Volume Reduction of Highway Runoff in Urban Areas: Final Report and NCHRP Report 802 Appendices C through F (2015)

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.

Annex A Volume Performance Tool Technical Documentation 1 Introduction and Purpose ..................................................................................................................... 1 2 Model Overview ..................................................................................................................................... 1 2.1 Functionality and Applicability ........................................................................................................ 1 2.2 User Experience .................................................................................................................................. 1 3 Technical Framework ........................................................................................................................... 2 3.1 Volume Reduction Mechanisms ........................................................................................................ 2 3.2 Capture Efficiency.............................................................................................................................. 3 3.3 Volume Reduction Efficiency ............................................................................................................ 4 3.4 Total Volume Reduction Performance ............................................................................................. 5 3.5 Accounting for Multi-Compartment VRAs and Treatment Trains .............................................. 5 3.6 Approach for Localizing Model Estimates ...................................................................................... 7 4 Modeling Methodology and Parameters for Tool Development ....................................................... 9 4.1 Types of Modeling Conducted .......................................................................................................... 9 4.2 SWMM Model Inputs ...................................................................................................................... 10 4.2.1 Precipitation Data ..................................................................................................................................... 10 4.2.2 Evapotranspiration Data ........................................................................................................................... 11 4.2.3 SWMM Hydrologic Parameters ............................................................................................................... 12 4.2.4 Storage Volume Increments ..................................................................................................................... 13 4.2.5 Drawdown Increments .............................................................................................................................. 14 4.2.6 ET Depth Increments ................................................................................................................................ 14 4.2.7 Dispersion Model Runs ............................................................................................................................ 14 4.2.8 Summary of Supporting Model Runs ....................................................................................................... 15 5 Summary of Tool Calculations ........................................................................................................... 16 6 Simplifying Assumptions and Reliability .......................................................................................... 18 7 Supplemental Case Study Analyses to Evaluate Reliability of Tool Methodology ........................ 20 7.1 Evaluation of Simplifying Assumption 1 ........................................................................................ 21 7.1.1 Hypothesis ................................................................................................................................................ 21 7.1.2 Analyses ................................................................................................................................................... 22 7.1.3 Summary of Findings ............................................................................................................................... 23 7.2 Evaluation of Assumption 2 ............................................................................................................ 23 7.2.1 Hypothesis ................................................................................................................................................ 23 7.2.2 Analysis .................................................................................................................................................... 23 7.2.3 Summary of Findings ............................................................................................................................... 24 7.3 Evaluation of Simplifying Assumption 3 ........................................................................................ 24 7.3.1 Hypothesis ................................................................................................................................................ 24 7.3.2 Analysis .................................................................................................................................................... 24 7.3.3 Summary of Findings ............................................................................................................................... 26 7.4 Evaluation of Simplifying Assumption 4 ........................................................................................ 26 A-i

7.4.1 Hypothesis ................................................................................................................................................ 26 7.4.2 Analysis .................................................................................................................................................... 26 7.4.3 Summary of Findings ............................................................................................................................... 27 7.5 Summary ........................................................................................................................................... 27 8 References ............................................................................................................................................. 27 9 Precipitation Gages Supported by Tool ............................................................................................. 28 A-ii

1 Introduction and Purpose The performance of volume reduction approaches (VRAs) is a function of many factors, including local climate and hydrology, storage volume, VRA design (i.e., footprint, depth, and discharge rates), underlying soil properties, and other factors. Because volume reduction occurs to different degrees in storms with different sizes, shapes, and antecedent conditions, it is necessary to utilize long term hydrologic and hydraulic modeling methods (i.e., continuous simulation, rather than design event simulation) to provide a reliable estimate of long term volume reduction. The Volume Performance Tool (tool) was developed as part of NCHRP Project 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas to allow a user to estimate the approximate volume reduction performance of a VRA or series of VRAs (i.e., treatment train), given the site location and planning level information about the site conditions and the VRA design. The purpose of this annex is to provide technical documentation for the Volume Performance Tool. 2 Model Overview 2.1 Functionality and Applicability The tool is intended to be used to evaluate a wide variety of single and multi-component VRA design scenarios for a user-defined location. To provide “real-time” estimates of volume reduction performance (i.e., immediately available, without creating and executing a continuous simulation model run), the tool queries the results of thousands of pre-executed long term hydrologic and hydraulic model simulations executed as part of the development of the tool. Using scaling and interpolation methods developed for this purpose (as discussed further in Sections 3.5 and 3.6), site-specific estimates can be derived for a specific design scenario based on a pre-determined array of hypothetical simulations conducted at the location of gages within each climate division. The tool itself is an Excel spreadsheet application in which the user selects a location, provides planning level project information and the tool provides an estimate of long term volume reduction. The tool is intended to allow DOTs to quickly evaluate the relative benefits of various conceptual design scenarios as well as to assist in develop sizing criteria. The tool is not intended solely for highway projects; however, assumptions have been made as part of the development of the tool such that the tool is expected to be most reliable for the conditions typically encountered in highway environment (specifically smaller catchments with higher percentage impervious area). Volume reduction performance of VRAs does not generally exhibit significant sensitivity to catchment size, as volume reduction processes tend to occur at somewhat longer timescales than peak flow phenomena; therefore the tool is considered to be reliable for planning level analysis of other types of catchments as well (e.g., larger, less impervious catchments). The tool is intended to provide a planning level estimate of volume reduction performance to assist in prioritizing VRAs at a planning level. Where more rigorous site-specific estimates of volume reduction performance are desired, the results of the tool should be refined through the use of a more detailed model that is developed specifically to represent local conditions. 2.2 User Experience The tool is intended to be accessible to a wide variety of practitioners around the country without significant prior effort required in reviewing user guidance or developing inputs. The tool consists of a A-1

macro-enabled Excel spreadsheet that contains a “lookup database” – a directory of data files that contain summary statistics from the long term hydrologic and hydraulic model runs completed as part of developing the tool. The user navigates through the tool using navigation buttons or tabs, populating information about the specific project and VRA scenario and viewing estimated volume performance results. The User’s Guide for the tool (Appendix B of the Guidance Manual) provides step by step guidance for using the tool and provides more explanation of the user interface and experience. 3 Technical Framework 3.1 Volume Reduction Mechanisms The total amount of volume reduction achieved in a VRA is a function of the amount of water that enters the VRA and does not immediately overflow (i.e., the amount of water that is captured), and the portion of the captured water that is “lost” via infiltration, ET, and/or consumptive use (i.e., the total of all three is the volume reduction), such that it does not discharge directly to surface water. When evaluating capture efficiency and volume reduction, each VRA can be considered to consist of a set of storage compartments, each with a distinct storage volume, discharge rate, and pathway by which water discharges (i.e., surface discharge, infiltration, ET). Figure 1 illustrates this concept. When storage capacity is available in a given compartment, then that compartment of the VRA can capture additional inflow. When storage capacity is not available in a given compartment to accept additional inflow, then inflowing water either fills the next storage compartment of the VRA, or bypasses the system (if no additional storage is available). The capture and volume reduction performance of a VRA is primarily a function of the amount of storage volume provided and the rate at which the storage drains to volume reduction pathways (i.e., infiltration, ET, consumptive use) versus surface discharge pathways. The menu of VRAs that is supported as part of the Guidance Manual includes VRAs with retention storage only, as well as systems with retention and detention storage. In the case of systems with retention storage only (e.g., infiltration trenches), the volume reduction performance is a function of the capture efficiency only – all water that is captured is lost, as there is no surface discharge from these VRAs besides the overflow that occurs when retention storage is filled. In the case of VRAs with a combination of detention and retention storage (e.g., bioretention with underdrains), the total volume reduction performance is the product of the capture efficiency (the portion of overall runoff volume that is captured and managed) and the volume reduction efficiency (the portion of the captured water that is lost). A-2

Figure 1. Schematic Representation of VRAs for Purpose of Capture Efficiency and Volume Reduction Analysis 3.2 Capture Efficiency Capture efficiency (or “percent capture”) is a metric that measures the percent of rainfall that is captured and managed by a VRA (i.e., does not bypass or immediately overflow). Captured stormwater may be infiltrated, evapotranspired, or retained for harvest and use, and/or treated and release. Capture efficiency is typically expressed as an average capture rate over a long period of time, for example annual average percent capture. Runoff volume that is not captured by a VRA is referred to as bypass or overflow. Volume reduction processes can only occur in a VRA when water is captured. Long term capture efficiency is primarily a function of the VRA storage volume (relative to the size and runoff potential of the watershed), the drawdown rate and pattern of the storage compartment, and local precipitation patterns. Practically, this means that the following parameters can be isolated as primary predictors of capture efficiency for the purpose of developing an approximate predictive tool: • Normalized storage volume, expressed as an equivalent precipitation depth over the watershed that would produce a runoff volume equivalent to the VRA storage volume. For example, a 3,000 cu-ft storage volume for a watershed that is 1 acre with a runoff coefficient of 0.9 would translate to an equivalent precipitation depth of 0.92 inches [3,000 cu-ft × 12 in/ft / (1 ac × 43,560 sq-ft/ac × 0.9)]. • Drawdown time of the storage volume. For VRA storage elements with nominally consistent drawdown rates regardless of season (i.e., infiltration, filtration, orifice-controlled surface discharge), the representative drawdown time can be expressed in hours. For example, a bioretention area with a storage depth of 18 inches and an underlying design infiltration rate of 0.5 inches per hour would have a nominal drawdown time of 36 hours (18 inches / 0.5 in/hr). For VRA storage elements with seasonally varying drawdown rates (i.e., storage drained by ET or irrigation-based consumptive use), the concept of a representative drawdown time is not applicable. In this case, the ET storage depth (i.e., the amount of potential ET that must occur for the stored water to empty) is a more appropriate indicator of how quickly storage is recovered and can be used (along with local climate data input to the model) as a predictor of long term capture efficiency. A-3

By isolating these two most important predictive variables, a limited number of continuous simulation model runs and associated results can be used to describe the expected long term performance of a wide range of VRA types and configurations. For example, the results of a long term model simulation for a 0.75-inch normalized storage depth with 24-hour drawdown would be representative of a wide range of different VRA configurations. The two examples would both be reliably represented by this single model run. Example 1: 20,000 cu-ft infiltration basin draining 8.2 acres of pavement (equates to 0.75-inch equivalent storm), with 3-foot ponding depth and a design infiltration rate of 1.5 inches per hour (equates to 24 hour drawdown time). Example 2: 300 cu-ft bioretention area with underdrains with a tributary area of 0.122 acres of pavement (equates to 0.75-inch equivalent storm), with 12 inches of ponding storage depth and a design media filtration rate of 0.5 inches per hour (equates to 24-hour drawdown time). It be seen that an infinite number of potential design combinations could be reflected by this single model run. An array of continuous simulation runs were executed in the EPA Storm Water Management Model (SWMM, version 5.0.022), as described further in Section 4, to encompass the range of normalized storage volumes and drawdown times (or ET depths) that the tool supports. For each of the combinations of storage volume and drawdown time (or ET depth), the capture efficiency was calculated using results from the long term SWMM model. For each combination of design variables, the percent capture was calculated as: Percent Capture = 100[1− (𝑉𝑉𝑏𝑏𝑏𝑏/𝑉𝑉𝑐𝑐)] Where: Vby = the total volume bypassed or overflowed over the simulation period Vc = the total runoff volume flowing into the VRA over the simulation period 3.3 Volume Reduction Efficiency Volume reduction efficiency refers to the portion of the “captured” volume that is lost to infiltration, ET, or consumptive use and does not discharge directly to surface water. Within the tool, the following assumptions were made: • For storage compartments without a surface discharge pathway (i.e., retention storage), the volume reduction efficiency was set to 100 percent of the capture efficiency (i.e., complete retention of all water that is captured). • For storage compartments with surface discharge as well as significant volume loss pathways, the volume reduction efficiency was estimated by computing the average loss rate as a fraction of the average total discharge rate. For example, if the average surface discharge rate during the drawdown period is 2 inches per hour and the average infiltration plus ET loss rate during that period is 0.5 inches per hour, then the volume reduction efficiency would be estimated as 20 percent (0.5 / (2 + 0.5)). • For storage elements with only surface discharge pathways (i.e., lined systems with limited ET), the volume reduction efficiency was assumed to be zero. A-4

