in the Aquatic Ecosystem. Water Quality Research Journal of Canada, Vol. 35, No. 3 (2000) pp. 365–397. 3.5.5. Modeling of Water Quality Impacts Modeling has become an important part of stormwater management. Nonetheless, modeling development efforts have focused mostly on surface runoff models, and little attention has been devoted to developing sophisticated groundwater models. James and Ulan (1997) present dis- cussions about the utilization of shallow ground routing routines in SWMM4.3 and HSPF to model infiltration BMPs. Beckers and Frind (2000) developed a model adapted to simulate situations where groundwater recharge may be impacted significantly by heterogeneity above the water table. The model accepts precipitation and evapo-transpiration as direct inputs. Hvitved-Jacobsen et al. (1996) simulated oxygen deple- tion from 35 years of rain events using a modified oxygen sag theory. Balmforth et al. (2002) discuss the Leeds Urban Pol- lution Management Study, which modeled an area with more than 500,000 people and 130 inadequate combined sewer overflow (CSO) systems. The model was used to gen- erate BOD, ammonia, and suspended solids loads, which then were compared to water quality standards to determine the contribution of individual discharges to the failure of standards. The researchers observed that modeling and data collection costs can be reduced through careful management and through the application of the bespoke model. Also, a detailed model allows simulation of vital processes such as first flush. Temperature is a significant water quality pa- rameter for cold-water aquatic habitats. Haq and James (2002) present a thermal enrichment model for Portage Creek, a cold-water habitat for fish located in Portage, Michigan. Using 11⁄2 years’ of continuous tempera- ture data, they created a model to simulate the heat budget for pavement runoff. The researchers concluded that pavement runoff impacts stream temperature. Temperature change associated with runoff from paved areas has been documented, as has the effect of detention basins on receiving water temperature conditions. However, temperature often is overlooked as a physical characteristic of receiving waters. It is possible to model temperature effects of urban runoff, but when temperature is related to chemical or biological processes in receiving waters a number of issues are unresolved; available models are often site-specific or lim- ited in scope (e.g., addressing only summer season issues). Nevertheless half of state DOTs ranked this as a low-priority research area. Only 10% of the DOTs ranked it as a high pri- 128 ority; in most of these cases, impacts on cold-water fish species are a driving factor. Bioavailability of metals in sediments is linked directly to pore-water metal activity, which is influenced by physi- cal, chemical, and biological processes. Wood and Shelly (1999) developed a system dynamics model to represent these processes and the major influences affecting pore water metal activity in a treatment wetland receiving stormwater influent. The model structure and behavior were tested and validated using several system dynamics validation tech- niques. The model was run using metal specific parameter values typical of metals commonly found in stormwater run- off. Simulation results demonstrated that chemical processes of acid volatile sulfide and organic carbon in binding metal in reduced sediments are the greatest influences in control- ling metal bioavailability. As represented in the model, the effect of bioturbation was negligible. The amount of organic carbon in the sediment plays the most substantial role in con- trolling metal bioavailability in the long run. Using 5 years’ of highway runoff characterization data, including 500 storms at nine locations in Washington State, Horner and Mar (1983) developed a model that relates high- way segment length to cumulative pollutant loadings. The model incorporates the effects of high traffic density and the mitigative effects of draining highways through roadside swales. The model can perform three levels of analysis rang- ing from detailed analysis to a simple screening method. 3.5.5.1. Identification of Research Needs The area of water quality modeling shares a few of the research gaps identified under the BMP Modeling section (sec- tion 3.2.10) of this effort. These research gaps include the availability of data for accurate and representative parameter estimation, the ability to accurately measure and analyze unit processes, model calibration, and the need for an expert sys- tem for model evaluation and selection. Other potential knowl- edge gaps pertinent to water quality modeling include guid- ance on modeling temperature change impacts from pavement runoff, further development and enhancement of stochastic water quality models, evaluation of the limitations imposed by snow on water quality modeling methodologies, and the development of solutions for more accurate simulation of the effects of snow in water quality models. 3.5.5.2. Primary References Balmforth, D., Barker, C., and P. Myerscough. Integrated Model- ing of CSO Impacts in Large Urban Areas. Global Solutions for Urban Drainage, Proc., 9th International Conference on Urban Drainage (2002) 15 pp. Beckers, J., and E. O. Frind. Simulating Groundwater Flow and Runoff for the Oro Moraine Aquifer System: Part I—Model For-

