Click for next page ( 52


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 51
52 CHAPTER FOUR FINDINGS AND SUGGESTIONS FOR RESEARCH Statewide travel forecasting is becoming a more common The 1995 American Travel Survey (ATS), the last ma- activity in transportation planning. There is an increase in jor source of information on long distance passenger the number of states with models and many states are in the travel, has not been updated. process of revising their models. The impetus for develop- Models tend to still have a time period of 24 h; none of ing a statewide model varies greatly from state to state. In the states have implemented the dynamic methods nec- some states models were created to address the needs of a essary for good forecasts of peak-hour travel in larger specific large project; in other states models were created for states. general planning needs. Statewide models have become es- There has been little progress in the creation and use of sential in some states for intercity corridor and statewide transferable parameters within any of the model steps. system planning. With the exception of a few geographically small states, there has not been a full integration of statewide and ur- Most statewide models are similar in structure to four-step ban models. Integration is easier to achieve in small urban transportation planning models. Statewide models dif- states where there are only a few metropolitan planning fer from urban models primarily in how the steps are config- organizations in close proximity or one that spans the ured. There exists no well-accepted definition of best practice state. Statewide models defer to urban models within in statewide models. Models range greatly in cost, staffing re- urban areas. quirements, development time frame, and capabilities. There has been little progress in integrating statewide models with national models, particularly the Freight Special data collection efforts, apart from National Analysis Framework (FAF). Household Travel Survey (NHTS) add-ons, are sporadic. Most states are making efficient use of a wide variety of sec- This review identified two particular issues that are limit- ondary data sources. ing progress in statewide model development. The following several distinct trends are apparent in re- Many databases are organized by county or other spa- cent statewide model development. tial units that are too coarse. More experience is needed with modeling multiday, Many newer models have network detail at about the long distance trip making. same level of precision as urban models. There are more freight components that are commodity- The following innovations in both statewide and national based, rather than being truck-only. modeling may lead to better planning practice. There is a greater and more effective use of geographic information systems to manage and acquire model data. Some states use nested zone structures to better tailor There is more of a tendency to hold statewide models to the level of spatial aggregation to the needs of a given the same standards of validation accuracy as urban model step. For example, several states have imple- models. mented subzones during traffic assignment to eliminate There has been a doubling of states (from one to two) lumpy loadings. Other states have adopted dual sets of that are pursuing models with integrated economic ac- zones and networks to model both national and local tivity components. travel effects. Traffic assignments are less likely to be all-or-nothing The full integration of freight, passenger, and economic and more likely to be equilibrium. activity offers a worthwhile direction for the next gen- There is a greater emphasis on multiclass traffic as- eration of statewide travel forecasting models. signment for combining freight and passenger traffic Some states have implemented tour-based passenger forecasts. components within their statewide models. Proposed improvements to FHWA's FAF may enable There are planning needs that have not been fully realized more rapid development of statewide freight compo- because of deficiencies in either data or algorithms. nents that are more accurate and more policy sensitive.