3.4 Total Volume Reduction Performance The total volume reduction performance was expressed in terms of: • Watershed relative volume reduction – the relative reduction in surface discharge volume compared to the same tributary area without controls, calculated as: Watershed relative volume reduction = Σ (% Cap) × (%VolRed) Where: % Cap = long term average annual capture efficiency % VolRed = long term volume reduction as a percentage of captured volume • Watershed absolute volume reduction – the difference in average annual runoff volume between the project condition without controls and the project condition with controls, calculated as: Watershed absolute volume reduction = Watershed relative volume reduction (%, computed per above) × Baseline Avg Annual Runoff Volume (without controls) Where: Baseline Avg Annual Runoff Volume (cu-ft) = Average Annual Precipitation Depth (inches) × Area (ac) × Runoff Coefficient / 12 inches/ft 3.5 Accounting for Multi-Compartment VRAs and Treatment Trains Many VRAs include a combination of retention and detention storage compartments that fill and drain in different orders and at different rates, and therefore are best represented as two or more discrete storage compartments. For each storage compartment, the methods described above can be used to estimate capture efficiency and volume reduction for that individual compartment. However, because the response between storage volume and capture efficiency is non-linear (i.e., above some storage volume there is a trend of increasingly diminishing incremental returns with incremental addition of storage), it is not reliable to simply add the capture efficiency achieved by each compartment as if each was an independent VRA. Similarly, where two VRAs are to be placed in series, it is not reliable to analyze each independently and simply sum the independent results. Holding all else equal, the first VRA compartment will tend to provide greater incremental capture efficiency than the second, and so on. To account for this non-linearity, a graphical/tabular method was implemented by a macro within the tool to estimate the total performance of multi-compartment VRAs and treatment trains, including compartments with different discharge pathways and different drain times for each discharge pathway. The method consists of the following steps and is illustrated graphically in Figure 2. These steps are conducted by the tool. The user is not responsible for implementing the steps listed below. 1. Order VRA compartments in terms of which compartment would fill first, before others would fill. For example, in the case of a bioretention area with a gravel sump below the underdrains, assume that the ET storage in soil pores would fill first and the sump would fill second before water would pool up into the detention storage compartments. Generally, it can be assumed that ET storage will fill first (as soil wets), infiltration storage will fill next (as the facility fills from the bottom up), and detention storage will be the last to fill. A-5

2. For the first compartment, the tool computes the equivalent storm depth (i.e., the storm depth that would be required to produce runoff from the watershed equal to the storage volume of the compartment) and drawdown time based on user inputs and finds the corresponding percent capture by querying the applicable lookup database. For the purpose of this example, assume the first compartment is the stone sump below a bioretention area with elevated underdrains and has storage equal to the runoff from a storm depth of 0.5 inches and a drawdown time of 72 hours. This combination yields a capture efficiency of 27 percent from the example nomograph (Figure 2). 3. Calculate the storage volume (as an effective storm depth) and drawdown time for the second compartment. In this example, this is the detention storage above the underdrains of the facility. This compartment has storage equal to the runoff from a storm depth of 0.75 inches and drains in 12 hours. 4. Using the percent capture of 27 percent obtained from Step 2, find the equivalent storm depth from the lookup database corresponding to the drawdown time of the next storage compartment (12 hours). In this example, the 27 percent capture achieved by the first compartment corresponds to an equivalent precipitation depth of 0.2 inches for the 12-hour drawdown scenario. 5. Next, “traverse” the nomograph curve (or table) up the 12-hour drawdown line a distance corresponding to the storage volume provided in the next compartment. In the example below, the second compartment has a storage volume equivalent to storm depth of 0.75 inches and a drawdown time of 12 hours. Therefore, the nomograph is traversed from 0.20 to 0.95 inches in the x-axis, corresponding to an increase from 27 percent capture to 75 percent capture in the y- axis. 6. Find the corresponding percent capture from the current position on the nomograph/table. In this case, the two compartments collectively achieve approximately 75 percent capture (27 percent by the first compartment plus 48 percent by the second compartment). 7. Repeat steps 2-5 for each remaining BMP compartment or treatment train element. The validity of this approach has been demonstrated as part of previous work (Orange County Public Works, 2011, Ventura County Watershed Protection District, 2011) and was tested and demonstrated for this specific application in comparison to more explicit model representations as part of development of the tool (See Section 7). A-6

Figure 2: Example application of method for incorporating multi-compartment VRAs (hypothetical location) 3.6 Approach for Localizing Model Estimates The tool was pre-packaged with lookup databases containing model results for 344 climate divisions (represented by a single point location precipitation/weather gage for each division) representing a wide range of conditions across the contiguous US. However, within each climate division there can be significant variability in precipitation patterns that may influence capture efficiency and volume reduction performance of VRAs. For example, within Los Angeles County alone, the average annual precipitation depth ranges from less than 10 inches up to more than 30 inches per year, and the 85th percentile, 24-hour storm depth (the local regulatory storm for sizing of water quality facilities) ranges from less than 0.75 inches to more than 1.5 inches (Los Angeles County Department of Public Works, 2012). As a result of intra-region variability, which likely occurs to different degrees in different climate divisions, there is potential for error to be introduced into the estimates provided by the tool, which are based on only one point location within each climate division. In areas of the climate division where storms are larger and/or more intense than the location of the gage that was analyzed, the tool would tend to over-estimate site- specific performance (actual performance would be less than what the tool predicts), and vice versa. However, it was not practicable to conduct continuous simulation at every possible project location because of the lack of reliable hourly data in some areas and the computational burden and associated quantity of data that would need to be pre-packaged with the tool. To address this limitation, the research team developed and tested an approach for “localizing” estimates within each climate division to improve site-specific reliability. This approach was based on the finding that the 85th percentile, 24-hour storm depth provides a reliable scaling factor for translating model results within a climate division -- when the sizes of VRAs are scaled from point to point based on 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Pe rc en t C ap tu re Equivalent Storm Depth, inches 12 72 Drawdown Time, hrs Compartment 1: 0.5 inch storage at 72 hour Compartment 2: 0.75 inch storage at 12 hour drawdown A-7

the relative magnitudes of the 85th percentile, 24-hour storm event at each point, then similarity in long term capture efficiency and volume reduction appear to be maintained. For example the capture efficiency would be expected to be similar for the following two points because their design volume as a fraction of the 85th percentile, 24-hour runoff volumes are the same: Point A: Runoff volume from 85th percentile, 24-hour storm = 3,200 cu-ft Actual VRA sized for 2,000 cu-ft Design volume as fraction of 85th percentile, 24-hour runoff volume = 2,000/3,200 = 0.625 Point B: Runoff volume from 85th percentile, 24-hour storm = 5,000 cu-ft Actual VRA sized for 3,125 cu-ft Design volume as fraction of 85th percentile, 24-hour runoff volume = 3,125/5,000 = 0.625 In other words, in order to achieve the same capture efficiency at Point B as Point A, the actual VRA size would need to be scaled up by the ratio of the 85th percentile, 24-hour storm depths. Alternately, if a VRA of the same size was provided at both locations, higher capture efficiency would be expected at Point A than Point B because Point A has a smaller 85th percentile, 24-hour storm event. The reliability of this method is discussed in Section 7. This relationship provides a method for estimating the performance at a given point in a division based on continuous simulation model conducted elsewhere within the division (i.e., at the main gage selected for each division). The scaling and localization method was implemented as follows; calculations are conducted within the tool (i.e., not by the user). 1. From the lookup database supporting the tool, the 85th percentile, 24-hour depth for each of the 344 gages are provided to the user. The 85th percentile, 24-hour depth was calculated by tabulating all daily precipitation totals (hourly or daily precipitation records can be used), filtering the list to exclude daily totals less than 0.11 inches, and ranking the list to identify the 85th percentile value (USEPA 2009, EISA Technical Guidance). 2. The user has the option to either use the default storm depth for that climate division (implicitly this means that the project is represented by the rainfall of the primary gage for that division) or enter the 85th percentile, 24-hour storm depth for their specific project location. This can be relatively easily calculated by the user from local gage data (hourly or daily) or looked up from a local reference. 3. The lookup database of pre-packaged simulation results was indexed both by the absolute storage volumes associated with each simulation (precipitation depths, inches) as well as the storage volume as a fraction of the 85th percentile, 24-hour storm depth at the gage location that was modeled (unitless fraction). 4. The storage volume provided in the VRAs at the project is divided by the user-entered 85th percentile, 24-hour storm depth for their project location (if provided) to compute the normalized storage volume (fraction of the 85th percentile storm depth, unitless) for the model scenario. The normalized storage volume is then used to query the lookup database to estimate capture efficiency. A-8

For example: 1. The 85th percentile, 24-hour depth at the climate division gage that is used to represent the overall division in the modeling is 0.95 inches. 2. The 85th percentile, 24-hour depth at the actual project location is estimated to be 0.80 inches based on the user’s analysis of precipitation records in the vicinity of the site. 3. The user enters the 0.80 inches into the field in the tool that allows for location-specific override. 4. A project proposes to construct a VRA that has a storage volume equivalent to the runoff from a 0.6 inch storm event. 5. Therefore, the proposed VRA has a normalized unitless storage volume of 0.75 (fraction of the 85th percentile, 24-hour storm depth = 0.6 inches / 0.8 inches). 6. The results of the pre-packaged model simulations corresponding to a normalized unitless storage volume of 0.75 are used to estimate the capture efficiency of the project-specific VRA (interpolation is required if an exact match is not returned). In comparison, if “localization” was not implemented, the tool would look up the model result for a normalized unitless storage volume of 0.63 (0.6/0.95), which would tend to underestimate the actual performance at the project location. 4 Modeling Methodology and Parameters for Tool Development 4.1 Types of Modeling Conducted Three sets of modeling runs were completed in the EPA Storm Water Management Model (SWMM) Version 5.0.022 to develop the underlying “lookup databases” to support the tool. This section describes the inputs to the SWMM model and the model increments that were used in these supporting model runs. • Consistent drawdown runs: Consistent drawdown runs were used to represent VRA compartments that can be approximated as draining at a relatively consistent rate throughout a long term continuous simulation (e.g., infiltration, media filtration, orifice discharge). The template model setup developed for these runs included a tributary subcatchment draining to a storage unit of a given size (varied between runs) modeled with a drawdown rate (varied between runs) that was held constant throughout each simulation. Continuous rainfall-runoff processes were simulated to estimate the continuous runoff hydrograph. Routing through the storage unit was simulated to estimate the long term capture efficiency associated with the given configuration. • ET drawdown runs: ET runs were used to represent VRA compartments that drain via ET processes, at rates that inherently vary with climatic factors throughout the year. The template model setup developed for these runs included a tributary subcatchment draining to a storage unit of a given size (varied between runs) modeled with a given stored water depth (varied between runs) that was drawn down at the applied ET rate (varies on a monthly basis and between locations). Continuous rainfall-runoff processes were simulated to estimate the continuous runoff hydrograph. Routing through the storage unit was simulated to estimate the long term ET loss associated with the given configuration. A-9

• Dispersion runs: Dispersion runs were used to represent VRA types that cannot be simply divided into compartments because water is dispersed in a thin layer and is acted upon by both infiltration and ET processes. The template model setup developed for these runs included a tributary subcatchment draining to two broad, shallow storage units in series (area varied between runs to represent different proportions of pervious area receiving dispersion). The first storage unit was used to represent water stored in the “suction storage” of soil pores that did not freely drain via gravity. This was filled first and was drawn down at the rate established by ET inputs. This storage unit also received flow from a “dummy catchment” with 100 percent imperviousness and zero depression storage; effectively representing precipitation directly on the dispersion area. The second storage unit had the same footprint as the first storage unit (i.e., equal to the size of the dispersion area) and received flow when the first storage unit overflowed. These storage units were effectively “stacked” in the model. This storage unit represented the freely drained pore storage (i.e., drained by gravity) in the amended media and any surface ponding in closed depressions. This storage unit was drained via Green-Ampt infiltration processes based on the assigned infiltration parameters (varied between runs). The depth of stored water in the first and second storage compartments was calculated based on the assumed depth of soil amendments (varied between runs) and typical amended soil properties. Continuous rainfall-runoff processes were simulated to estimate the runoff hydrograph. Routing through the storage units was simulated to estimate the long term capture efficiency associated with the given configuration. 4.2 SWMM Model Inputs A consistent set of SWMM inputs were used for each of the model runs described in Section 4.1. 4.2.1 Precipitation Data Hourly precipitation datasets from the National Climatic Data Center (NCDC) were obtained for all available gages in the conterminous United States (approximately 6,700 gages). These gages were analyzed to estimate the period of record of observations, the fraction of the record that was missing or qualified, and the resolution of precipitation depth measurements. For each climate division (defined by NOAA, see Figure 4), a precipitation gage was selected that had at least 30 years of precipitation data, less than 5 percent missing or qualified data, and precipitation depth resolution of 0.01 inches, where possible. Where no gage within the climate division met these criteria, the gage that most closely met these criteria was selected. Section 9 lists the precipitation gages that were selected. Precipitation datasets were used directly as inputs to the EPA SWMM model. A-10

Figure 3. NCDC Climate Divisions 4.2.2 Evapotranspiration Data Reference potential evapotranspiration data were obtained from the United States Geologic Service Oak Ridge National Laboratory (Vogel and Sankarasubramania 2005). The dataset has 40 years (1950 to 1990) of derived monthly ET estimates for 1,469 stations in the conterminous US. The closest available gage to each of the selected precipitation gages was selected from this dataset. Reference potential ET data were averaged for each month to estimate monthly normal potential ET valves. These monthly normal values were used in SWMM simulations, with an adjustment factor of 0.7 to account for actual potential ET being typically less than the reference potential ET for most cover types. For VRAs that are sensitive to ET rates, the user has the option to set the ET adjustment factor (also referred to as “crop coefficient”) to reflect the plant palette that is proposed. A-11