mulation and Conceptual Analysis. Journal of Hydrology, Vol. 229, No. 3 (2000) pp. 265–280. Haq, R., and W. James. Thermal Enrichment of Stream Tempera- ture by Urban Stormwater. Proc., 9th International Conference on Urban Drainage (2002) 11 pp. Horner, R. R., and B. W. Mar. Guide for Assessing Water-Quality Impacts of Highway Operations and Maintenance. In Trans- portation Research Record 948, TRB, National Research Coun- cil, Washington, DC (1983) pp. 31–40. Hvitved-Jacobsen, T., Portielje, R., and J. Schaarup-Jensen. Sto- chastic Reliability Methods in Modelling Effects of Urban Pol- lution Runoff. Proc., 7th International Conference on Urban Storm Drainage (1996) pp. 1491–1496. James, W., and J. A. Ulan. Towards a Shallow Groundwater Rou- tine for Modeling Infiltration BMPs in Urban Stormwater Mod- els. Advances in Modeling the Management of Stormwater Impacts, Vol 6 (February 1997). Wood, T. S., and M. L. Shelley. A Dynamic Model of Bioavail- ability of Metals in Constructed Wetland Sediments. Ecological Engineering, Vol. 12, No. 3-4 (1999) pp. 231–252. 3.5.6. Water Quality Impacts of Combined Sewer Overflows CSOs are drainage systems that discharge excess untreated sewage and stormwater directly into marine waters, lakes, rivers, and other water bodies when the system capacity is reached. In most cases CSOs are legacy systems built in the past and left unchanged as a result of cost constraints. In the old days, CSO systems were not considered as a major source of pollutants; it was assumed that by the time sewage systems overflowed, the majority of the contaminants would have been flushed already. Also, the increased volume of the receiving water body was assumed to provide adequate dilution. These assumptions are hard to verify (Villeneuve and Lavallee, 1985). Since the advent of wastewater treatment plants in the 1950s, CSO impacts to stormwater have been subject to increasing scrutiny. In some municipalities, interceptor pipes have been built to convey all wastewater, including dis- charges from CSOs, to wastewater treatment plants. How- ever, in many older cities, CSOs still remain an issue. Accord- ing to Seidl et al. (1996), the main impacts of CSOs include bacteria loads and increased consumption of oxygen due to organic matter. The water quality impacts of CSOs need to be better understood in order to facilitate development of appropriate regulations for the protection of receiving waters. According to Kaunelis and Johnson (2000), areas of interest and research questions concerning CSOs include • Is CSO discharge sufficiently clean to meet water qual- ity standards? • What impacts should be measured? • How can CSO impacts be isolated and measured inde- pendently of other impacts to receiving waters? • How much data is needed to support CSO decision making? 129 A number of research projects have investigated the impact of CSOs on receiving waters. Many more monitoring proj- ects are still in progress. Kaunelis and Johnson (2000) dis- cuss an ongoing evaluation of nine facilities built by the Rouge River National Wet Weather Demonstration Project to store and treat CSO effluent in metropolitan Detroit. The results of this effort will determine if more capital investment is needed to mitigate CSO impacts. An investigation by Hvitved-Jacobsen and Harremoes (1981) found that CSO impacts on dissolved oxygen occur in two phases. The imme- diate phase is oxygen depletion attributed to the soluble frac- tion of organic matter in the discharge. The delayed phase is potentially more serious and attributed to “adsorption of sol- uble, colloidal and fine particulate fractions.” Widera and Podraza (1996) describe chemical analyses as “spot checks” that show water quality at a definite time while stream biota analyses reveal long-term effects. Several studies provide insight into CSO effluent composi- tion. The Rouge River National Wet Weather Demonstration Project monitored CSOs for 2 years. Kaunelis and Johnson (2000) summarized the methodology and the results of this study. Monitored pollutants included carbonaceous biological oxygen demand (CBOD), TSS, ammonia, and total phospho- rus. EMCs for four basins were as follows: CBOD, 4.5–43.2 mg/l; TSS, 24–82.7 mg/l; ammonia, 0.14–4.47 mg/l; and total phosphorus, 0.58–1.26 mg/l. The study concluded that the impacts of CSOs can be isolated from other sources of pol- lution to quantify the effectiveness of CSO mitigation mea- sures. Two years’ of monitoring data and/or 10 overflow events can provide adequate data to support CSO-related decision making. Data from a 5-year study of the Cumberland River in Nashville, Tennessee, showed that dissolved oxygen depletion was not an issue with the CSO system. This study by Thackston and Murr (1999) also ruled out the CSO sys- tem as the source of the fecal coliform bacteria problem in the river. Data from the study saved $106,000,000 in planned redundant improvements to the drainage system. Seidl et al. (1996) presented a discussion on monitoring data collected as input to run the model Prose and monitored six rainfall events and parameters including conductivity, tur- bidity, TSS, COD, ammonia, DOC, BOD5, and bacteria. Researchers found similar ratios between the parameters under various conditions and concluded that high DOC may origi- nate from urban surface deposits or resuspension of sewer deposits. They observed a decrease of DOC for the big rainfall events; they attributed the decrease to dilution. Fluctuations in bacterial levels made observed bacteria biomass level results less conclusive and harder to correlate to other parameters. A number of studies focus on creating new CSO pollutant models or modifying existing models. Hvitved-Jacobsen and Schaarup-Jensen (1990) discussed the application of a dis- solved oxygen stream simulation model. O’Connor et al. (1993) presented a discussion of the use of EPA’s SWMM to model pollutants in a CSO abatement study of Newton Creek in New York. Villeneuve and Lavallee (1985) presented