OCR for page 51
53 At this time there is no pressing need for best practice Potential sources of data to support and evaluate standards for statewide models. In states with integrated statewide models. models, the state of the practice exceeds the curricular con- Multimodal performance standards for trip genera- tent of transportation planning graduate programs. tion and activity, trip length and duration, mode choice, corridor assignment, low-volume roadway The review of current practice supports the four principal assignment, rural areas and facilities, multimodal de- research suggestions of the Statewide Travel Demand Mod- mand, and multimodal assignment. els Peer Exchange. These research suggestions have been Comparison of urban and statewide planning model previously identified as being of high priority. results and sensitivities. Estimates of the time and costs for various options. Rural Area Trip-Making Characteristics. Many urban Long-Distance Travel Data Collection. Many states models have benefited greatly from the existence of found the ATS to be an invaluable source of informa- transferable parameters for forecasting travel within ur- tion on long distance travel within and across their ban areas. Notable sources of such parameters are borders. However, the latest data from the ATS is now NCHRP Report 187, NCHRP Report 365, and the 10 years old, and although the 2001 NHTS also con- Quick Response Freight Manual (QRFM). Similar data tains data on long distance, infrequent trips, the data set have not been compiled for intercity or rural travel. is limited in the number of samples and the number of Some statewide models have used urban parameters for trips reported. It is suggested that the ATS be repeated rural travel, perhaps introducing an unnecessary error to or the NHTS be upgraded to a comparable level of forecasts. Research is needed to define trip generation detail for long distance trip making. rates, trip distribution friction factors, vehicle occu- pancy rates, time-of-day factors, and mode-split model In addition, the state of the practice suggests that addi- coefficients. This information is needed principally for tional research be undertaken in the following areas. passenger travel. A potential source of much of this in- formation is the NHTS. The QRFM should be updated Improvements in Traffic Assignment. As with urban to include rural commercial trip characteristics. models, traffic assignment is the step closest to the re- Development of a National Passenger Travel Model. sults that influence decision making. However, produc- The United States does not have a national model of pas- ing a traffic assignment is much more difficult in senger travel, although it does have a national freight statewide models because of the larger sizes of the net- model (FHWA's FAF). Currently, most statewide mod- works, the distance between origins and destinations, els have networks that extend well into neighboring and the coarseness of zone systems. In particular, there states and beyond. A national passenger model would go are three issues that need further investigation. a long way toward relieving statewide planners from the Peak periods and traffic dynamics. In states where in- burden of modeling vast areas outside their borders to tercity trip durations greatly exceed 1 h, static traffic properly account for external travel. The main purpose assignment is incapable of directly performing peak- of a national model would be to obtain reliable forecasts hour forecasts. Dynamic traffic assignment can track of passenger vehicle flows between states on major U.S. groups of vehicles in both time and space; therefore, highways and passenger volumes through major airports it potentially can estimate traffic volumes and delays and rail and bus terminals. Local detail in such a model for short periods of time. Because no state is cur- would not be needed. rently using dynamic traffic assignment, its applica- Development of Validation Performance Standards for bility should be tested on full-sized networks. Statewide Models. There are well-recognized quality Spatial aggregation. To cover the full land area of a standards for urban travel forecasting models, but none state, zone systems have been coarse. A few states have for statewide travel forecasting models. Because experimented with subzones during the assignment statewide models tend to be coarser than urban models step to remove errors associated with large zones; how- and because statewide models are used to study a nar- ever, more experience is necessary. Research is needed rower range of policies and project options, there is a to determine the best methods for establishing sub- sentiment within some states that statewide models do zones and to ascertain the potential benefits. not need to meet strict urban standards for validation. Speed of execution. Some newer models have very Research is needed in these areas, as identified by the large networks, causing very slow path building and Peer Exchange. traffic assignment. There is a need for faster algo- Acceptable ranges of parameters and values used as rithms, either by writing better algorithms or by fully inputs to statewide models. exploiting computer hardware. Speed of execution Key market segments that should be addressed in will be of increased concern as states adopt dynamic statewide models. and multiclass traffic assignment methods. Suitable and unsuitable applications of statewide Intermodal Freight Networks. As with urban models, models. many statewide models have truck networks. Networks

OCR for page 51
54 for other freight modes are rare, and no state has re- Better Public Source Commodity Flow Information. ported networks capable of handling intermodal freight. The Commodity Flow Survey has been an invaluable Additional research and experience is necessary to source of information on freight shipments within the determine the best way to build networks for handling United States. However, the survey does not provide freight that use more than one mode. complete information. Innovative methods are needed Cost Models for Freight and Freight Modal Choice Pa- for combining existing data sources and economic rameters. Cost is the most important factor in freight models for filling in the gaps in the Commodity Flow mode choice. The knowledge of mode choice for long Study. These methods need to be expressed as simple distance freight is inadequate, principally because the procedures that can be executed by modeling staffs at costs of transporting freight are not well understood. state departments of transportation. The potential of NCHRP Report 260 contains detailed methods for esti- FHWA's FAF for providing better commodity flow in- mating the costs of freight; however, those methods are formation could be explored. now outdated. New research is needed to ascertain the Better Information on Non-Freight Commercial Vehicle costs of moving one ton of a particular commodity from Movements. Both urban and statewide models could origin to destination by a variety of competing models. benefit from a better understanding of commercial vehi- Additional research is needed to determine the sensi- cle movements that are not transporting freight. A tivity of cost relative to other factors within the modal means of acquiring such information might be a Na- choice process. The effect of changing logistics prac- tional Business Travel Survey, which would be analo- tices on long distance freight movements needs to be gous to the NHTS. Such a survey would also be an op- quantified. portunity to learn more about business logistics practices Innovative Methods of Estimating OriginDestination and supplement the information from the Commodity Tables from Ground Counts. Developing very large Flow Survey. This information could also be helpful for origindestination tables for specific purposes, modes, developing default commercial trip making characteris- and commodities is currently difficult because of the tics for an update of the QRFM. amount of required information and the amount of com- Improved Curricula for Transportation Planning putation time. Better methods, suitable for highly Graduate Programs. The emergence of integrated detailed and multiclass models, are needed to find accu- transportation/land use/economic activity models sug- rate tables that use more then simple ground counts as gests that related topics might be elevated in importance inputs. Such methods need to be validated for accuracy. within graduate program curricula.