4.2.3 SWMM Hydrologic Parameters Table 1 contains the parameters that were used in the SWMM simulation of rainfall-runoff. Table 1. SWMM Runoff Generation Parameters SWMM Runoff Parameters Units Values Source/Rationale Wet time step minutes 15 Standard assumption when using hourly precipitation inputs Dry time step hours 4 Standard assumption when using hourly precipitation inputs Routing time step seconds 120 Balance between stability and runtimes. Precipitation Time Resolution Hours 1.0 Hourly precipitation data available at each gage, mostly at 0.01 inch depth resolution. Period of Record years 1/1/1980 to 12/31/2009 NOAA National Climatic Data Center. Evapotranspiration in/ month Varies by location USGS Oak Ridge National Laboratory (Vogel and Sankarasubramanian, 2005) Area acres 1.0 Typical of smaller catchments in urban highway environment; not a sensitive parameter for volume reduction or capture performance of VRAs from 0.1 to more than 50 acres. This lack of sensitivity was demonstrated in model results from NCHRP 25-20(01). Imperviousness % 100 Simulations were conducted with a range of imperviousness to bracket the range that may be present in tributary areas. This was used to develop an adjustment method such that the results of 100% impervious catchments could be used to represent the performance of VRAs treating catchments less than 100% impervious (See Section 7). Characteristic Flow Path Length ft 86 86 ft path length; 500 ft width, per assumptions used in NCHRP 25-31. Not significantly sensitive for analysis of volume reduction processes. Overland Slope ft/ft 0.02 Represents typical cross slope on roadways; varies by site. Consistent with assumptions used in NCHRP 25-31. Not significantly sensitive for analysis of volume reduction processes with hourly precipitation inputs. Depression storage, impervious inches 0.05 ASCE 1992 (low range of estimate for impervious surface) Depression storage, pervious inches 0.1 ASCE 1992 (low range of estimates for lawns; characteristic of highway embankments, usually greater slope than lawns) Impervious Manning’s n 0.012 McCuen et al. 1996 – Based on concrete (typical of concrete and asphalt wearing coarse) A-12

SWMM Runoff Parameters Units Values Source/Rationale Pervious Manning’s n 0.15 McCuen et al. 1996 – Based on short grass Infiltration Model NA Green-Ampt Based on availability of parameters and supporting documentation. Saturated Hydraulic Conductivity in/hr Varies by soil texture class, see Table 2 below. Suction Head inches Varies by soil texture class, see Table 2below. Initial Moisture Deficit fraction Varies by soil texture class, see Table 2 below. Table 2. SWMM Green-Ampt Infiltration Parameters SWMM Runoff Parameters Units Soil Texture Class Groupings Source/Rationale Sand – Sandy Loam Sandy Loam – Sandy Clay Loam Sandy Clay Loam – Silty Clay Clay Assumed Texture Class- > (Rawls et al. 1983) Loamy Sand Silt Loam Sandy Clay Loam Clay Hydraulic Conductivity (mid-point and range) in/hr 1.2 (0.43 – 4.7) 0.26 (0.06 to 0.43) 0.04 (0.02 to 0.06) 0.01 (0 to 0.02) Rawls et al. 1983, based on assumed texture classes Suction Head inches 2.4 6.7 8.7 11.4 Rawls et al. 1983, based on assumed texture classes Initial Moisture Deficit fraction 0.36 0.29 0.21 0.16 Rawls et al. 1983, based on assumed texture classes (Porosity – Avg (Field Capacity, Wilting Point)) 4.2.4 Storage Volume Increments For each precipitation gage, storage volume increments were defined by 10 standard multiples of the 85th percentile, 24-hour storm depth for that gage (unitless) that bracket the range of typical VRA sizing: ⇒ 0.1 ⇒ 0.2 ⇒ 0.4 ⇒ 0.6 ⇒ 0.8 ⇒ 1.0 ⇒ 1.25 ⇒ 1.5 ⇒ 2.0 ⇒ 3.0 For each model run, the 85th percentile, 24-hour storm depth was known at the modeled gage. For each configuration, the storage volume was calculated using the equation: V (cu-ft) = 1.0 acre × 43,560 sq-ft/ac × Volumetric Runoff Coefficient × 85th pctl Storm Depth (inches) × Fraction of 85th pctl Storm Depth /12 inches/ft The volumetric runoff coefficient used for the purpose of computing storage volume was assumed to be 0.9 for full impervious catchments. A-13

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 4.2.5 Drawdown Increments For compartments of VRAs that can be approximated by a consistent drawdown time (i.e., drained by infiltration or by a surface discharge outlet), model increments were defined by 10 standard drawdown times that are relevant for most VRAs (hours): ⇒ 2 ⇒ 3 ⇒ 6 ⇒ 12 ⇒ 24 ⇒ 48 ⇒ 72 ⇒ 120 ⇒ 180 ⇒ 360 The tool is not intended for VRAs with very rapid drain times or very long drain times. Increments were developed based on our experience with the spacing of increments needed to obtain reliable results. Drawdown times were implemented in the model by setting the discharge rate from the SWMM storage unit (Q) to: Q (cfs) = V (cu-ft) / (Drawdown Time (hours) × 3600 sec/hr) 4.2.6 ET Depth Increments For compartments of VRAs that hold water and are drained primarily by ET, the “ET depth”, defined as the amount of ET that must occur for the ET storage to be substantially recovered, were used as model increments. The following ET depth increments were modeled (inches): ⇒ 0.5 ⇒ 0.75 ⇒ 1 ⇒ 2 ⇒ 3 ⇒ 5 ⇒ 10 The low range was based on 3 inches of amended soil media with typical ET storage capacity (approximately the difference between the field capacity and the wilting point). The upper range was based on cistern that captures enough water to meet irrigation demand for 10 inches worth of ET over the area to be irrigated. 4.2.7 Dispersion Model Runs Where VRAs operate primarily via dispersion of water over pervious areas (i.e., filter strips, vegetated swales) the concept of a “storage volume” and “drawdown time” can be difficult for the user to estimate from planning level design parameters. A separate lookup database was developed based on (1) the ratio of pervious area receiving flow to impervious area contributing flow, (2) the underlying soil infiltration rate of the receiving pervious area, and (3) the depth of amended soils. Increments of these parameters were simulated to develop the lookup database as shown in Table 3. A-14

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Table 3. Increments of Input Parameters for Dispersion Runs Ratio of Impervious Area Draining to VRA to Pervious Area Receiving Dispersion Soil Properties of Underlying Soils Below Amended Soil in Dispersion Area Depth of Soil Amendment or Decompaction, inches Saturated Hydraulic Conductivity, in/hr Initial Moisture Deficit, inches/ inch Suction Head, inches 50 0.00 0.16 11.4 0 20 0.01 0.16 11.4 3 10 0.05 0.21 8.7 6 5 0.10 0.21 8.7 12 2.5 0.20 0.29 6.7 24 1.0 0.50 0.29 6.7 0.5 1.00 0.36 2.4 0.25 2.00 0.36 2.4 4.2.8 Summary of Supporting Model Runs Table 4 provides a summary of the supporting model runs that were executed to provide the back-end database to support the tool. Sensitivity testing indicated that these increments provide reliable basis for interpolation. Table 4. Summary of Supporting Model Runs Parameter Number of Increments Consistent Drawdown Model Runs (Infiltration, Surface Discharge) Climate Divisions 344 Modeled Imperviousness of Tributary Area 1 Supported Imperviousness 0 to 100% (continuous scale; more reliable above 50%) Modeled Soil Type Not Applicable (100% impervious) Supported Soil Type User can select between 4 soil texture classes or enter a user-defined soil infiltration rate within the range supported. Storage Volume 10 Drawdown Time 10 Total – Consistent Drawdown Runs 34,400 ET Drawdown Model Runs Climate Divisions 344 Modeled Imperviousness of Tributary Area 1 Supported Imperviousness 0 to 100% (continuous scale; more reliable above 50%) Modeled Soil Type Not Applicable (100% impervious) Supported Soil Type User can select between 4 soil texture classes or enter a user-defined soil infiltration rate within the range supported. Storage Volume 10 ET Depth Increments 7 Total – ET Runs 24,080 A-15

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Parameter Number of Increments Dispersion Model Runs Climate Divisions 344 Ratio of Impervious to Pervious 8 Soil Properties below Amended Soil 8 Amended Soil Depth 5 Total – Dispersion Model Runs 110,080 Total Number of Model Runs 168,560 Model runs were executed using macros that automate the development of SWMM files and execute these files in batch simulation model. Computationally it would have been possible to conduct more runs, however the size of the lookup database would increase and potentially create excessive file sizes for the Tool. Additionally, the opinion of the research team was that additional model runs would not substantially improve the reliability of estimates. However, if an individual DOT wanted to develop a customized version of the spreadsheet for their state, additional densities of meteorological stations could be run and/or additional more refined increments, etc. Each SWMM run produced a summary report file (.rpt) that contained summary results of the simulation. Among many reported parameters, the rpt file included a summation of the volume that enters the storage element and overflows the storage element, which can be used to determine the baseline runoff (without controls) and the average annual percent capture for that scenario. A simple Excel Macro was customized to “mine” the relevant data from each of the rpt files and populate a database of summary results where each line of the database describes the model run and includes the summary output statistics. 5 Summary of Tool Calculations The tool conducts a number of key calculations to translate the results of the SWMM model runs to estimates of long term volume reduction. This section provides a summary of calculations. Full details of calculations can be inspected via review of Excel formulas and macros. • Based on user inputs about project location, the precipitation summary statistics are accessed from a precipitation data lookup table in the tool. The 85th percentile, 24-hour storm depth and the average annual storm depth are later used in tool calculations. • User inputs about the tributary area to a VRA are used to estimate baseline average annual runoff volume. They are also later used to normalize the VRA design parameters relative to the catchment and precipitation parameters. • VRAs are selected by the user from a drop down menu. This loads an input form that is unique to the selected VRA type. Each input form contains the design parameters, default parameters, and the associated calculations. Guidance is provided within the tool for specifying input parameters. Intermediate calculations are hidden from the user. Each VRA template has a unique set of rules about the compartments it contains, the order in which these compartments are analyzed, and any checks or limits that are enforced between compartments. • Each VRA template has a unique set of calculations embedded within it to translate VRA design parameters into underlying lookup indices that can be used to obtain values from the lookup database. For example, these calculations perform functions such as: A-16

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas o Estimate the volume of each compartment based on the design parameters specified by the user and the default parameter assumptions. o Normalize the volume of each compartment in terms of the runoff volume produced from an equivalent storm depth over the tributary area. This calculation involves both the volumes of each compartment as well as the tributary area parameters. This is the step where the effects of VRA design, catchment size, imperviousness, and soil type are each considered: when a catchment is smaller, less impervious, and/or has soils with less runoff potential, a VRA compartment of a given volume is equivalent to the runoff from a larger storm event, thereby achieves higher capture efficiency, and vice versa. o Normalize the equivalent storm depth as a fraction of the local 85th percentile, 24-hour precipitation depth. This is the step where the effect of the local precipitation patterns are considered. When the local 85th percentile, 24-hour storm is larger, a VRA compartment of a given volume is smaller as a fraction of the runoff from this event, thereby achieves lower capture efficiency, and vice versa. o Calculate the drawdown parameters for each compartment (drawdown time, ET depth) based on specified parameters such as infiltration rate, depth of compartments, etc. o Calculate intermediate information about the specified VRA design, such as calculated VRA footprint, to provide feedback to the user and support iterative use of the Tool. • After the lookup indices have been calculated (e.g., compartment volume as a the runoff volume from a certain fraction of the 85th percentile, 24-hour storm volume), two-dimensional interpolation algorithms are applied to obtain estimates from the lookup database for the exact combination of lookup indices (i.e., interpolation on two key indices within the lookup table). For compartments where some amount of volume reduction has already been provided in a previous compartment or VRA in a treatment train, these lookup/interpolation algorithms also account for “upstream” capture by implementing the algorithm described in Section 3.5. • In some cases, two different design elements have the potential to control the performance. For example, in a bioretention area with underdrains, capture efficiency can be controlled by either the surface ponding volume and the rate at which water enters the amended media at the surface or by the total volume VRA volume and the rate at which water leaves the system as a whole. In cases such as this, both calculations are conducted to determine which computation controls. • Validation checks are conducted to provide warnings when the specified set of design parameters and defaults results in lookup indices that are out of the bounds of the lookup database. Reference calculations are also conducted for key parameters to provide recommended ranges for design parameters so that the lookup indices remain within the supported range. • Composite capture efficiency and volume reduction performance of the VRA or treatment train is tabulated, and these percentages are applied to the baseline runoff volume to calculate long term average quantities of volume reduced, volume treated, and volume bypassed (cu-ft per year). A-17