methodologies to characterize CSO wastewater and to define lateral and longitudinal diffusion of wastewater. The paper also discussed mitigation of intermittent CSO discharges. Michelbach et al. (1999) presented a method for estimating nutrient loads to Lake Constance in Europe from CSOs. The authors developed new functions to calculate nutrient loads from average nutrient concentrations, annual overflow rate, and solids transport in sewers. 3.5.6.1. Identification of Research Needs CSO systems are widely variable, and water quality impacts depend on a host of site-specific parameters. Measurement of impacts is based mostly on computer simulations. There is a need for better monitoring of CSO effluent quality in relation to meteorological factors. What prevailing conditions or fac- tors increase or decrease CSO impacts? What methods can be used to mitigate impacts (structural and nonstructural)? 3.5.6.2. Primary References Hvitved-Jacobsen, T., and K. Schaarup-Jensen. Analysis of CSO Impacts on the Dissolved Oxygen Concentration of Receiving Streams. Proc., 5th International Conference on Urban Storm Drainage (June 1990) pp. 517–522. Hvitved-Jacobsen, T., and P. Harremoes. Impact of Combined Sewer Overflow on Dissolved Oxygen in Receiving Streams. Proc., 2nd International Conference on Urban Storm Drainage (June 1981) pp. 226–235. Kaunelis, V. P., and C. R. Johnson. Evaluation of In-Stream Impacts of CSO Control Facilities. Proc., Watershed 2000 Con- ference, Vancouver, BC (2000) 20 pp. Michelbach S., Weib, G., and H. Brombach. Nutrient Impact from CSOs on Lake Constance. Proc., 8th International Conference on Urban Storm Drainage (August 1999) pp. 474–481. O’Connor, A., Schuepfer, Z., and L. Kloman. Hydraulic and Pollu- tant Modelling of CSOs Using SWMM’s EXTRAN Block. Proc., 1993 Stormwater and Water Quality Management Model- ing Conference (February 1993) pp. 189–204. Seidl, M., Belhomme, G., Servais, P., Mouchel, J. M., and G. De- mortier. Biodegradable Organic Carbon and Heterotrophic Bac- teria in Combined Sewer during Rain Events. Proc., 7th Interna- tional Conference on Urban Storm Drainage (1996) pp. 229–234. Thackston, E. L., and A. Murr, A. CSO Control Project Modifica- tions Based on Water Quality Studies. ASCE Journal of Envi- ronmental Engineering, Vol. 125, No. 10 (1999) pp. 979–987. Villeneuve, J. P., and P. Lavallee. Measured CSO Contribution to River Quality Deterioration and Methodologic Approach for Negative Influence Evaluation (August 1985) pp. 379–422. Widera, J., and P.C. Podraza. The Impact of a Combined Sewer Over- flow on the Ecology of Benthic Protozoa and Macro-invertebrates in a Small Urban Stream. Proc., 7th International Conference on Urban Storm Drainage (September 1996) pp. 1847–1852. 3.5.7. Deicing Agent Impacts Deicing agents were discussed briefly under the context of Highway Runoff Characterization and Assessment. This sec- 130 tion evaluates the impacts of deicing agents on stormwater runoff. Potential research questions include • How do deicing agents impact receiving waters, and what are the least toxic alternatives? • What factors influence or compound receiving waters impacts, and how can these factors be minimized? A comprehensive study performed by Michigan DOT investigated the environmental and economic impacts of de- icing agents. The study includes an analysis of various deicing materials: sodium chloride, calcium magnesium acetate (CMA), Motech, calcium chloride, and proprietary products such as CMS-B, CG-90 Surface Saver, and Verglimit. Deicers containing chloride salts were found to have similar impacts, which in turn appear different from impacts from deicers containing CMA. The Michigan DOT study and pre-existing research concur that only in rare situations can road salts cause significant direct impacts. Results from the model developed for the Michigan DOT study suggest that chloride concentrations in the Great Lakes will not reach toxic levels even in the worst-case scenario; however, chlorides can cause density stratification in smaller water bodies and CMA decomposition can lead to depletion of dissolved oxygen lev- els. The authors suggest diversion of runoff to less sensitive areas to attenuate impacts on receiving waters (Public Sector Consultants, 1993). Other research has been done to investigate the impacts of deicing agents in correlation to external variables such as meteorological and climatological factors. In a study near Jamesville, New York, Champagne (1977) found that pre- cipitation and temperature have an effect on the release of salts into receiving waters. Researchers also found that road salts can infiltrate into soils and can impact chloride levels in the receiving water long after road salt application. Research on modeling and simulation of the impacts of road salts has been performed in a number of studies. Halm (1997) discussed the development of a finite difference model and its application to airport deicing activities. Lewis (1999) conducted an evaluation of the environmental effects of the deicer magnesium chloride, widely used by Colorado DOT during winter highway maintenance activities. The literature review preceding the investigation indicated that magnesium chloride deicers are unlikely to produce adverse environ- mental effects. However, magnesium chloride may contain other chemicals such as rust inhibitors, which may consist of organic compounds that increase the oxygen demand. These chemicals have not been studied adequately. Results of the Lewis study found that no significant increases in BOD could be detected as a result of the addition of 0.3% deicer solution. Biotoxicity testing was conducted on boreal toad tadpoles, juvenile rainbow trout, Ceriodaphnia (aquatic invertebrate), and Selenastrum (algae). Tadpoles and juvenile rainbow trout showed no mortality over 96-hour intervals at deicer con- centrations of 0.1%, which is close to the expected median deicer concentration within short distances from the road-

