Click for next page ( 2


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 1
STATEWIDE TRAVEL FORECASTING MODELS SUMMARY Statewide travel forecasting models attempt to meet some of the same goals for statewide transportation planning that urban travel forecasting models have met for urban transporta- tion planning. The earliest experiments in statewide travel forecasting in the 1970s adapted methods that had been developed specifically for urban travel forecasting; however, those early statewide modeling efforts were severely hampered because of difficulties in ade- quately covering large geographic areas in sufficient detail. In the past 10 years statewide transportation planners have seen dramatic improvements in socioeconomic and network databases, tools for accessing these databases, and computational power. Consequently, interest in fully capable statewide travel forecasting models has steadily increased. Each succeeding generation of models within a state has become more ambitious. Approximately one-half of the states now have functional models. This synthesis is particularly directed toward those states that want to develop a model from scratch and those states that are interested in upgrading their existing models. The core of this synthesis is the results of surveys received from every state that has a statewide travel forecasting model. Information about modeling activities was provided by 49 states returning at least one questionnaire. The responses to the synthesis questionnaires, along with those from an earlier questionnaire distributed by the TRB Statewide Travel De- mand Models Peer Exchange in 2004, allow for a general assessment of the state of the prac- tice. The questionnaires focused on individual components of the models and the modeling process. To achieve a better understanding of how all the pieces fit together, five case stud- ies are presented to provide a broader overview. Two of the case studies, Indiana and Kentucky, deal specifically with passenger components. Two other case studies, Virginia and Wisconsin, cover freight components. One additional case study from Ohio explores a newer approach to statewide models that integrates the traditional passenger and freight components with forecasts of economic activity and land use that had been pioneered in Oregon. In ad- dition, several states were asked to expand on their questionnaire responses with regard to how statewide models have been successfully applied. Preceding the survey results are reviews of the literature and key concepts. The reviews provide the basis for a full understanding of current practice without duplicating literature re- views contained in other readily available documents. These reviews focus on standard ref- erences on statewide models, recently published research on intercity travel in the United States, and key databases. Chapter one finishes with a glossary of terms that are often used by those individuals who build statewide travel forecasting models, but which might be less familiar to others. Additional reviews of literature in passenger and freight modeling are found in Appendixes C and D, respectively. The state of the practice has matured over the last 10 years. Many statewide travel fore- casting models now have network detail similar to urban models. Validation standards have increased, such that some models are now able to achieve the same level of accuracy as ur- ban models. With the exceptions of Ohio and Oregon, statewide models still closely follow urban models in structure within their passenger travel components. However, there is a trend away from truck-only freight components toward commodity-based freight components,

OCR for page 1
2 which better exploit available freight databases. Ohio and Oregon are implementing a new modeling paradigm that integrates forecasts of economic activity and land use into the travel model. Statewide models have proven to be versatile tools in assisting in the development of both statewide and metropolitan area plans. Such models are primarily used for intercity corridor planning, statewide system planning, and bypass studies; however, they are also frequently used for providing input to metropolitan planning organization (MPO) models, replacing MPO mod- els, or serving as the main forecasting means for rural projects. Statewide models have been used in several states for air quality conformity analysis, freight planning, traffic impact stud- ies, economic development studies, project prioritization, and many other planning needs. Given that there are no best-practice standards for statewide models, different states have taken different approaches to building their models to meet their particular needs. Develop- ment times range from approximately 6 months to 8 years, and development costs have ranged from less than $100,000 to many millions of dollars. The level of detail in both net- works and zone systems also varies greatly. Models fall into five general categories: (1) origindestination (OD) table estimation and assignment, (2) freight only, (3) passenger only, (4) combined passenger and freight, and (5) integrated passenger/freight/economic activity. Most states with models have avoided original data collection; tending to rely heavily on secondary data sources. Important data sources for passenger components include the Cen- sus Transportation Planning Package, the National Household Travel Survey, MPO data- bases, the American Travel Survey, and in-house traffic counts. Some states have purchased National Household Travel Survey add-ons. Freight data often came from the Vehicle In- ventory and Use Survey, a particular freight data vendor, the Commodity Flow Survey, and the Rail Carload Waybill Sample. Most passenger components are multimodal, and all passenger components include auto- mobiles. Other commonly found modes are intercity railroad, intercity bus, local bus, and commuter railroad. The geographic size of states means that many statewide models are still spatially and temporally coarser than urban models. The coarseness is exacerbated by the need to con- sider long distance trips that start or end in other states. Indiana, Ohio, and Texas have the largest zone systems, with more than 4,500 zones each. Most models have avoided the use of special generators. Networks with more than 200,000 links have been created. Although smaller states are capable of running peak-period traffic assignments, most states run 24-h traffic assignments. Statewide models tend to have several trip purposes, covering both the traditional urban trip purposes and assorted long distance trip purposes. A few states have been able to use Fratar factoring for trip distribution, because of the availability of OD matrices for certain trip purposes. Otherwise, the various steps of passenger components tend to be similar in structure to those found in urban models. There are two fundamentally different styles of freight forecasting: (1) direct forecasting of vehicle flows without reference to commodities and (2) forecasting of commodities, and then using the commodity flow forecast to estimate vehicle flows. Three-fourths of states re- porting freight components base their forecasts on commodities. Some states have explored innovative methods, such as estimating OD tables from traffic counts to fill in gaps in sec- ondary data sources. Only three states reported using mode split expressions for freight. Most states rely on the historical share of tonnage carried by each freight mode. Commodities carried by truck are

OCR for page 1
3 also assigned to vehicle types by fixed shares based on historical data. None of the models were directly concerned with truckrail intermodal. Statewide models are not yet fully integrated with urban models within the state. Ap- proximately half of the statewide models are capable of providing independent estimates of traffic within urban areas; however, statewide models invariably yield to urban models if there is a disagreement. Also, about half of the statewide models are capable of developing external station forecasts for urban models. Many statewide models base their zone systems and network on urban models, although simplifications are often necessary. Validation of statewide models exploits many of the same techniques and data sources as urban models. However, most states do not expect their models to validate as well as urban models. Prominent validation data sources are passenger vehicle counts, truck counts, na- tional default trip generation values, OD flows from the Census Transportation Planning Package, and locally collected survey data. The five case studies help to illustrate the wide range of reasonable approaches to statewide travel forecasting. The cases studies concentrate on the more promising approaches and indicate how even modest expenditures of resources can result in powerful tools for statewide transportation planning. The Ohio case study, in particular, shows what might be accomplished when budgets and time permit a full treatment of the interaction between trans- portation supply, transportation demand, land use, and economic activity. There have been many successful applications of statewide models; however, modelers still struggle to overcome many obstacles to achieve good results within a reasonable bud- get. Ongoing problems include: Scales of statewide models; Zone systems that are coarser than urban models within a given state; Databases that are geocoded to county-sized geographical areas or larger; and Many models that are unable to do peak-hour forecasts, because intercity trip lengths are too long for the static traffic assignments currently in use.