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas • A sensitivity analysis interface allows the user to specify a high and low bound for up to three key parameters per VRA. An underlying sensitivity analysis algorithm is then executed to cycle through the tool calculations for the high and low bounds of each specified sensitivity parameter, holding the other specified parameters fixed at their original assigned value. 6 Simplifying Assumptions and Reliability A number of simplifying assumptions were made as part of developing the tool. These are summarized below: 1. General rainfall patterns are assumed to be approximately homogenous within a climate division. While scaling to localize model estimates is recommended within climate divisions, as discussed above, and should significantly improve site-specific estimates, this approach assumes that the overall patterns of rainfall are relatively similar within a climate division such that the overall trends developed from modeling at a single gage location is representative throughout the climate division that the gage represents. The use of the 85th percentile, 24-hour storm depth as a basis for scaling has been demonstrated to be reliable in several areas (OCPW 2011; VCWPD 2011). Additional analysis was conducted as part of developing this tool (described in Section 7) to demonstrate that this approach is generally reliable across the contiguous US. 2. An approximate method was used to isolate and account for multiple VRA compartments instead of explicit modeling of all potential VRA configurations. The graphical method of summing multiple compartments of a VRA has been demonstrated to provide reliable results versus more explicit representation of VRAs (OCPW 2011; University of Missouri 2012). The validity of this approach has been evaluated against more explicit representations as part of this project (described in Section 7) and was found to provide reasonably reliable results. 3. Each VRA type was assumed to follow a prescribed drawdown pattern that does not change over the duration of the model or as a function of water depth/hydraulic head in the VRA. Time- variable factors such as temperature effects on infiltration rate and decline in infiltration rate as a result of clogging are not represented by the model. However, guidance is provided for selecting design infiltration rates that reflect the average long term conditions, accounting for seasonal variability in infiltration rates and inevitable declines in infiltration rates with time. 4. Snowfall and snowmelt were not simulated as part of developing lookup tables. A water equivalent approach was used where all precipitation was assumed to fall as liquid rain. While this assumption may introduce error in some climates, there are limited portions of the contiguous US that receive more than 10 to 20 percent of annual precipitation by snowfall. Failure to account for snowfall/snowmelt would potentially introduce considerable errors in peak flow estimation in some areas (i.e., rain on snow events); however, cumulative long term errors in runoff volume introduced for this portion of precipitation would generally have a relatively minor influence on long term volume reduction estimates. In some cases, rain on snow events could overwhelm a VRA in a way that would not be predicted by a simple water equivalent model. In other cases, snow pack could effectively serve as “detention” storage and would allow a greater amount of runoff to be captured (and then volume lost) than would be predicted by using the water equivalent assumptions. Therefore errors may approximately balance for purposes of estimating volume reduction. A-18

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 5. Model scenarios were based on discrete increments of input parameters, and interpolation is required between entries in the lookup database. This approach assumes that response is approximately linear between discrete model scenarios. This assumption is generally reliable with appropriate selection of simulation increments to provide a greater number of increments where the response is most non-linear. 6. VRAs are assumed to follow an ordered filling and draining regime, where ET storage is assumed to fill first, followed by the lowest freely drained storage elements, followed by storage compartments located at a higher elevation. In essence, a VRA fills from the bottom to the top after satisfying ET storage. For the majority of storm events, this assumption is representative. However, under peak runoff conditions, it may be possible for upper storage elements to fill prior to lower storage elements because of limitations on the rate at which water is conveyed between elements. For example, in the case of a bioretention area with elevated underdrains, it may be possible for the surface ponding area to fill to a point of overflow before water can flow through the amended media to entirely fill the underdrain sump. Where this condition is possible, a secondary check is provided in the tool to determine if it controls the capture efficiency calculations. 7. Hourly precipitation records were used. For small catchments, these records are known to mask peak short term runoff rates, which occur as a response to sub-hourly, high intensity precipitation. This may be especially important for peak event overflows of VRAs. However, long term volume reduction performance is a function of processes that occur at a longer time interval than short- interval peak intensities and the cumulative total of many smaller events as well as less frequent larger events. Additionally, most VRAs include volumetric storage capacity that tends to equalize short-duration peak runoff rates. Therefore the use of hourly precipitation inputs is considered reliable for assessment of volume reduction performance of volume-based VRAs. 8. Estimates are derived from model runs of 100 percent impervious catchments. By developing a distinct runoff coefficient equation (runoff coefficient as a function of impervious cover) for each soil type, the normalized results of performance simulations conducted for 100 percent impervious catchments are translated to represent actual catchments that are less than 100 percent. For example, with correction for runoff coefficient in the application of the lookup database, the following example design scenarios (and many others) are represented by the same continuous simulation results: • A VRA sized at 3,000 cu-ft receiving runoff from 2 acre catchment at a runoff coefficient of 0.45 is equivalent to the runoff from a 0.91 inch storm event [(3,000 cu-ft × 12 in/ft) / (2 ac × 43,560 sq-ft/ac × 0.45) = 0.91 inches]. • A VRA sized at 6,000 cu-ft receiving runoff from a 2 acre catchment with a runoff coefficient of 0.9 is equivalent to the runoff from a 0.91 inch storm event [(6,000 cu- ft × 12 in/ft) / (2 ac × 43,560 sq-ft/ac × 0.90) = 0.91 inches]. This simplification is necessary to reduce the number of scenarios that must be simulated to develop the lookup database and maintain a reasonable file size for the tool. While minor non- linearity may be expected between capture efficiency and runoff coefficient, the selection of an appropriate runoff coefficient equation allows this simplification to be acceptable for planning level purposes. This simplification was evaluated (see Section 7) and found to be generally reliable. A-19

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Overall, these simplifying assumptions are expected to have a relatively minor impact on the model results and are considered to be reliable for planning level estimates of volume reduction performance. As with any modeling tool, a major source of error can arise from the user entering erroneous input parameters or entering input parameters that do not reflect realistic design scenarios. 7 Supplemental Case Study Analyses to Evaluate Reliability of Tool Methodology Section 6 identified several simplifying assumptions for which further evaluation of reliability was deemed necessary. As part of early tool development and validation efforts, the research team performed focused analyses to test the reliability of these simplifying assumptions. For each simplifying assumption, example analyses were conducted explicitly (i.e., without simplification) and the results of these analyses were compared to what would be obtained using the simplification that was proposed. The simplifying assumptions that were evaluated include: Simplifying Assumption 1: Normalized results from performance simulations conducted for 100 percent impervious catchments can be used to represent catchments with different imperviousness and soil types via post-processing methods. This allows for fewer model permutations to be run and reduces the file size of the tool. Simplifying Assumption 2: The 85th percentile, 24-hour storm depth can be used as a normalizing and scaling factor to “localize” model results to account for intra-climate division variability as described in Section 3.6. This provides improved spatial resolution by allowing reliable scaling within the influence area of each precipitation gage that was modeled. Simplifying Assumption 3: The graphical method presented in Section 3.5 can account for the non-linearity of capture response observed in multi-compartment VRAs and treatment trains and reliably approximate the performance of these types of systems. This allows the effects of standard single-compartment runs to be summed after model runs are complete, rather than running models for all potential configurations of multi-compartment or treatment train VRAs. This significantly reduces the number of modeling runs required and reduces the file size of the tool. Simplifying Assumption 4: Results of evapotranspiration (ET) runs using monthly normal ET data are comparable to model estimates developed using a daily time series of ET data and therefore monthly data can be utilized. To test the validity of these assumptions, three clusters of precipitation gages were selected to represent distinctly different climate zones across the country. Within each cluster, gages were selected to represent distinctly different local geographic settings that could lead to different VRA performance (i.e., elevation, topography, distance from major water bodies). Precipitation gages used in the analysis are shown in Figure 4. A-20

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Figure 4. Precipitation gages used for testing the reliability of simplifying assumptions (with NCDC climate division boundaries shown) 7.1 Evaluation of Simplifying Assumption 1 7.1.1 Hypothesis Our hypothesis was that the results of performance simulations conducted for 100 percent impervious catchments can be translated to represent actual catchments that are less than 100 percent by developing a distinct event-based runoff coefficient equation for each soil type. The runoff coefficient equation would be used in the “backend” of the tool to estimate the “equivalent precipitation depth” over the watershed that would produce a runoff volume equivalent to the VRA storage volume. In other words, we hypothesized that two catchments with varying characteristics (imperviousness, soil type, area) that have the same “equivalent precipitation depth” will achieve approximately the same capture efficiency. For example, with correction for runoff coefficient in the application of the lookup database, the following example design scenarios (and many others) could be represented by the same continuous simulation results: • A VRA sized at 3,000 cu-ft receiving runoff from 2 acre catchment at a runoff coefficient of 0.45 is equivalent to the runoff from a 0.91 inch storm event [(3,000 cu-ft × 12 in/ft) / (2 ac × 43,560 sq-ft/ac × 0.45) = 0.91 inches]. • A VRA sized at 6,000 cu-ft receiving runoff from a 2 acre catchment with a runoff coefficient of 0.9 is equivalent to the runoff from a 0.91 inch storm event [(6,000 cu-ft × 12 in/ft) / (2 ac × 43,560 sq-ft/ac × 0.90) = 0.91 inches]. The form of the runoff coefficient equation proposed for the purpose of computing the “equivalent precipitation depth” is: A-21

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Runoff Coefficient = m × imp + b Where: imp = impervious cover as a fraction from 0 to 1; m and b are coefficients ranging from 0 to 1. Different coefficients were developed for different soil types. 7.1.2 Analyses In order to test this hypothesis, consistent drawdown runs were performed for combinations of (i) 4 soil types; (ii) 5 impervious fractions; (iii) 10 storage volume increments and (iv) 10 drawdown increments (2,000 runs for each gage) using the SWMM parameters listed Section 4 for the following three precipitation gages: (i) Washington D.C. Reagan Airport (COOP ID: 448906); (ii) Los Angeles International Airport (COOP ID: 45114) and (iii) Portland International Airport (COOP ID: 356751). Model results allowed us to compare between (1) the percent capture achieved based on modeling of explicit combinations of imperviousness and soil type, and (2) the percent capture achieved for each distinct combination using the hypothesized approach based on 100 percent imperviousness runs normalized the using the runoff coefficient equation. Coefficients m and b were developed for each soil type by minimizing the sum of range of percent capture estimates for five “equivalent precipitation depths” (0.2”, 0.5”, 1”, 1.5” and 2”) using m and b as variables (solver function in Excel). Each combination of “equivalent precipitation depth” and VRA drawdown time had a range of percent capture calculated as the difference between maximum and minimum percent captures from 25%, 50%, 75% and 100% impervious catchment runs. In other words, where the range was zero, the approach worked perfectly to normalize for different soil types and imperviousness. Where the range was greater, the normalization routine did not account for some aspects of the response. Coefficients estimated in this analysis for Washington D.C. gage are shown in Table 5. The reliability of the developed coefficients for a different precipitation gage was tested by using the coefficients developed using Washington D.C. gage in the runs conducted for Los Angeles and Portland gages. The average range of capture efficiency observed at each gage for each soil type (across combinations of 5 equivalent precipitation depths and 10 drawdown increments = 50 total data points) is presented in Table 5. Table 5. Analysis Performed to Test Simplifying Assumption 1 Based on Washington D.C. Runs Average (Max) Range of Differences in Capture Efficiency1 Soil Type m b Washington D.C. Los Angeles Portland Loamy Sand 0.90 0.00 0.1% (0.7%) 0.1% (0.3%) 0.1% (0.2%) Silt loam 0.86 0.04 1.9% (3.5%) 2.0% (4.5%) 2.3% (5.1%) Sandy Clay Loam 0.37 0.53 3.6% (10%) 3.4% (6.3%) 5.5% (20%) Clay 0.15 0.75 1.5% (6%) 1.2% (4.2%) 3.3% (11%) 1 – Range of differences in capture efficiency is the absolute difference between min and max percent capture (after normalization) for a given precipitation depth, drawdown time, and given soil type, based on a range of model runs from 25 to 100 percent imperviousness. For example, if the 25 percent impervious run yielded 45 percent capture after normalization and the 100 percent impervious run yielded 48 percent capture after normalization, then the reported range for that combination of inputs would be 3 percent. Note: Runoff coefficient equation m and b are not intended to be used to yield estimates of long term runoff coefficient; they have been developed specifically to provide a “best fit” for this application. A-22

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 7.1.3 Summary of Findings Based the summary presented in Table 5 and evaluation of individual runs (not reported), average ranges of capture efficiency differences observed in the analysis for Washington D.C. are relatively small and are generally within the anticipated uncertainty associated with selection of model input parameters. When runoff coefficient formulae are transferred to other regions, maximum differences tended to increase slightly. In all three regions, results of model runs with lower imperviousness (25 percent) tended to control the ranges (i.e., tended to be most different from the mean), while results from 50, 75, and 100 percent imperviousness runs tended to be more closely spaced. For low imperviousness, where errors appear to be greatest, the simplification resulted in under-prediction of capture which is considered to be conservative. These errors are generally within the range of certainty in model inputs and highway catchments tend to have imperviousness well above 50 percent. Overall the post-processing approach for addressing differences in imperviousness performs relatively well and tends to result in a minor underestimate of capture for lower imperviousness watersheds. This validates the usage of this simplifying assumption. 7.2 Evaluation of Assumption 2 7.2.1 Hypothesis Our hypothesis was that the 85th percentile, 24-hour precipitation depth can be used to localize model results as described in Section 3.6. 7.2.2 Analysis In order to test the “localization” approach presented in Section 3.6, a consistent drawdown run was modeled with 10 storage increments and 10 drawdown increments (100 runs) for 9 precipitation gages from three climate divisions (3 from each climate division; shown in Figure 4 and Table 6). A one acre 100 percent impervious catchment was used as the contributing watershed for these runs. The results of model runs from each of the 3 rain gages within each of the three clusters were first extracted, then were compared to the estimate that would have been obtained had only the first gage in each cluster been run and results had normalized to the other gages based on the relative ratios of the 85th percentile, 24-hour precipitation depths (as proposed in Section 3.6). The range of capture efficiencies indicates how much error is introduced via the localization approach versus explicit runs of each gage. The average range of capture efficiencies estimated from the analysis for the three gages in each climate division from the 100 runs is presented in Table 6. A-23