way. Ceriodaphnia had a 48-hour threshold of mortality at 0.1%, and Selenastrum showed significant suppression of division rate for algal cells at deicer concentrations slightly in excess of 0.1%. The overall conclusion of the study is that application of magnesium chloride deicer having a chemical composition and application rate similar to those typically used by Colorado DOT is highly unlikely to cause or con- tribute toward environmental damage at distances greater than 20 yards from the roadway. Even very close to the road- way, the potential of magnesium chloride deicer to cause environmental damage is probably much smaller than that of other factors related to road use and maintenance, including pollution of highway surfaces by vehicles and use of salt and sand mixtures to promote traction in winter. Magnesium chloride deicer may offer net environmental benefits if its use leads to a reduction in the quantity of salt and sand applied to roadways. A study on the effects of runoff from Chautauqua Lake Bridge, in western New York, on sunfish further illustrates this toxicity (Adams-Kszos et al., 1990). NaCl appeared to be the major contributor to the toxicity of runoff from the Chautauqua Lake Bridge in laboratory bioassays. However, concentrations of zinc and cadmium present in the 50% win- ter runoff were in the range reported to be toxic to fish and may have been additive or synergistic with the NaCl toxicity in the laboratory bioassays. Because runoff from the Chau- tauqua Lake Bridge is diluted greatly when it enters the lake, it is unlikely that bridge runoff will be toxic. However, if runoff comparable to that entering Chautauqua Lake during the winter were to enter a much smaller body of water, the metals and NaCl would probably cause significant harm to freshwater organisms. The effects of the highway deicing activities on the Peshastin Creek watershed in Washington were studied over a 6-month period from December 1999 to May 2000. Steelhead (Oncorhynchus tshawytscha), Chinook salmon (Oncorhynchus mykiss), and bull trout (Salvelinus confluen- tus), three threatened/endangered species, inhabit the stream, and therefore a study of the effects of deicing activities was warranted. Five reaches along Peshastin Creek and its tribu- taries were selected for the collection of weekly grab samples and three of these reaches were outfitted with continuous monitoring equipment. Water quality tests, Microtox® toxi- city tests, benthic macro invertebrate enumeration, and stream- bed substrate sieve analyses were used to evaluate the influ- ence of deicing activities (application of traction sand and IceBAN, a liquid deicer) on Peshastin Creek. Chloride exhib- ited signs of preferential elution and was found to be signif- icantly higher in concentration in areas adjacent to the US Highway 97. The maximum recorded chloride concentration in Peshastin Creek was 3.3 mg/L and 2.7 mg/L at reach 2 and reach 4, respectively. The nonimpacted reaches of Peshastin yielded an average chloride concentration of 0.62 mg/L. Heavy metals concentrations (soluble and total) were much lower than EPA’s recommended limits. The benthic macro invertebrate study, although qualitative in nature, suggested 131 that the deicing activities did not adversely impact the three fish food organisms that were quantified. Streambed substrate analyses indicated that the traction sand used in deicing activ- ities had no measurable negative impact on known spawning locations. The physical, chemical, and biological parameters evaluated in this study indicate that deicing activities along SR 97 had no measurable negative impact on Peshastin Creek. 3.5.7.1. Identification of Research Needs With regard to the receiving water impacts associated with deicing agents, it appears there is a need for a database con- taining an evaluation of the human health and receiving water impacts along with toxicity test results for all existing deicing agents to aid in the selection of deicing agents. Other potential research needs include the evaluation of the persis- tence and implications of various deicing agents in roadside soils, evaluation of the factors that influence or compound receiving water impacts, and the development of strategies to minimize impacts. Recommendations suggested by Fischel (2001) include the development and implementation of deic- ing strategies that reduce the amount of chemicals required and the development of decision support systems based on weather conditions to optimize deicing operations. Minimiz- ing the amount of deicing chemicals used in deicing opera- tions results in a corresponding reduction of the impacts to receiving waters. 3.5.7.2. Primary References Adams-Kszos, L., Winter, J. D., and T. A. Storch. Toxicity of Chau- tauqua Lake Bridge Runoff to Young-of-the-Year Sunfish (Lep- omis macrochirus). Bulletin of Environmental Contamination and Toxicology, Vol. 45, No. 6 (1990) pp. 923–930. Champagne, D. M. Impact of Highway Deicing Salts on Rural Stream Water Quality. In Transportation Research Record 647, TRB, National Research Council, Washington, DC (1977) pp. 47–52. Fischel, M. Evaluation of Selected Deicers Based on a Review of the Literature. Final Report No. CDOT-DTD-R-2001-15, Col- orado Department of Transportation Research Board (October 2001) 117 pp. Halm, M. J. A Finite Difference Model to Predict Stream Water Quality Impacts as a Result of Airport Deicing Activities. Proc., Water Environment Federation 70th Annual Conference and Exposition (October 18–22, 1997) pp. 319–330. Lewis, W. M. Jr. Studies of Environmental Effects of Magnesium Chloride Deicer in Colorado. Final Report No. CDOT-DTD-R- 99-10, Colorado Department of Transportation Research Board, U.S. Department of Transportation, Federal Highway Adminis- tration (1999) 101 pp. Public Sector Consultants, Inc. The Use of Selected Deicing Mate- rials on Michigan Roads: Environmental and Economic Impacts. Michigan Department of Transportation, Lansing (December 1993).

