National Academies Press: OpenBook

Statewide Travel Forecasting Models (2006)

Chapter: Chapter One - Introduction

« Previous: Summary
Page 4
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 4
Page 5
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 5
Page 6
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 6
Page 7
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 7
Page 8
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 8
Page 9
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 9
Page 10
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 10
Page 11
Suggested Citation:"Chapter One - Introduction." National Academies of Sciences, Engineering, and Medicine. 2006. Statewide Travel Forecasting Models. Washington, DC: The National Academies Press. doi: 10.17226/13958.
×
Page 11

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.

5Statewide travel forecasting models address numerous planning needs by estimating, for a future date, the num- ber of vehicles that use major transportation facilities within a state. Statewide models can encompass both pas- senger and freight issues, and provide forecasts for a vari- ety of modes including highways, urban transit systems, intercity passenger services, airports, seaports, and rail- roads. Statewide models are particularly useful for fore- casting in rural areas that are not covered by urban travel forecasting models. Statewide models provide a consistent way to forecast travel on transportation facilities across a state in a manner that reflects current understanding of travel behavior. Only about half of the 50 states have created statewide models. Most of these models resemble urban transportation planning (UTP) models in structure. However, almost all states with models have faced unusual challenges resulting from the large sizes respective of their geographic areas and the large amounts of data required to adequately describe these areas. With few exceptions, several characteristics of statewide planning demonstrate the need for distinguishing between statewide and urban models. • Statewide models cover far more land area than urban models within the same state. • Statewide models cover far more facilities than urban models within the same state. • Statewide models are often concerned with economic developments that extend well beyond the borders of a region, such as national and international trade issues and trends. • There is less experience with statewide models than ur- ban models. • There is less research on intercity travel patterns than on urban travel patterns. • Statewide models must incorporate long distance, mul- tiday trips. • Statewide freight components require recognition of many modes, whereas most urban freight components only focus on trucks. • Software products do not address the special needs of statewide models. • The legal impetus for statewide models is insufficient when compared with metropolitan planning organization (MPO) models. • There are data opportunities for statewide models that are not available for urban models, such as freight flow databases and economic forecasts for subareas. Statewide travel forecasting models seek to determine the amount and location of travel by looking at parts of the traveler decision processes. A model based on behavioral principles would differ substantially from one based entirely on empirical findings, such as growth factor methods. Nonetheless, some states feel that purely empirical models still meet their needs. MAJOR SOURCES OF INFORMATION ON STATEWIDE TRAVEL FORECASTING MODELS There are few general sources of information on statewide or intercity passenger components; however, there have been two significant NCHRP studies on statewide freight model- ing. This section highlights important historical and recent documents that have been useful to individuals or groups building statewide travel forecasting models. Appendix C is an excerpt from the literature review sec- tion of the Guidebook on Statewide Travel Forecasting (Horowitz 1999) that concerns passenger and intercity travel forecasting. Appendix D is an annotated bibliography of the statewide and national freight forecasting techniques. Refer- ences to the literature found in this section are intended to up- date these earlier literature reviews. Guidebook on Statewide Travel Forecasting 1999 The Guidebook (Horowitz 1999) was the last major reference that covered both passenger and freight components of statewide models. This resource contains extensive advice on individual steps within the models, network preparation, and data sources. A comprehensive literature review and several short case studies are included in an appendix. One chapter is devoted entirely to time series methods such as Box–Jenkins techniques. The Guidebook emphasizes three- or four-step modeling approaches. Details are given on how urban transportation mod- eling software packages could be adapted for statewide models. It recommends that freight forecasting be commodity-based, although there was some treatment of truck-only models. CHAPTER ONE INTRODUCTION