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas Table 6. Analysis Performed to Test Simplifying Assumption 2 Climate Division Coop Id Station Name 85th Percentile (in) Elevatio n (ft) Average (Max) Range of Differences in Capture Efficiency1 4404 448906 WASHINGTON REAGAN AP 0.94 33 1.6% (4.2%) 448046 STAR TANNERY 0.90 3117 446712 PIEDMONT RSCH STN 1.10 1706 3502 356751 PORTLAND INTL AP 0.63 62 2.2% (10%) 353705 HASKINS DAM 1.25 2480 352345 DISSTON 1 NE LAYING CK 0.82 3996 0406 45114 LOS ANGELES INTL AP 1.02 318 2.6% (14%) 47740 SAN DIEGO WSO AP 0.78 49 46162 NEWHALL S FC32CE 1.70 4045 1 – Range of differences in capture efficiency is the difference between the percent capture from each gage-specific analysis and the percent capture estimated by “localizing” results from the first gage in each cluster for each combination of equivalent precipitation depth, drawdown time. 7.2.3 Summary of Findings Based on the summary in Table 6 and evaluation of individual runs (not reported), the localization approach appears to provide a reliable basis on average for scaling within each of the climate divisions. We do not see reason to believe that a significantly better localization, on average, would be achieved using a different percentile rainfall depth. Maximum differences occur where drawdown times are very short (2 to 3 hours) and storage volumes are very small (0.1 to 0.3 inch equivalent storm depths). 7.3 Evaluation of Simplifying Assumption 3 7.3.1 Hypothesis Our hypothesis was that the graphical approach for summing multiple compartments and treatment trains presented in Section 3.5 is reliable for applications of the tool to quantify percent capture and percent volume reduction provided by typical VRA configurations that have multiple compartments. 7.3.2 Analysis In order to test the approach presented in Section 3.5, the following two analyses were performed and then the results from each analysis were compared. Both “Analysis 1” and “Analysis 2” (described below) were performed for the following three precipitation gages: (i) Washington D.C. Reagan Airport (COOP ID: 448906); (ii) Los Angeles International Airport (COOP ID: 45114) and (iii) Portland International Airport (COOP ID: 356751). A one acre 100 percent impervious catchment was selected in the analyses as the contributing watershed. Analysis 1 – Explicit Representation of Multi-Compartment VRAs: SWMM was used to estimate composite capture efficiency for a two compartment VRA. The model in this analysis was set up such that the first compartment receives runoff from the catchment and the second compartment receives overflow A-24

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas from the first compartment (or fills when the capacity of the first is exceeded). Capture efficiency in this analysis was estimated using the equation: Capture efficiency = 1 - (Combined overflow leaving system)/(Runoff) Two multi-compartment scenarios (A and B) were modeled as explicit configurations in SWMM. Increments used in the analysis are presented in Table 7. Scenario A has a constant drawdown compartment followed by another consistent drawdown compartment. This could be representative of a permeable shoulder reservoir that overflows to a bioretention area, or could be representative of a bioretention area with elevated underdrains that has two distinct storage compartments that fill in a pre-determined order. Scenario B has an ET compartment followed by a consistent drawdown compartment. This could represent a filter strip that flows to an infiltration trench, or a bioretention area where water is first intercepted in soil pores before filling the gravity storage elements of the system. Each scenario has a total of 256 runs for each precipitation gage. Table 7. Increments used in Explicit SWMM Runs Compartment 1 Compartment 2 Scenario A Storage Increment Drawdown (hrs.) Storage Increment Drawdown (hrs.) 0.2 12 0.2 12 0.6 24 0.6 24 1 48 1 48 2 72 2 72 Scenario B Storage Increment ET Depth (in) Storage Increment Drawdown (hrs.) 0.2 0.5 0.2 12 0.6 1 0.6 24 1 2 1 48 2 5 2 72 Analysis 2 – Graphical Method: The same two scenarios were evaluated using the lookup database generated from runs with single storage compartments by applying the graphical method described in Section 3.5. The results of the graphical method (Analysis 2) were compared to the results of the explicit representation (Analysis 1). The average absolute difference in capture efficiency between the two analyses for each precipitation gage is presented in Table 8. Table 8. Analysis Performed to Test Simplifying Assumption 3 Absolute Difference in Capture Efficiency – Average (Max) of 256 Parallel Model Runs Scenario A Scenario B Washington D.C. 0.5% (2%) 0.7% (4.7%) Los Angeles 0.5% (2.4%) 0.7% (3.4%) Portland 0.8% (3.3%) 0.7% (4.1%) A-25

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 7.3.3 Summary of Findings The analysis performed to test the graphical method for two typical VRA scenarios supports this simplifying assumption and validates the usage of the graphical approach presented Section 3.5. The relative error produced using this approach appears to be much smaller than the uncertainty in model inputs (based on the summary in Table 8 and evaluation of individual runs, not reported). Greater error may be introduced in some configurations; additionally this approach does not account for the details of all routing configurations that could be used. However, conducting explicit simulations of all potential combinations of storage compartments and routing configurations could yield model runs in the range of millions or billions and would be computationally prohibitive. Alternatively, simply adding the capture efficiency of multiple compartments (rather than using the graphical method) is theoretically flawed. Therefore, this represents a reliable approach for balancing rigor with computational constraints. 7.4 Evaluation of Simplifying Assumption 4 7.4.1 Hypothesis Our hypothesis was that monthly normal ET data are adequate for model runs and provide a reasonable approximation of model results that would be obtained using a daily time series of data. 7.4.2 Analysis In order to understand the sensitivity of temporal resolution of ET data used in the simulations on model results, parallel scenarios were evaluated. Each scenario consisted of a compartment that holds water and drains primarily by ET (storage equivalent to the runoff volume from the 24-hour, 85th percentile storm event with a 2-inch ponding depth that drains to ET only) and receives runoff from a 100 percent impervious 1 acre catchment. Scenarios were modeled using both monthly normal ET data and daily ET data for a 10 year record (10/1/1999 to 9/30/2009) for the following three precipitation gages: (i) Washington D.C. Reagan Airport (COOP ID: 448906); (ii) Los Angeles International Airport (COOP ID: 045114) and (iii) Portland International Airport (COOP ID: 356751). Results from this analysis are summarized in Table 9 below. Table 9: Estimates of Capture Efficiency with Usage of Different Temporal Resolution ET Data Total Runoff (10^6 gallon) Capture Efficiency Daily ET data Monthly Normal ET data % Difference Daily ET data Monthly Normal ET data Difference in Capture Washington D.C. 9.73 9.38 3.6% 39% 40% -1.0% Los Angeles 2.47 2.41 2.3% 49% 51% -2.2% Portland 7.35 7.06 3.8% 20% 21% -0.8% A-26

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 7.4.3 Summary of Findings Simulations with monthly normal ET data generated slightly smaller runoff quantities and slightly greater capture percentages than simulations with daily ET data. However, considering the small differences in predicted estimates and the simplicity offered by the use of monthly normal ET data, monthly normal data was considered most appropriate for this analysis. 7.5 Summary Based on the analyses performed, the simplifying assumptions that were used are expected to have a relatively minor impact on the model results and are considered to be reliable for planning level estimates of volume reduction performance. The greatest errors are typically encountered in watersheds with lower imperviousness and for design configurations where drawdown times and equivalent storm event storage volumes are small (i.e., systems are more sensitive to short-duration intensities/volumes). In general, these conditions are likely to be relatively rare in the highway environment for typical VRAs designs. More rigorous analytical tools are recommended in cases where a greater degree of control is needed over analyzing specific watershed or VRA details and/or where the simplifying assumptions of this Tool are not acceptable. 8 References American Society of Civil Engineers (ASCE) (1992). Design and Construction of Urban Stormwater Management Systems, New York, NY Los Angeles County Department of Public Works (LACDPW) (2012). Los Angeles County Water Resources, Precipitation Website. http://www.ladpw.org/wrd/precip/. Last accessed: 12/31/2012. McCuen, R., P. Johnson, and R. Ragan (1996). Highway Hydrology, FHWA-SA-96-067, Federal Highway Administration, Washington, DC National Climatic Data Center (NCDC) (2012). http://www.ncdc.noaa.gov/ Orange County Public Works (OCPW) (2011). Technical Guidance Document for the Preparation of Conceptual/Preliminary and/or Project Water Quality Management Plans (WQMPs). http://www.ocwatersheds.com/DocmgmtInternet/Download.aspx?id=638 Rawls, W., D. Brakensiek, and N. Miller (1983). ”Green‐ampt Infiltration Parameters from Soils Data.” J. Hydraul. Eng., 109(1), 62–70. University of Missouri (2012). Stormwater Quality Master Plan for the University of Missouri at Columbia. Prepared by Geosyntec Consultants. US Environmental Protection Agency (USEPA) (2009). Technical Guidance on Implementing Section 438 of the Energy Independence and Security Act of 2008. http://www.epa.gov/owow/NPS/lid/section438/pdf/final_sec438_eisa.pdf Ventura County Watershed Protection District (VCWPD) (2011). Ventura County Technical Guidance Manual for Stormwater Control Measures, July 13, 2011. www.vcstormwater.org/technicalguidancemanual.html A-27

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas 9 Precipitation Gages Supported by Tool State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches AL 1 NORTHERN VALLEY 014064 HUNTSVILLE INTL AP MADISON 624 1.24 2.00 54.9 AL 2 APPALACHIAN MOUNTAIN 010831 BIRMINGHAM AP ASOS JEFFERSON 615 1.20 1.93 45.1 AL 3 UPPER PLAINS 010063 ADDISON WINSTON 818 1.30 2.10 51.2 AL 4 EASTERN VALLEY 014209 JACKSONVILLE CALHOUN 688 1.20 1.92 45.6 AL 5 PIEDMONT PLATEAU 012124 DADEVILLE 2 TALLAPOOSA 733 1.30 2.00 51.7 AL 6 PRAIRIE 015550 MONTGOMERY AP ASOS MONTGOMERY 202 1.25 2.01 51.0 AL 7 COASTAL PLAIN 010140 ALBERTA WILCOX 175 1.40 2.20 55.2 AL 8 GULF 015478 MOBILE REGIONAL AP MOBILE 215 1.45 2.42 58.6 AR 1 NORTHWEST 032356 EUREKA SPRINGS 3 WNW CARROLL 1,420 1.20 1.91 43.2 AR 2 NORTH CENTRAL 032794 GILBERT SEARCY 620 1.18 1.95 42.9 AR 3 NORTHEAST 030458 BATESVILLE LVSTK INDEPENDENCE 571 1.30 2.10 45.9 AR 4 WEST CENTRAL 032574 FT SMITH RGNL AP SEBASTIAN 449 1.20 1.92 42.3 AR 5 CENTRAL 034248 LITTLE ROCK ADAMS FLD PULASKI 258 1.28 2.10 29.3 AR 6 EAST CENTRAL 036920 STUTTGART 9 ESE ARKANSAS 198 1.30 2.20 47.8 AR 7 SOUTHWEST 032810 GILLHAM DAM POLK 520 1.50 2.40 54.3 AR 8 SOUTH CENTRAL 030220 ARKADELPHIA 2 N CLARK 196 1.44 2.30 52.5 AR 9 SOUTHEAST 035754 PINE BLUFF JEFFERSON 215 1.40 2.20 49.3 AZ 1 NORTHWEST 024645 KINGMAN #2 MOHAVE 3,539 0.80 1.20 9.0 AZ 2 NORTHEAST 029439 WINSLOW AP NAVAJO 4,886 0.48 0.75 7.3 AZ 3 NORTH CENTRAL 020487 ASH FORK 3 YAVAPAI 5,075 0.69 1.10 12.3 AZ 4 EAST CENTRAL 026323 PAYSON GILA 4,850 0.90 1.50 19.3 A-28