3.5.8. Groundwater Quality Analysis and Impacts Increase in impervious area due to urbanization interferes with groundwater recharge. Developed areas that allow some form of infiltration are likely to inject contaminants carried in surface runoff into groundwater. New sources of groundwater recharge that have resulted from urbanization include domes- tic septic tanks, industrial waste injection wells, agricultural and residential irrigation, and infiltration BMPs (Pitt et al., 1994). All of these new sources of groundwater recharge have the potential to cause negative groundwater impacts. After development, if most runoff is infiltrated, it is likely that over- all infiltration volumes will be higher than before redevelop- ment, as the water loss from evapotranspiration is reduced. Therefore, the potential for increased loadings as well as increased concentrations is created. Widespread adoption and acceptance of infiltration BMPs as stormwater runoff treatment and control methods have spawned questions as to whether contaminants are treated adequately before runoff mixes with groundwater. Natural organic matter (NOM) present in stormwater reacts with heavy metals to form complexes that have shorter transport times. To investigate the implications of this phenomenon on ground- water impacts, Hathhorn and Yonge (1995) performed a two- phase study to investigate heavy metal–NOM transport mechanics. They found that dissolved organic matter enhanced the transport of lead through NOM-metal com- plexation. To minimize groundwater impacts, they recom- mended that in siting an infiltration facility, the organic con- tent of the soil and the background metal content should be determined. Also, the distance to groundwater should be increased from 3 feet to approximately 10 feet. A study by Barraud et al. (1999) investigated the potential impacts to groundwater from two soakaway (i.e., underground injection control) facilities receiving urban runoff. One facil- ity was 2 years old and one facility was more than 30 years old. Groundwater was monitored during storm events 1 m and 1.5 m down-gradient from the newer and older facilities, respectively. Soil quality also was measured. The results for the newer facility indicated that metal and hydrocarbon con- centrations were high near the injection surface but decreased rapidly a few decimeters down. However, the older facility indicated that heavy metals and mineral oils can contaminate the soil over a radius of at least 1 meter around the infiltra- tion facility. Impacts to groundwater were low, but there were measurable increases in copper, lead, and zinc concen- trations as compared to background groundwater concentra- tions. The authors noted that since the data were highly vari- able and few data points were monitored, it was difficult to draw any definite conclusion from the study. In a study to assess impacts of an exfiltration pipe, a detention pond, a retention pond, and two swales in Florida, Schiffer (1989) found the concentrations of chromium, cop- per, and lead in groundwater to be below detection limits. Groundwater near the ponds had the highest TKN levels, 132 while the highest levels for nitrate nitrogen and phosphorus were observed near the swales and the exfiltration pipe. Con- taminant concentrations were monitored from 1984 through 1986. The results of this Florida DOT study showed an atten- uation of inorganic contaminants; yet, the researchers con- cluded that organic compounds in the retention pond sedi- ments may eventually impact groundwater quality. Sela (1994) presented graphical methods that were applied to identify areas of high groundwater sensitivity in a 14-mile- long highway widening project in New Jersey. This infor- mation could find potential applications in BMP siting and design. Little research has been done on DOT impacts to ground- water. The most notable DOT groundwater impacts histori- cally have arisen from maintenance yard contamination to wells, in which case the DOT has sometimes bought wells, homes, or even larger developments. In Pennsylvania DOT’s case, such expensive impacts ultimately led to the agency’s development of ISO14001-certified environmental manage- ment systems in maintenance districts. In cases of special danger where spills contaminate water recharge areas, DOTs have been known to develop agreements and “double ditch” a facility separating water that came underneath the road from water that was coming off the highway to prevent any type of highway spills from affecting the groundwater and endan- gered species. There have been numerous studies on MTBE. MTBE is an oxygenate used to increase oxygen levels in gas, thereby enhancing combustion and decreasing carbon monoxide emis- sions (Delzer et al., 1996). In a study to investigate the extent of MTBE contamination, USGS collected 592 stormwater samples in 16 cities and metropolitan areas from 1991 through 1995. The results of this study as summarized by Delzer et al. (1996) detected MTBE in 7% of the samples analyzed. Another study presented by Squillace et al. (1996) found MTBE to be detected most frequently in shallow groundwater; MTBE was detected in 27% of the 210 shallow groundwater wells sam- pled in eight areas versus 1% of 412 deep groundwater wells sampled in nine areas. A significant number of shallow wells in urban areas were contaminated as compared to shallow wells in agricultural areas. The majority of the studies on MTBE encountered in this effort appear to be centered on groundwater. However, the Metropolitan Water District of Southern California surveyed six reservoirs, which serve as sources of drinking water in Southern California, to deter- mine the level of MTBE impacts. This study by Dale et al. (2000) found that motorized watercraft can be a significant source of MTBE. Infiltration systems usually are not designed with any concern for pollution retention. However, it is sometimes proposed that polluted stormwater should pass through a humic layer at the soil surface to effectively screen off any present well-absorbable or degradable pollutants, whereas clean stormwater should be allowed to infiltrate directly into the underground. The implications of such procedures have