The Guidebook does not include discussions of two emerging topics: tour-based passenger components and com- bining freight and passenger components with a built-in eco- nomic activity component. Transportation Research Circular E-C011: Statewide Travel Demand Forecasting (Conference Proceedings) 1999 This specialty conference, roughly coinciding time-wise with the publication of the Guidebook, heard reports from several states about their existing statewide models and plans for new statewide models. Breakout sessions also provided research recommendations. Presentations were made by rep- resentatives from California, Florida, Indiana, Kentucky, Michigan, New Hampshire, New Jersey, Oregon. Rhode Island, Washington, and Wisconsin. NCHRP Report 260: Application of Statewide Freight Demand Forecasting Techniques 1983 This report (Memmott 1983) provided a methodology for building a freight component for a statewide model. The re- port recommended a commodity-based approach and sug- gested that commodity distribution be performed by a gravity expression. An all-or-nothing mode-split step was proposed. Modal costs were to be determined with the help of regulated tariffs then in effect. The location of commodity consumption was to be determined by an input–output (IO) model. NCHRP Report 8-43, Methods for Forecasting Statewide Freight Movements and Related Performances 2005 This draft report (Cambridge Systematics, Inc., et al. 2005) is a comprehensive reference on the current state of the prac- tice in statewide freight forecasting models. It describes sev- eral approaches to model building, depending on project needs and data availability, including the direct facility flow factoring method, origin–destination (OD) factoring method, truck model, four-step commodity model, and economic ac- tivity model. Ten case studies of freight components are pre- sented: Minnesota Trunk Highway 10 truck trips, Florida ports, Ohio’s interim freight component, FHWA’s Freight Analysis Framework (FAF), New Jersey truck trip table, Southern California Association of Governments heavy-duty trucks, Indiana commodity transport, Florida statewide freight, Cross-Cascades Corridor Analysis Project, and Ore- gon combined passenger and freight. Statewide Travel Demand Models Peer Exchange 2004 The attendees of the Peer Exchange (Longboat Key, Florida, September 23–24, 2004) included representatives from 14 states and consultants who reported on their statewide model- ing efforts. The proceedings are distilled from a questionnaire 6 completed by many of the representatives. States reporting on their models were Florida, Indiana, Kentucky, Louisiana, Massachusetts, Missouri, New Hampshire, New Jersey, Ohio, Oregon, Virginia, and Wisconsin. The proceedings of the Peer Exchange have been published by TRB as Transportation Research Circular E-C075. Presentations at the 2004 Annual Meeting of the Transportation Research Board, Session on Statewide Travel Forecasting Models This session involved presentations by six authors on new developments in statewide travel forecasting models. • Ohio Statewide Travel Demand Forecasting Model • An update on the Transportation and Land Use Model Information Project (TLUMIP) in Oregon • Wisconsin Statewide Model • Statewide Modeling: The New Frontier • The Trouble with Intercity Travel Demand Models • A Brief Synthesis of the State of the Practice in Statewide Travel Forecasting. The PowerPoint slides can be found on the Statewide Travel Forecasting website: http://www.uwm.edu/~horowitz/ statewide.html. On-Line Documents About Statewide Travel Forecasting Models Numerous states or their consultants have web pages that contain documents about their statewide travel forecasting models. These web pages are volatile, and a fresh web search is required to find the most current information. Here are a few web pages that were active at the time of this report. • Florida Statewide Freight Model: webservices. camsys.com/freightmodel/freightmodel.htm. • Vermont Statewide Model: http://www.aot.state.vt.us/ planning/TDModel.htm. • Virginia Statewide Model: http://www.wilbursmith.com/ vdotmodel/howandwhen.html. • Virginia Statewide Freight Model: http://www.wilbur smith.com/vdotmodel/attachments/082902/Freight% 20Report%20(Draft%2008-20-02).pdf. • Ohio Statewide Model: http://www.dot.state.oh.us/ urban/AboutUs/Statewide.htm. • Connecticut Statewide Model: http://www.ct.gov/dot/ cwp/view.asp?a=1383&q=259806. • Oregon Statewide Model: http://www.oregon.gov/ ODOT/TD/TP/TMR.shtml. Quick Response Freight Manual 1996 Although not specifically for statewide models, the Quick Response Freight Manual (QRFM) (Cambridge Systematics, Inc., et al. 1996) (Travel Model Improvement Program,