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches AZ 5 SOUTHWEST 029660 YUMA WSO AP YUMA 206 0.58 1.11 2.9 AZ 6 SOUTH CENTRAL 026481 PHOENIX AP MARICOPA 1,107 0.67 1.02 7.2 AZ 7 SOUTHEAST 028820 TUCSON INTL AP PIMA 2,549 0.66 1.10 11.3 CA 1 NORTH COAST DRAINAGE 042910 EUREKA WFO WOODLEY IS HUMBOLDT 20 0.88 1.37 39.2 CA 2 SACRAMENTO DRNG. 040161 ALTURAS MODOC 4,400 0.50 0.72 12.3 CA 3 NORTHEAST INTER. BASINS 048873 TERMO 1 E LASSEN 5,300 0.49 0.80 10.0 CA 4 CENTRAL COAST DRNG. 047769 SAN FRANCISCO WSO AP SAN MATEO 8 0.89 1.39 20.2 CA 5 SAN JOAQUIN DRNG. 043257 FRESNO YOSEMITE INT'L FRESNO 333 0.68 0.97 10.9 CA 6 SOUTH COAST DRNG. 045114 LOS ANGELES INTL AP LOS ANGELES 97 1.02 1.60 12.3 CA 6 SOUTH COAST DRNG. 047740 SAN DIEGO WSO AP SAN DIEGO 15 0.78 1.25 9.8 CA 7 SOUTHEAST DESERT BASIN 044232 INDEPENDENCE INYO 3,950 0.70 1.30 4.8 CO 1 ARKANSAS DRAINAGE BASIN 053477 GRANADA PROWERS 3,484 0.80 1.40 14.6 CO 2 COLORADO DRAINAGE BASIN 053488 GRAND JUNCTION WALKER MESA 4,858 0.42 0.61 8.5 CO 3 KANSAS DRAINAGE BASIN 050304 ARAPAHOE CHEYENNE 4,020 0.80 1.30 15.5 CO 4 PLATTE DRAINAGE BASIN 051179 BYERS 5 ENE ADAMS 5,100 0.70 1.20 14.5 CO 5 RIO GRANDE DRAINAGE BASIN 057337 SAGUACHE SAGUACHE 7,701 0.44 0.64 8.1 CT 1 NORTHWEST 065445 NORFOLK 2 SW LITCHFIELD 1,340 1.04 1.69 50.8 CT 2 CENTRAL 063456 HARTFORD HARTFORD 190 1.01 1.60 44.9 CT 3 COASTAL 060806 BRIDGEPORT SIKORSKY AP FAIRFIELD 5 0.96 1.60 41.7 DE 1 NORTHERN 079595 WILMINGTON NEW CASTLE NEW CASTLE 79 0.99 1.62 37.7 DE 2 SOUTHERN 073570 GEORGETOWN 5 SW SUSSEX 45 1.10 1.70 40.0 FL 1 NORTHWEST 080211 APALACHICOLA AP FRANKLIN 20 1.40 2.39 54.9 FL 2 NORTH 080975 BRANFORD SUWANNEE 30 1.29 2.20 25.8 FL 3 NORTH CENTRAL 082158 DAYTONA BEACH INTL AP VOLUSIA 31 1.20 2.03 49.1 FL 4 SOUTH CENTRAL 085612 MELBOURNE WFO BREVARD 35 1.30 2.10 48.1 FL 5 EVERGLADES 083186 FT MYERS PAGE FLD AP LEE 15 1.40 2.30 54.9 A-29

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches FL 6 LOWER EAST COAST 085663 MIAMI INTL AP MIAMI-DADE 29 1.28 2.20 58.9 GA 1 NORTHWEST 092485 DALLAS 7 NE PAULDING 1,100 1.20 1.98 49.3 GA 2 NORTH CENTRAL 090451 ATLANTA/MUN., GA. FULTON 1,010 1.14 1.80 49.0 GA 3 NORTHEAST 091619 CARNESVILLE 4 N FRANKLIN 866 1.28 1.95 46.6 GA 4 WEST CENTRAL 092166 COLUMBUS METO AP MUSCOGEE 392 1.18 1.90 48.8 GA 5 CENTRAL 095443 MACON MIDDLE GA AP BIBB 343 1.13 1.86 44.8 GA 6 EAST CENTRAL 090495 AUGUSTA BUSH FLD AP RICHMOND 132 1.10 1.77 43.4 GA 7 SOUTHWEST 093028 EDISON CALHOUN 294 1.30 2.15 48.3 GA 8 SOUTH CENTRAL 096879 PEARSON ATKINSON 205 1.29 2.00 46.5 GA 9 SOUTHEAST 097847 SAVANNAH INTL AP CHATHAM 46 1.18 1.95 48.8 IA 1 NORTHWEST 136975 REMSEN PLYMOUTH 1,330 0.97 1.50 28.1 IA 2 NORTH CENTRAL 137602 SHELL ROCK BUTLER 912 0.92 1.50 30.5 IA 3 NORTHEAST 138009 STRAWBERRY POINT CLAYTON 1,200 0.93 1.52 34.4 IA 4 WEST CENTRAL 137708 SIOUX CITY AP WOODBURY 1,095 0.87 1.35 25.9 IA 5 CENTRAL 132203 DES MOINES AP POLK 957 0.91 1.50 32.3 IA 6 EAST CENTRAL 130608 BELLEVUE L&D 12 JACKSON 603 0.98 1.54 33.1 IA 7 SOUTHWEST 131245 CARSON 3NNE POTTAWATTAMIE 1,090 1.06 1.86 32.7 IA 8 SOUTH CENTRAL 132195 DERBY LUCAS 1,190 1.00 1.63 34.2 IA 9 SOUTHEAST 138688 WASHINGTON WASHINGTON 690 1.00 1.60 33.9 ID 1 PANHANDLE 101079 BONNERS FERRY BOUNDARY 2,075 0.59 0.90 23.3 ID 3 NORTH CENTRAL PRAIRIES 105241 LEWISTON AP NEZ PERCE 1,436 0.41 0.62 12.1 ID 3 NORTH CENTRAL CANYONS 103143 FENN RS IDAHO 1,560 0.68 1.00 36.4 ID 4 CENTRAL MOUNTAINS 107327 PRAIRIE ELMORE 4,780 0.68 1.05 22.2 ID 5 SOUTHWESTERN VALLEYS 101022 BOISE AIR TERMINAL ADA 2,814 0.42 0.61 11.5 ID 6 SOUTHWESTERN HIGHLANDS 103811 GRASMERE 3 S OWYHEE 5,140 0.50 0.80 10.9 ID 7 CENTRAL PLAINS 103677 GOODING 1 S GOODING 3,643 0.50 0.80 10.3 A-30

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches ID 8 NORTHEASTERN VALLEYS 105169 LEADORE LEMHI 6,000 0.40 0.60 8.5 ID 9 UPPER SNAKE RIVER PLAINS 107211 POCATELLO RGNL AP POWER 4,478 0.42 0.63 11.2 ID 10 EASTERN HIGHLANDS 104456 IDAHO FALLS 16 SE BONNEVILLE 5,828 0.50 0.70 16.4 IL 1 NORTHWEST 115751 MOLINE WSO AP ROCK ISLAND 592 0.96 1.56 36.9 IL 2 NORTHEAST 111577 CHICAGO MIDWAY AP 3SW COOK 620 0.90 1.50 36.4 IL 3 WEST 110082 ALEXIS 1 SW WARREN 680 0.97 1.60 33.2 IL 4 CENTRAL 116711 PEORIA GTR PEORIA RGNL PEORIA 650 0.93 1.47 35.3 IL 5 EAST 114198 HOOPESTON 1NE VERMILION 710 0.90 1.50 36.1 IL 6 WEST SOUTHWEST 118179 SPRINGFIELD LINCOLN AP SANGAMON 594 0.90 1.47 34.9 IL 7 EAST SOUTHEAST 116159 NEWTON 6 SSE JASPER 510 1.00 1.60 39.7 IL 8 SOUTHWEST 115983 MURPHYSBORO 2 SW JACKSON 550 1.10 1.74 43.9 IL 9 SOUTHEAST 112353 DIXON SPRINGS AG CTR POPE 527 1.15 1.82 45.8 IN 1 NORTHWEST 125535 MEDARYVILLE 5 N PULASKI 695 0.90 1.50 36.2 IN 2 NORTH CENTRAL 128187 SOUTH BEND AP ST. JOSEPH 773 0.80 1.33 37.7 IN 3 NORTHEAST 123037 FORT WAYNE AP ALLEN 791 0.82 1.30 36.5 IN 4 WEST CENTRAL 120922 BRAZIL CLAY 680 1.00 1.61 41.7 IN 5 CENTRAL 124259 INDIANAPOLIS INTL AP MARION 790 0.91 1.47 40.4 IN 6 EAST CENTRAL 120132 ALPINE 2 NE FAYETTE 850 0.92 1.46 41.1 IN 7 SOUTHWEST 122738 EVANSVILLE REGIONAL AP VANDERBURGH 400 1.03 1.70 44.0 IN 8 SOUTH CENTRAL 126580 OOLITIC PURDUE EX FRM LAWRENCE 650 1.05 1.60 43.6 IN 9 SOUTHEAST 120482 BATESVILLE WTR WKS RIPLEY 970 1.00 1.50 43.4 KS 1 NORTHWEST 143153 GOODLAND RENNER FLD SHERMAN 3,656 0.74 1.27 17.5 KS 2 NORTH CENTRAL 141767 CONCORDIA BLOSSER MUNI CLOUD 1,469 0.97 1.60 27.5 KS 3 NORTHEAST 143810 HORTON BROWN 1,030 1.10 1.90 35.4 KS 4 WEST CENTRAL 141730 COLLYER 10 S TREGO 2,407 0.90 1.40 19.8 KS 5 CENTRAL 144178 KANOPOLIS LAKE ELLSWORTH 1,492 1.07 1.70 25.5 A-31

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches KS 6 EAST CENTRAL 148167 TOPEKA BILLARD MUNI AP SHAWNEE 881 1.07 1.78 34.7 KS 7 SOUTHWEST 142164 DODGE CITY RGNL AP FORD 2,582 0.86 1.44 20.5 KS 8 SOUTH CENTRAL 148830 WICHITA MID-CONTINENT SEDGWICK 1,321 1.10 1.76 30.2 KS 9 SOUTHEAST 141351 CASSODAY 2SW BUTLER 1,440 1.15 1.80 32.2 KY 1 WESTERN 151631 CLINTON 4 S HICKMAN 350 1.20 1.90 46.5 KY 2 CENTRAL 154954 LOUISVILLE INTL AP JEFFERSON 488 0.95 1.51 43.9 KY 3 BLUE GRASS 154746 LEXINGTON BLUEGRASS AP FAYETTE 980 0.94 1.50 45.5 KY 4 EASTERN 151080 BUCKHORN LAKE PERRY 780 0.90 1.44 45.8 LA 1 NORTHWEST 168440 SHREVEPORT AP CADDO 254 1.32 2.16 42.5 LA 2 NORTH CENTRAL 169803 WINNFIELD 3 N WINN 160 1.40 2.30 56.0 LA 3 NORTHEAST 169806 WINNSBORO 5 SSE FRANKLIN 80 1.46 2.30 54.0 LA 4 WEST CENTRAL 165266 LEESVILLE VERNON 28 1.40 2.40 52.8 LA 5 CENTRAL 169357 VIDALIA #2 CONCORDIA 60 1.50 2.55 54.5 LA 6 EAST CENTRAL 160549 BATON ROUGE METRO AP EAST BATON ROUGE 64 1.36 2.30 57.7 LA 7 SOUTHWEST 165078 LAKE CHARLES AP CALCASIEU 9 1.47 2.55 55.6 LA 8 SOUTH CENTRAL 165021 LAFAYETTE LAFAYETTE 25 1.50 2.70 60.3 LA 9 SOUTHEAST 166660 NEW ORLEANS AP JEFFERSON 4 1.44 2.40 61.3 MA 1 WESTERN 193985 KNIGHTVILLE DAM HAMPSHIRE 630 1.00 1.68 44.5 MA 2 CENTRAL 190736 BLUE HILL NORFOLK 625 1.08 1.75 50.2 MA 3 COASTAL 190770 BOSTON SUFFOLK 12 0.96 1.57 43.1 MD 1 SOUTHEASTERN SHORE 188005 SALISBURY FAA AP WICOMICO 48 0.92 1.86 44.1 MD 2 CENTRAL EASTERN SHORE 183090 FEDERALSBURG CAROLINE 20 1.10 1.90 43.8 MD 3 LOWER SOUTHERN 186915 PATUXENT RIVER ST. MARY'S 38 1.10 2.00 14.3 MD 4 UPPER SOUTHERN 180465 BALTIMORE WASH INTL AP ANNE ARUNDEL 156 0.99 1.62 41.7 MD 6 NORTHERN CENTRAL 180470 BALTIMORE CITY BALTIMORE (CITY) 14 1.05 1.72 41.2 MD 7 APPALACHIAN MOUNTAIN 184030 HANCOCK WASHINGTON 384 0.90 1.30 34.4 A-32