not yet been investigated thoroughly, and it is rarely realized that they may lead to an unacceptable contamination of sur- face soils. In order to investigate the potential impacts to soil and groundwater that have received runoff water from highly trafficked roads for several decades, a field study of a surface infiltration system and a subsurface infiltration system was conducted (Mikkelsen et al., 1996). The results of the inves- tigation found that the infiltration systems served as effective pollutant traps for copper, zinc, cadmium, lead, PAHs, and adsorbed organically bound halogens, and the potential for groundwater contamination caused by leaching of heavy met- als was of low concern. The authors noted that soluble con- stituents such as many pesticides and deicing salts may pass directly through infiltration systems with little or no retention in the soil matrix and should be investigated further. Differ- ences between the ability of surface and subsurface infiltra- tion systems to retain pollutants were not found to be signif- icant, but it was indicated that retention capacity was largely a function of neutral to weakly alkaline pH conditions, and a similar observation may not occur in other types of geology. 3.5.8.1. Identification of Research Needs Based on the review of literature pertaining to potential impacts to groundwater caused by infiltration of stormwater runoff, there appears to be a need for more research. The methods used to assess impacts are difficult to implement, and the results are difficult to assess. State DOTs need a pro- cedure to estimate the potential extent and magnitude of groundwater quality degradation from transportation BMPs, particularly those that rely on infiltration at their primary treatment mechanism. This guidance would include proce- dures for identifying and evaluating current and potential uses of groundwater and water quality requirements that could be affected by transportation BMPs. The direction of flow movement in groundwater aquifers needs to be identi- fied. Any pollutant plumes in aquifers must be evaluated, including direction of flow and concentrations. Treated storm- water quality from transportation BMPs that could infiltrate groundwater should be identified in terms of flows, con- stituents, and concentrations. The role of geology in pollu- tant retention appears to be a research gap that needs to be filled. The distance between BMP invert and the maximum groundwater elevation must be determined, as must the rate of flow downward to the groundwater. With regard to the persistence of MTBE in groundwater, potential research questions and areas of interest outlined by Delzer et al. (1996) include the following: • How persistent is MTBE in streams, what is the rate of degradation, and what are the potential impacts on aquatic life? • What proportions of MTBE are contributions from pre- cipitation versus runoff contributions? 133 • Do other oxygenates behave in a similar manner? • How do factors such as land use relate to MTBE occur- rence? • What are the proportions of contributions from storm- water recharge and precipitation to MTBE in ground- water? In summary, the fate and transport of stormwater con- stituents from BMPs as the constituents move through the soil mantle and ultimately move through groundwater must be determined. Past guidance for siting infiltration BMPs has focused on minimum depth to groundwater; however, the fil- tering and sorption capacity of the soils the water passes through are more important considerations. 3.5.8.2. Primary References Barraud, S., Gautier, A., Bardin, J. P., and V. Riou. The Impact of Intentional Stormwater Infiltration on Soil and Groundwater. Water Science and Technology, Vol. 39, No. 2 (1999) pp. 185–192. Dale, M. S., Koch, B., Losee, R. F., Crofts, E. W., and M. K. Davis. MTBE in Southern California Water. Journal of the American Water Works Association, Vol. 92, No. 8 (2000) pp. 42–51. Delzer, C., Zogorski, J. S., Lopes, T. J., and L. R. Bosshart. Occur- rence of the Gasoline Oxygenate MTBE and BTEX Compounds in Urban Stormwater in the United States, 1991–1995. Water- Resource Investigations Report 96-4145 (1996). Hathhorn, W. E., and D. R. Yonge. The Assessment of Groundwater Pollution Potential Resulting from Stormwater Infiltration BMPs. Technical Report WA-RD-389.2, Washington Department of Transportation (August 1995) 181 pp. Mikkelsen, P. S., Haefliger, M., Ochs, M., Jacobsen, P., Tjell, J. C., and M. Boller. Pollution of Soil and Groundwater from Infiltra- tion of Highly Contaminated Stormwater, A Case Study. Proc., 7th International Conference on Urban Storm Drainage (Sep- tember 1996) pp. 707–712. Pitt, R., Clark, S., and K. Parmer. Protection of Groundwater from Intentional and Non-Intentional Stormwater Infiltration. U.S. EPA Report No.EPA/600/SR-94/051 and No.PB94-165354AS, U.S. Environmental Protection Agency Storm and Combined Sewer Program, Cincinnati, OH (May 1994) 187 pp. Schiffer, D. M. Impacts of Stormwater Management Practices on Groundwater. Florida Department of Transportation Final Report FL/DOT/SMO/90-378 (1989) 79 pp. Sela, E. Mitigation of Non-Point Source Pollution Impacts on Groundwater Aquifers: A Case Study. Proc., ASCE 1st Annual Conference on Water Policy, New York, NY (1994) pp. 246–249. Squillace, P. J., Zogorski, J. S., Wilber, W. G., and C. V. Price. A Preliminary Assessment of the Occurrence and Possible Sources of MTBE in Groundwater of the United States, 1993–94. Open File Report No. 95-456, U.S. Geological Survey, Washington, DC (1996). 3.5.9. Wetland Impacts Wetlands provide numerous benefits that include flood and erosion control and water quality improvement. Wet-