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches MD 8 ALLEGHENY PLATEAU 188315 SINES DEEP CREEK GARRETT 2,040 0.74 1.16 46.9 ME 1 NORTHERN 171175 CARIBOU WFO AROOSTOOK 624 0.70 1.06 37.0 ME 2 SOUTHERN INTERIOR 178641 SWANS FALLS OXFORD 400 1.00 1.60 43.1 ME 3 COASTAL 176905 PORTLAND JETPORT CUMBERLAND 45 0.97 1.62 43.7 MI 1 WEST UPPER 204090 IRON MT KINGSFORD WWTP DICKINSON 1,071 0.79 1.20 28.3 MI 2 EAST UPPER 207366 SAULT STE MARIE SNDRSN CHIPPEWA 722 0.62 1.02 32.6 MI 3 NORTHWEST 200662 BELLAIRE ANTRIM 625 0.70 1.10 31.2 MI 4 NORTHEAST LOWER 200164 ALPENA CO RGNL AP ALPENA 684 0.66 1.04 27.7 MI 5 WEST CENTRAL LOWER 205712 MUSKEGON CO AP MUSKEGON 625 0.71 1.15 32.1 MI 6 CENTRAL LOWER 203170 GLADWIN GLADWIN 775 0.80 1.40 18.3 MI 7 EAST CENTRAL LOWER 203580 HARBOR BEACH HURON 600 0.70 1.10 29.5 MI 8 SOUTHWEST LOWER 203333 GERALD R FORD INTL AP KENT 803 0.82 1.33 35.8 MI 9 SOUTH CENTRAL LOWER 204155 JACKSON 3N JACKSON 950 0.80 1.26 30.6 MI 10 SOUTHEAST LOWER 202846 FLINT BISHOP INTL AP GENESEE 770 0.71 1.16 26.2 MN 1 NORTHWEST 218235 THIEF LAKE REFUGE MARSHALL 1,142 0.80 1.30 23.2 MN 2 NORTH CENTRAL 214026 INTERNATIONAL FALLS AP KOOCHICHING 1,183 0.68 1.14 24.3 MN 3 NORTHEAST 212248 DULUTH INTL AP ST. LOUIS 1,433 0.79 1.31 29.4 MN 4 WEST CENTRAL 210112 ALEXANDRIA CHANDLER FLD DOUGLAS 1,416 0.83 1.31 11.1 MN 5 CENTRAL 217294 ST CLOUD MUNI AP SHERBURNE 1,018 0.83 1.39 26.7 MN 6 EAST CENTRAL 215435 MINNEAPOLIS/ST PAUL AP HENNEPIN 872 0.80 1.33 27.7 MN 7 SOUTHWEST 218323 TRACY LYON 1,403 0.90 1.40 25.2 MN 8 SOUTH CENTRAL 215987 NORTHFIELD 2 NNE DAKOTA 890 0.90 1.42 29.1 MN 9 SOUTHEAST 217004 ROCHESTER INTL AP OLMSTED 1,304 0.85 1.44 29.5 MO 1 NORTHWEST PRAIRIE 234358 KANSAS CITY INTL AP PLATTE 1,005 1.01 1.76 37.2 MO 2 NORTHEAST PRAIRIE 237455 ST LOUIS LAMBERT AP ST. LOUIS 531 0.96 1.57 37.2 MO 3 WEST CENTRAL PLAINS 235987 NEVADA WTP VERNON 820 1.20 1.90 39.3 A-33

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches MO 4 WEST OZARKS 237976 SPRINGFIELD WSO AP GREENE 1,259 1.10 1.78 42.0 MO 5 EAST OZARKS 238620 VIENNA 2 WNW MARIES 770 1.11 1.74 41.3 MO 6 BOOTHEEL 233999 HORNERSVILLE DUNKLIN 250 1.25 2.07 46.4 MS 1 UPPER DELTA 221743 CLEVELAND 3 N BOLIVAR 140 1.40 2.30 50.4 MS 2 NORTH CENTRAL 227815 SARDIS DAM PANOLA 303 1.32 2.16 53.9 MS 3 NORTHEAST 229003 TUPELO RGNL AP LEE 361 1.30 2.13 54.4 MS 4 LOWER DELTA 228445 STONEVILLE EXP STN WASHINGTON 127 1.40 2.10 50.0 MS 5 CENTRAL 225062 LEXINGTON HOLMES 285 1.40 2.23 54.1 MS 6 EAST CENTRAL 228374 STATE UNIV OKTIBBEHA 185 1.30 2.11 51.8 MS 7 SOUTHWEST 227714 RUTH 1 SE LINCOLN 443 1.43 2.35 56.4 MS 8 SOUTH CENTRAL 227220 PURVIS 2N LAMAR 378 1.40 2.30 59.8 MS 9 SOUTHEAST 225776 MERIDIAN AP LAUDERDALE 294 1.30 2.12 55.2 MS 10 COASTAL 227840 SAUCIER EXP FOREST HARRISON 229 1.50 2.50 66.9 MT 1 WESTERN 245745 MISSOULA INTL AP MISSOULA 3,192 0.41 0.64 13.2 MT 2 SOUTHWESTERN 241309 BUTTE 8 S SILVER BOW 5,700 0.50 0.80 14.8 MT 3 NORTH CENTRAL 241737 CHOTEAU TETON 3,845 0.63 1.00 10.6 MT 4 CENTRAL 244055 HELENA AP ASOS LEWIS AND CLARK 3,828 0.45 0.73 11.0 MT 5 SOUTH CENTRAL 240807 BILLINGS INTL AP YELLOWSTONE 3,581 0.51 0.85 13.7 MT 6 NORTHEASTERN 241088 BREDETTE ROOSEVELT 2,638 0.65 1.02 12.3 MT 7 SOUTHEASTERN 248169 TERRY 21 NNW PRAIRIE 3,142 0.63 1.06 13.9 NC 1 SOUTHERN MOUNTAINS 310301 ASHEVILLE BUNCOMBE 2,238 0.82 1.37 37.8 NC 2 NORTHERN MOUNTAINS 319675 YADKINVILLE 6 E YADKIN 875 1.00 1.63 43.1 NC 3 NORTHERN PIEDMONT 313630 GREENSBORO AP GUILFORD 890 0.98 1.56 42.1 NC 4 CENTRAL PIEDMONT 317069 RALEIGH AP WAKE 416 0.99 1.57 42.2 NC 5 SOUTHERN PIEDMONT 311690 CHARLOTTE DOUGLAS AP MECKLENBURG 728 1.01 1.67 42.4 NC 6 SOUTHERN COASTAL PLAIN 319457 WILMINGTON INTL AP NEW HANOVER 33 1.24 2.19 54.7 A-34

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches NC 7 CENTRAL COASTAL PLAIN 319476 WILSON 3 SW WILSON 110 1.10 1.74 38.5 NC 8 NORTHERN COASTAL PLAIN 311458 CAPE HATTERAS AP DARE 11 1.26 2.16 55.8 ND 1 NORTHWEST 329425 WILLISTON SLOULIN FLD WILLIAMS 1,902 0.57 1.02 13.4 ND 2 NORTH CENTRAL 320492 BALFOUR 3 SW MCHENRY 1,615 0.76 1.30 15.9 ND 3 NORTHEAST 321435 CAVALIER 7NW PEMBINA 890 0.90 1.50 19.3 ND 4 WEST CENTRAL 327585 RIVERDALE MCLEAN 1,977 0.80 1.35 16.6 ND 5 CENTRAL 322018 DAWSON KIDDER 1,730 0.79 1.38 17.5 ND 6 EAST CENTRAL 322859 FARGO HECTOR INTL AP CASS 900 0.74 1.30 20.0 ND 7 SOUTHWEST 327530 RICHARDTON ABBEY STARK 2,470 0.70 1.21 17.0 ND 8 SOUTH CENTRAL 320819 BISMARCK MUNI AP BURLEIGH 1,651 0.66 1.12 15.9 ND 9 SOUTHEAST 320382 ASHLEY MCINTOSH 2,014 0.80 1.40 17.6 NE 1 PANHANDLE 257665 SCOTTSBLUFF AP SCOTTS BLUFF 3,945 0.61 0.99 14.6 NE 2 NORTH CENTRAL 258760 VALENTINE MILLER AP CHERRY 2,590 0.73 1.22 18.3 NE 3 NORTHEAST 255995 NORFOLK AP MADISON 1,551 0.88 1.48 24.8 NE 5 CENTRAL 253395 GRAND ISLAND CTR NE AP HALL 1,840 0.88 1.45 23.9 NE 6 EAST CENTRAL 254795 LINCOLN AP LANCASTER 1,190 0.94 1.55 17.8 NE 7 SOUTHWEST 256065 NORTH PLATTE RGNL AP LINCOLN 2,778 0.81 1.25 19.4 NE 8 SOUTH CENTRAL 252560 EDISON FURNAS 2,120 0.90 1.41 21.2 NE 9 SOUTHEAST 258395 SYRACUSE OTOE 1,100 1.00 1.70 30.0 NH 1 NORTHERN 275639 MT WASHINGTON COOS 6,267 1.02 1.77 92.2 NH 2 SOUTHERN 271683 CONCORD ASOS MERRIMACK 346 0.85 1.33 38.1 NJ 1 NORTHERN 286026 NEWARK INTL AP ESSEX 7 0.99 1.61 44.0 NJ 2 SOUTHERN 280311 ATLANTIC CITY INTL AP ATLANTIC 60 1.01 1.63 41.2 NJ 3 COASTAL 281351 CAPE MAY 2 NW CAPE MAY 20 1.00 1.66 37.3 NM 1 NORTHWESTERN PLATEAU 293142 FARMINGTON AG SCI CTR SAN JUAN 5,625 0.50 0.70 7.7 NM 2 NORTHERN MOUNTAINS 292837 EL VADO DAM RIO ARRIBA 6,740 0.50 0.80 14.3 A-35

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches NM 3 NORTHEASTERN PLAINS 292030 CONCHAS DAM SAN MIGUEL 4,244 0.80 1.30 13.7 NM 4 SOUTHWESTERN MOUNTAINS 292250 CUBERO CIBOLA 6,195 0.60 0.90 10.9 NM 5 CENTRAL VALLEY 290234 ALBUQUERQUE INTL AP BERNALILLO 5,310 0.51 0.77 8.6 NM 6 CENTRAL HIGHLANDS 297094 PROGRESSO TORRANCE 6,297 0.62 1.02 12.8 NM 7 SOUTHEASTERN PLAINS 290600 ARTESIA 6S EDDY 3,366 0.78 1.30 10.7 NM 8 SOUTHERN DESERT 294426 JORNADA EXP RANGE DONA ANA 4,266 0.65 1.07 9.2 NV 1 NORTHWESTERN 269171 WINNEMUCCA AP HUMBOLDT 4,296 0.39 0.58 7.9 NV 2 NORTHEASTERN 262631 ELY AIRPORT WHITE PINE 6,262 0.44 0.63 9.1 NV 3 SOUTH CENTRAL 268170 TONOPAH AIRPORT NYE 5,395 0.42 0.63 2.3 NV 4 EXTREME SOUTHERN 264436 LAS VEGAS AP CLARK 2,131 0.58 0.88 4.2 NY 1 WESTERN PLATEAU 303983 HORNELL ALMOND DAM STEUBEN 1,325 0.70 1.20 32.2 NY 2 EASTERN PLATEAU 300687 BINGHAMTON GREATER AP BROOME 1,595 0.74 1.16 37.1 NY 3 NORTHERN PLATEAU 303851 HIGHMARKET LEWIS 1,763 0.90 1.50 52.5 NY 4 COASTAL 305811 NEW YORK LA GUARDIA AP QUEENS 11 1.00 1.68 43.1 NY 5 HUDSON VALLEY 300042 ALBANY INTL AP ALBANY 275 0.80 1.27 36.7 NY 6 MOHAWK VALLEY 308586 TRIBES HILL MONTGOMERY 300 0.80 1.30 36.4 NY 7 CHAMPLAIN VALLEY 309389 WHITEHALL WASHINGTON 119 0.87 1.35 36.1 NY 8 ST. LAWRENCE VALLEY 301185 CANTON 4 SE ST. LAWRENCE 448 0.70 1.14 33.9 NY 9 GREAT LAKES 307167 ROCHESTER INTL AP MONROE 533 0.65 1.03 32.0 NY 10 CENTRAL LAKES 308383 SYRACUSE HANCOCK AP ONONDAGA 413 0.70 1.13 36.8 OH 1 NORTHWEST 338357 TOLEDO EXPRESS WSO AP LUCAS 669 0.76 1.22 32.6 OH 2 NORTH CENTRAL 336196 OBERLIN LORAIN 816 0.78 1.24 34.7 OH 3 NORTHEAST 330058 AKRON CANTON WSO AP SUMMIT 1,208 0.72 1.21 36.3 OH 4 WEST CENTRAL 337935 SPRINGFIELD NEW WWKS CLARK 930 0.90 1.40 38.6 OH 5 CENTRAL 331786 COLUMBUS WSO AP FRANKLIN 810 0.79 1.32 38.0 OH 6 EAST CENTRAL 334865 MANSFIELD WSO AP RICHLAND 1,295 0.80 1.30 34.2 A-36