lands are home to one-third of all federally listed endan- gered species. Unfortunately, the number of wetlands has been reduced drastically because of urbanization. Twenty- two states have lost approximately one-half of their wet- lands, while California, Iowa, and Ohio have lost about 90%. The bulk of wetland losses have occurred as a result of agricultural conversion, natural erosion, and urbanization— not as a result of highway construction. According to a study by Apogee Research Inc. (1997), between 310,000 and 570,000 acres of wetlands have been lost as a result of FAHP construction activities between 1955 and 1980. Replacement costs of such wetlands start between $153 mil- lion and $6 billion. A review of the literature pertinent to wetland impacts reveals that much research has been done on this subject. However, the discussion herein is limited to wetland impacts caused specifically by highway stormwater runoff. The most pertinent research questions with regard to highway runoff are • To what highway runoff pollutants are natural wetlands most sensitive? • Can constructed or mitigated wetlands successfully be used to treat highway runoff without impacting local biota? The impacts of highway construction and operation on wet- lands have been the subject of a number of studies. Harris et al. (1984) discuss a wetland monitoring program devel- oped by the Arkansas Highway and Transportation Depart- ment to document impacts to a wetland during the construc- tion of US-67 in White County Arkansas. Yu et al. (1998) evaluate two mitigated wetlands constructed by Virginia DOT. Highway runoff provided the primary source of water for both wetlands. The researchers found that the habitat and biota remained healthy and diverse for both wetlands. Highway operation and maintenance practices that may potentially impact groundwater are likely to impact nearby wetlands, too. A hydrogeologic investigation (Panno et al., 1999) indicated the migration of contaminants into two wetlands via groundwater. The investigation consisted of a 15-month-long hydrogeologic evaluation of a fen-wetland complex in northeastern Illinois. The origin of the high con- centrations of Na+ and Cl− ions in groundwater plumes were linked to a private septic tank and road salt operations. Observed impacts to fen vegetation included succession by salt-loving plant species. Large concentrations of sulphate in the second wetland were linked to oxidation of pyrite within underlying soils. There were no discernable impacts on fen vegetation from the high sulphate concentrations. The study demonstrated how easily septic systems and deicing opera- tions could negatively impact wetland vegetation. A comprehensive synthesis of federal programs that impact wetlands is presented in a two-volume report by the U.S. 134 Department of Interior (U.S. Department of the Interior, 1988 and 1994). Wetland impacts resulting from federal pro- grams such as agricultural programs; water development and management programs; infrastructure; local development and housing programs; and federal programs to promote resource use, extraction, and development are presented in the context of regional variability of impacts. The regions studied in the report include the Mississippi Delta Region; the Prairie Pot- hole Region; southeastern Alaska; the Central Valley in Cal- ifornia; the Everglades in Florida; Maryland’s Eastern Shore; Coastal Michigan; Northern Michigan; the Pocosins in North Carolina; New Jersey; the Puerto Rican Mangroves; the Texas Coast; and riparian areas in Idaho, Nevada, and New Mexico. The literature has established that highway runoff can have a negative impact on wetlands; however, the provision of some form of detention as pretreatment prior to the wetland has been shown to significantly alleviate impacts. An evalu- ation by Schiffer (1989) of the effects of highway runoff on two wetlands in central Florida showed that the concentra- tion of automobile-related contaminants and sediment can be reduced by detaining runoff before it is released into wet- lands. Spatial variations of pollutants within the freshwater marsh indicated that for most contaminants, concentrations decrease with increasing distance from the inlet. Color, total organic carbon, and chromium concentrations behaved in the opposite manner. The behavior of chromium may be due to the fact that chromium remains dissolved longer than some of the other metals and could also be linked to atmospheric deposition. The study concluded that detention structures larger than the 12′-by-25′ trash retainer used at the fresh- water marsh may provide significant sorption and settling of contaminants, thereby minimizing impacts to wetlands. Mitigated wetlands formed as a result of highway con- struction projects or for economic incentives or wildlife habi- tat creation can provide significant water quality benefits. A study (Yu et al., 1998) examined the feasibility of using mit- igated wetlands as stormwater BMPs. Two mitigated wet- lands were evaluated and monitored during storm events. Wetland vegetation density and wildlife diversity were used as metrics of highway runoff impacts. Peak flow reduction for both sites was observed to be in excess of 40%. Removal rates for TSS, COD, total phosphorous, orthophosphate, and zinc were as high as 90%, 65%, 70%, 70%, and 50%, respec- tively. Vegetation and wildlife at the two sites were observed to be healthy and diverse. According to Knight et al. (1998), the Greens Bayou Wetland Mitigation Bank, implemented in cooperation with Texas DOT, also was intended to pro- vide stormwater quality mitigation benefits. Approximately 220 acres of wetlands are included in the project. An intricate train of treatment was included as part of the design to pro- vide multiple benefits such as highway runoff water quality improvement, flood flow retention, and the creation of wild- life habitat. This project will provide treatment for a signifi-

cant portion of the additional flows resulting from the expan- sion of Beltway 8. Larson and Neill (1987) examined three main biophysical elements of wetlands (hydrology, soils, and vegetation) in relation to artificial wetlands constructed in fulfillment of mit- igation requirements. The importance of each of these ele- ments to basic wetlands function was evaluated, and the data requirements for assessing the significance of each of the ele- ments were defined. Hunt et al. (1999) discussed the use of an in-stream wetland for nitrogen removal in a contaminated stream. The authors concluded that in-stream wetlands are good landscape features that can be used to mitigate excess nitrogen and are a good complement to other BMPs. 3.5.9.1. Identification of Research Needs Based on the review of literature, the potential highway runoff impacts on natural and mitigated wetlands appear to be well documented. The tendency for many highway runoff pollutants to accumulate in wetland sediments and vegetation raises some concern with regard to long-term impacts on wet- land biota. Current regulatory requirements for monitoring and assessing impacts to existing wetlands ensure that water quality and sediment quality, as well as toxicity data, are and will continue to become available for analysis. However, as discussed in section 3.5.4, there is a general lack of applicable bioindicators for evaluating impacts associated with the episodic nature of stormwater runoff. A potential research need may be to develop indicators for assessing impacts to wetlands from highway runoff by conducting a detailed analy- sis of currently available data on wetlands receiving runoff from highway facilities. 135 3.5.9.2. Primary References Harris, J. L., Burnside, F. L., Richardson, B. L., and W. K. Welch. Methods for Analysis of Highway Construction Impacts on A Wetland Ecosystem—A Multidisciplinary Approach. In Trans- portation Research Record 969, TRB, National Research Coun- cil, Washington, DC (1984) pp. 8–17. Hunt, P. G., Stone, K. C., Humenik, F. J., Matheny, T. A., and M. H. Johnson. In-Stream Wetland Mitigation of Nitrogen Con- tamination in a USA Coastal Plain Stream. Journal of Environ- mental Quality, Vol. 28 (1999) pp. 249–256. Knight, R. L., Adams R., O’Brien, C., and E. R. Davis. Beltway 8 Wetland Water Quality Project: Constructed Wetlands for Storm- water Polishing and Wetland Mitigation Banking. In Transporta- tion Research Record 1626, TRB, National Research Council, Washington, DC (1998) pp. 11–20. Larson, J. S., and C. Neill. Mitigating Freshwater Wetland Alter- ations in the Glaciated Northeastern United States: An Assess- ment of the Science Base. The Environmental Institute, Vol. 87, No. 1, University of Massachusetts, Amherst (1987) 143 pp. Panno, S. V., Nuzzo, V. A., Cartwright, K., Hensel, B. R., and I. G. Krapac. Impact of Urban Development on the Chemical Compo- sition of Groundwater in a Fen-Wetland Complex. Wetlands, Vol. 19, No. 1 (1999) pp. 236–245. Schiffer, D. M. Effects of Highway Runoff on the Quality of Water and Bed Sediments of Two Wetlands in Central Florida. Water- Resources Investigations Report 88-4200 (1989) 63 pp. U.S. Department of the Interior. The Impact of Federal Programs on Wetlands, Volume 2. A Report to Congress by the Secretary of the Interior (March 1994) 333 pp. U.S. Department of the Interior. The Impact of Federal Programs on Wetlands, Volume 1: The Lower Mississippi Alluvial Plain and the Prairie Pothole Region. A Report to Congress by the Secre- tary of the Interior (October 1988) 114 pp. Yu, S. L., Earles, T. A., and G. M. Fitch. Aspects of Functional Analysis of Mitigated Wetlands Receiving Highway Runoff. In Transportation Research Record 1626, TRB, National Research Council, Washington, DC (1998) pp. 21–30.

Next: Chapter 4 - Summary of Identified Research Gaps and Needs »
Identification of Research Needs Related to Highway Runoff Management Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB's National Cooperative Highway Research Program (NCHRP) Report 521: Identification of Research Needs Related to Highway Runoff Management summarizes significant stormwater management practices and research efforts, and it identifies the most pressing gaps and needs in the current state of knowledge in over more than 30 subject areas. The report includes full research project statements for the topics considered to be of highest priority.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!