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches OH 7 NORTHEAST HILLS 334992 MASSILLON STARK 930 0.90 1.40 11.2 OH 8 SOUTHWEST 332075 DAYTON WSO AP MONTGOMERY 1,000 0.80 1.34 37.5 OH 9 SOUTH CENTRAL 334004 JACKSON 3 NW JACKSON 800 0.90 1.45 36.7 OH 10 SOUTHEAST 338378 JENKINS DAM BURR OAK ATHENS 760 0.80 1.30 35.1 OK 1 PANHANDLE 343002 EVA TEXAS 3,574 0.80 1.30 15.1 OK 2 NORTH CENTRAL 343304 FORT SUPPLY 3SE WOODWARD 2,030 1.00 1.71 21.4 OK 3 NORTHEAST 348992 TULSA INTL AP TULSA 650 1.23 1.95 39.1 OK 4 WEST CENTRAL 345648 MAYFIELD BECKHAM 2,005 1.10 1.90 23.9 OK 5 CENTRAL 346661 OKLAHOMA CITY WILL ROGERS AP OKLAHOMA 1,285 1.17 1.87 33.2 OK 6 EAST CENTRAL 348497 STIGLER 1 SE HASKELL 570 1.35 2.30 43.1 OK 7 SOUTHWEST 343281 FT COBB CADDO 1,285 1.20 2.01 28.8 OK 8 SOUTH CENTRAL 344865 KINGSTON 5 SSE MARSHALL 684 1.34 2.18 38.0 OK 9 SOUTHEAST 340670 BENGAL 4 NNW LATIMER 667 1.40 2.40 48.3 OR 1 COASTAL AREA 350328 ASTORIA AP PORT OF CLATSOP 9 0.88 1.42 67.8 OR 2 WILLAMETTE VALLEY 356751 PORTLAND INTL AP MULTNOMAH 19 0.63 0.98 36.7 OR 3 SOUTHWESTERN VALLEYS 355429 MEDFORD INTL AP JACKSON 1,297 0.60 0.97 19.1 OR 4 NORTHERN CASCADES 352697 ESTACADA 24 SE CLACKAMAS 2,200 0.88 1.40 54.0 OR 5 HIGH PLATEAU 353232 GERBER DAM KLAMATH 4,850 0.60 1.20 17.7 OR 6 NORTH CENTRAL 356546 PENDLETON E OR RGNL AP UMATILLA 1,486 0.41 0.62 12.1 OR 7 SOUTH CENTRAL 354670 LAKEVIEW 2 NNW LAKE 4,890 0.50 0.79 15.4 OR 8 NORTHEAST 356845 PRAIRIE CITY RS GRANT 3,540 0.47 0.70 14.6 OR 9 SOUTHEAST 354321 JORDAN VALLEY MALHEUR 4,390 0.45 0.71 12.5 PA 1 POCONO MOUNTAINS 369705 WILKES-BARRE INTL AP LUZERNE 930 0.78 1.29 37.0 PA 2 EAST CENTRAL MOUNTAINS 360106 ALLENTOWN AP LEHIGH 390 0.99 1.57 44.1 PA 3 SOUTHEASTERN PIEDMONT 366889 PHILADELPHIA INTL AP PHILADELPHIA 10 0.99 1.60 41.5 PA 4 LOWER SUSQUEHANNA 363699 HARRISBURG CAPITAL CY YORK 340 0.84 1.47 32.5 A-37

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches PA 5 MIDDLE SUSQUEHANNA 367931 SELINSGROVE 2 S SNYDER 420 1.00 1.70 24.6 PA 6 UPPER SUSQUEHANNA 368905 TOWANDA 1 S BRADFORD 760 0.80 1.21 32.7 PA 7 CENTRAL MOUNTAINS 362265 DU BOIS 7 E CLEARFIELD 1,670 0.75 1.15 40.6 PA 8 SOUTH CENTRAL MOUNTAINS 364001 HOLLIDAYSBURG 2 NW BLAIR 990 0.81 1.30 36.4 PA 9 SOUTHWEST PLATEAU 366993 PITTSBURGH ASOS ALLEGHENY 1,203 0.71 1.13 34.5 PA 10 NORTHWEST PLATEAU 362682 ERIE WSO AP ERIE 730 0.73 1.21 36.6 RI 1 ALL 376698 PROVIDENCE KENT 60 1.05 1.75 45.6 SC 1 MOUNTAIN 384581 JOCASSEE 8 WNW OCONEE 2,500 1.60 2.80 82.4 SC 2 NORTHWEST 383747 GRNVL SPART INTL AP SPARTANBURG 943 1.12 1.86 48.8 SC 3 NORTH CENTRAL 389327 WINNSBORO FAIRFIELD 560 1.20 1.90 35.6 SC 4 NORTHEAST 385306 LORIS 2 S HORRY 90 1.20 1.90 46.8 SC 5 WEST CENTRAL 386209 NEWBERRY NEWBERRY 476 1.20 1.90 44.9 SC 6 CENTRAL 381939 COLUMBIA METRO AP LEXINGTON 225 1.19 1.92 45.1 SC 7 SOUTHERN 381544 CHARLESTON INTL AP CHARLESTON 40 1.20 2.00 49.4 SD 1 NORTHWEST 394864 LEMMON PERKINS 2,567 0.71 1.20 16.7 SD 2 NORTH CENTRAL 396282 ONAKA 2N FAULK 1,610 0.80 1.40 18.1 SD 3 NORTHEAST 390020 ABERDEEN RGNL AP BROWN 1,297 0.79 1.32 18.6 SD 4 BLACK HILLS 396427 PACTOLA DAM PENNINGTON 4,720 0.70 1.20 18.5 SD 5 SOUTHWEST 392557 EDGEMONT FALL RIVER 3,610 0.62 1.02 15.0 SD 6 CENTRAL 396170 OAHE DAM STANLEY 1,660 0.80 1.30 15.0 SD 7 EAST CENTRAL 394127 HURON AP BEADLE 1,280 0.78 1.28 20.0 SD 8 SOUTH CENTRAL 395620 MISSION TODD 2,587 0.90 1.41 20.8 SD 9 SOUTHEAST 397667 SIOUX FALLS AP MINNEHAHA 1,428 0.85 1.41 23.9 TN 1 EASTERN 401656 CHATTANOOGA AP HAMILTON 671 1.12 1.79 53.1 TN 2 CUMBERLAND PLATEAU 406170 MONTEREY PUTNAM 1,860 1.19 1.82 55.7 TN 3 MIDDLE 406402 NASHVILLE ASOS DAVIDSON 600 1.05 1.75 47.8 A-38

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches TN 4 WESTERN 405954 MEMPHIS INTL AP SHELBY 254 1.30 2.02 51.6 TX 1 HIGH PLAINS 415411 LUBBOCK INTL AP LUBBOCK 3,254 0.91 1.55 15.8 TX 2 LOW ROLLING PLAINS 419729 WICHITA FALLS MUNI AP WICHITA 1,017 1.13 1.86 27.5 TX 3 NORTH CENTRAL 413284 FT WORTH MEACHAM FLD TARRANT 687 1.20 2.03 28.6 TX 4 EAST TEXAS 419665 WHEELOCK ROBERTSON 420 1.36 2.20 36.6 TX 5 TRANS PECOS 412797 EL PASO AP EL PASO 3,918 0.60 1.04 8.5 TX 6 EDWARDS PLATEAU 418845 TARPLEY BANDERA 1,390 1.30 2.22 31.3 TX 7 SOUTH CENTRAL 410428 AUSTIN CAMP MABRY TRAVIS 670 1.21 2.01 32.3 TX 8 UPPER COAST 419364 VICTORIA ASOS VICTORIA 115 1.32 2.30 30.5 TX 9 SOUTHERN 414191 HINDES ATASCOSA 360 1.30 2.38 23.5 TX 10 LOWER VALLEY 411136 BROWNSVILLE INTL AP CAMERON 24 1.17 2.28 25.7 UT 1 WESTERN 422090 DELTA MILLARD 4,620 0.45 0.69 7.7 UT 2 DIXIE 427516 ST GEORGE WASHINGTON 2,770 0.55 0.80 7.1 UT 3 NORTH CENTRAL 427598 SALT LAKE CITY INTL AP SALT LAKE 4,225 0.51 0.80 15.3 UT 4 SOUTH CENTRAL 426135 NEPHI JUAB 5,128 0.50 0.75 14.3 UT 5 NORTHERN MOUNTAINS 422385 ECHO DAM SUMMIT 5,470 0.50 0.70 14.0 UT 6 UINTA BASIN 427395 ROOSEVELT RADIO UINTAH 5,014 0.46 0.70 6.8 UT 7 SOUTHEAST 420738 BLANDING SAN JUAN 6,032 0.58 0.90 11.6 VA 1 TIDEWATER 446139 NORFOLK INTL AP NORFOLK (CITY) 30 1.01 1.76 43.9 VA 2 EASTERN PIEDMONT 447201 RICHMOND INTL AP HENRICO 164 1.01 1.69 43.4 VA 3 WESTERN PIEDMONT 445120 LYNCHBURG INTL AP CAMPBELL 940 0.93 1.55 40.6 VA 4 NORTHERN 448906 WASHINGTON/NAT., VA. ARLINGTON 10 0.94 1.50 39.6 VA 5 CENTRAL MOUNTAIN 447285 ROANOKE INTL AP ROANOKE 1,175 0.93 1.50 40.2 VA 6 SOUTHWESTERN MOUNTAIN 448547 TROUT DALE 3 SSE GRAYSON 2,839 0.90 1.44 45.4 VT 1 NORTHEASTERN 431565 CORINTH ORANGE 1,180 0.80 1.30 37.8 VT 2 WESTERN 431081 BURLINGTON WSO AP CHITTENDEN 330 0.69 1.10 34.5 A-39

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches VT 3 SOUTHEASTERN 438556 UNION VILLAGE DAM ORANGE 460 0.77 1.26 34.3 WA 1 WEST OLYMPIC COAST 456858 QUILLAYUTE AP CLALLAM 185 1.20 1.97 101.9 WA 2 NE OLYMPIC SAN JUAN 456678 PORT TOWNSEND JEFFERSON 100 0.46 0.72 22.3 WA 3 PUGET SOUND LOWLANDS 456114 OLYMPIA AP THURSTON 188 0.79 1.27 47.9 WA 4 E OLYMPIC CASCADE FOOTHILLS 453357 GREENWATER PIERCE 1,730 0.79 1.31 52.6 WA 5 CASCADE MOUNTAINS WEST 458009 STAMPEDE PASS KITTITAS 3,959 1.07 1.80 80.3 WA 6 EAST SLOPE CASCADES 454849 LUCERNE 1 N CHELAN 1,200 0.68 1.16 26.7 WA 7 OKANOGAN BIG BEND 453515 HARRINGTON 1 NW LINCOLN 2,160 0.50 0.66 12.7 WA 8 CENTRAL BASIN 458207 SUNNYSIDE YAKIMA 747 0.40 0.60 7.6 WA 9 NORTHEASTERN 457938 SPOKANE INTL AP SPOKANE 2,353 0.45 0.66 16.5 WA 10 PALOUSE BLUE MOUNTAINS 456789 PULLMAN 2 NW WHITMAN 2,545 0.51 0.80 21.4 WI 1 NORTHWEST 470349 ASHLAND EXP FARM BAYFIELD 650 0.80 1.34 28.9 WI 2 NORTH CENTRAL 476939 RAINBOW RSVR TOMAHAWK ONEIDA 1,600 0.74 1.20 27.9 WI 3 NORTHEAST 476510 PESHTIGO MARINETTE 600 0.80 1.30 28.7 WI 4 WEST CENTRAL 475948 NEW RICHMOND ST. CROIX 1,000 0.90 1.50 29.7 WI 5 CENTRAL 471676 CLINTONVILLE WAUPACA 802 0.82 1.39 30.1 WI 6 EAST CENTRAL 473269 GREEN BAY A S INTL AP BROWN 687 0.74 1.18 27.8 WI 7 SOUTHWEST 474546 LANCASTER 4 WSW GRANT 1,040 0.90 1.50 32.7 WI 8 SOUTH CENTRAL 474961 MADISON DANE CO AP DANE 866 0.83 1.42 32.1 WI 9 SOUTHEAST 475479 MILWAUKEE MITCHELL AP MILWAUKEE 670 0.79 1.33 31.9 WV 1 NORTHWESTERN 466859 PARKERSBURG WOOD 620 0.78 1.18 37.4 WV 2 NORTH CENTRAL 465002 LAKE LYNN MONONGALIA 900 0.75 1.20 38.4 WV 3 SOUTHWESTERN 461570 CHARLESTON YEAGER AP KANAWHA 910 0.79 1.24 43.0 WV 4 CENTRAL 462718 ELKINS RANDOLPH CY AP RANDOLPH 1,979 0.72 1.12 44.3 WV 5 SOUTHERN 469011 UNION 3 SSE MONROE 2,110 0.80 1.20 35.7 WV 6 NORTHEASTERN 465739 MATHIAS HARDY 1,540 0.80 1.30 33.6 A-40

NCHRP 25-41 Guidance for Achieving Volume Reduction of Highway Runoff in Urban Areas State Climate Division Number Climate Division Name Selected COOP Station COOP Station Name County Elevation, MSL Calculated 85th Percentile, 24-hour Storm Depth Calculated 95th Percentile, 24-hour Storm Depth Calculated Average Annual Precipitation Depth, inches WY 1 YELLOWSTONE DRAINAGE 485345 LAKE YELLOWSTONE TETON 7,870 0.43 0.70 20.2 WY 2 SNAKE DRAINAGE 486440 MORAN 5 WNW TETON 6,798 0.50 0.77 23.5 WY 3 GREEN AND BEAR DRAINAGE 487845 ROCK SPRINGS AP SWEETWATER 6,741 0.40 0.60 4.2 WY 4 BIG HORN 488852 TENSLEEP 4NE WASHAKIE 4,815 0.60 0.90 12.9 WY 5 POWDER, LITTLE MISSOURI, TONGU 488155 SHERIDAN AP SHERIDAN 3,945 0.52 0.83 14.3 WY 6 BELLE FOURCHE DRAINAGE 487270 PINE TREE 9 NE CAMPBELL 5,111 0.60 1.11 11.5 WY 7 CHEYENNE & NIOBRARA DRAINAGE 486660 NEWCASTLE WESTON 4,315 0.60 1.00 15.0 WY 8 LOWER PLATTE 481570 CASPER WSCMO NATRONA 5,338 0.45 0.80 11.9 WY 9 WIND RIVER 485390 LANDER AP FREMONT 5,557 0.60 0.97 12.9 WY 10 UPPER PLATTE 487105 PATHFINDER DAM NATRONA 5,918 0.50 0.75 9.2 A-41