7FHWA) has been used by states to implement the truck mode within a freight component. The QRFM provides default co- efficients for trip generation and trip distribution steps. NCHRP Report 187: Quick-Response Urban Travel Estimation Techniques and Transferable Parameters: User’s Guide 1978 and NCHRP Report 365: Travel Estimation Techniques for Urban Planning 1998. NCHRP Report 365 (Martin and McGuckin 1998) is essen- tially an update of NCHRP Report 187 (Sosslau et al. 1978). Although not specifically for statewide models, these two re- ports have allowed states to quickly implement passenger components when there were data deficiencies as to local travel patterns in urban areas. The reports provide transfer- able parameters for trip production estimation, trip attraction estimation, gravity expressions for trip distribution, time-of- day, automobile occupancy, and delay calculations. RECENT RESEARCH ON UNITED STATES INTERCITY TRAVEL FORECASTING “Critical Review of Statewide Travel Forecasting Practice” 1999 This article by Horowitz and Farmer (1999) is based primar- ily on the literature review section of the Guidebook. It offers suggestions for areas where statewide travel forecasting models can be improved. “The Trouble with Intercity Travel Demand Models” 2004 Miller (2004) critically reviews the literature on intercity passenger demand modeling. The article particularly con- trasts models of total demand with nested logit algorithms. Also described are the issues involved in applying intercity passenger demand models. “Evaluating Role of Distance and Location in Statewide Travel Demand Forecasting by Using American Travel Survey” 1999 O’Neill et al. (1999) present average distances of person travel, cross-tabulated by purpose and mode for California, Colorado, Florida, Massachusetts, and Michigan. Modes in- vestigated were personal vehicle, air, bus, train, and water. As expected, the study found that trips by air were much longer than the other modes; however, there was no clear break point of trip length that separated modes. “The Land Development Module of the Oregon2 Modeling Framework” 2004 Hunt et al. (2004a) explains one of the seven modules of the Oregon2 statewide travel forecasting model. The model allocates activities to grid cells (30 m × 30 m) once each year until reaching the planning horizon. “Driving to Distractions, Recreational Trips in Private Vehicles” 2000 Mallett and McGuckin (2004) present descriptive statistics of recreational trips by private automobile from the 1995 American Travel Survey (ATS) and the National Personal Transportation Survey (NPTS) from 1990. Comparisons were made of both urban and long distance recreational trips across racial and income groups. “Modeling the Competition Among Air Travel Itinerary Shares: GEV Model Development” 2005 This article by a research group from Northwestern Univer- sity (Coldren and Koppelman 2005) presents results from the creation of an itinerary share prediction model for air travel. Both multinomial logit expressions and nested logit algo- rithms were created to forecast choices of travelers when booking air travel based on service characteristics such as number of stops, connection quality, distance, competing carriers, aircraft type, and time of day. MAJOR DATABASES OF PARTICULAR INTEREST FOR STATEWIDE TRAVEL FORECASTING American Travel Survey 1995 The ATS was conducted in 1995 and early 1996 by the Bureau of Transportation Statistics (BTS). It is the only comprehensive national database on long distance (more than 100 mi) passenger travel. Approximately 54,000 households provided information, with each household re- porting on one year of travel in four quarterly surveys. Data about each trip include the reason for making the trip, prin- ciple mode (including vehicle type), mode of access or egress, origin, destination, intermediate stops, travel dates, duration, nights away from home, type of lodging, and travel distance. Origins and destinations are geocoded to states and metropolitan areas. Most surveys were obtained by telephone, although some personal visits were made. Individual trip records and complete household data are available on CD-ROM. There are no immediate plans to do another long distance survey similar to the ATS, although some information on long distance travel can be obtained from the NHTS. National Household Travel Survey Formerly known as the National Personal Transportation Survey, this survey of passenger travel has been conducted at varying times since 1969, with the last survey completed in 2002. Approximately 66,000 households were surveyed, of which about 40,000 were from 9 specific geographic ar- eas (who requested add-on samples) and the remainder was

a general coverage of the entire United States. Households were sampled by means of random-digit dialing and were interviewed by telephone. Data on all trips in a household over a 24-h period were collected as were data on long distance trips, defined as greater than 50 mi, over a 28-day period. Individual household and trip records are available on CD-ROM from the BTS. Daily trip data include trip times, modes, purposes, vehicles used, durations, lengths, day of the week, and the presence of other travelers for the same trip. Long distance trip data include dates of travel, whether the trips are recurring, purposes, primary modes, destinations, types of lodging, overnight stops, and access and egress information for air, bus, and rail modes. It should be noted that the definition of “long distance” is different from that used by the ATS; therefore, the data sets are not directly comparable. The results of the survey may have been affected by the September 11, 2001, terrorist attacks. Individual trip records from the NHTS are available and are easy to summarize or analyze. Much of the transferable parameters in NCHRP Report 365 were developed from the 1990 National Personal Transportation Survey. Planning is currently underway for the next survey in 2008. More infor- mation may be obtained from http://nhts.ornl.gov/2001/ html_files/introduction.shtml. Commodity Flow Survey The Commodity Flow Survey (CFS) is a survey of shippers in the United States. Shipments from most major industries are represented in the sample, last taken in 2002. It was composed of a stratified random sample of approximately 50,000 establishments with 2.6 million shipments. Estab- lishments reported a sample of their shipments (or all ship- ments for smaller establishments) for one week in each of four calendar quarters. Information about each shipment in- cluded the origin, destination, value, weight, mode, dis- tance estimated from a network, and commodity group. Modes covered by the survey included for-hire truck, pri- vate truck, rail, inland water, deep sea water, pipeline, air, and parcel delivery or U.S. Postal Service. Data are also available from the 1997 and 1993 surveys. The CFS does not contain data on imports, and its level of spatial detail is coarse. Industrial sectors included mining, manufacturing, wholesale trade, electronic shopping, and mail-order busi- nesses. The survey excluded services, transportation, construction, other retail, farms, fisheries, gas and oil extraction, and most government-owned establishments. The U.S. portions of imports that are transshipped from within the United States are included. Shipments passing entirely through the United States are excluded. Detailed tables can be obtained on CD-ROM from the BTS. Plan- ning is underway for the next CFS in 2007. More informa- tion on the survey may be found at http://www.bts.gov/ programs/commodity_flow_survey/. 8 Vehicle Inventory and Use Survey The Vehicle Inventory and Use Survey (VIUS), formerly known as the Truck Inventory and Use Survey, consists of data on the operation and physical characteristics of commercial ve- hicles. The survey was first done in 1963, and is currently con- ducted every 5 years. The latest survey was done in 2002, with the next on schedule for 2007. Operating characteristics include number of miles driven and commodities carried. Individual truck records are available for almost 100,000 trucks. Opera- tional characteristics that are of general interest for travel models are base state, average weight with payload, type of business, miles driven outside state, miles driven by trip length, miles driven by commodity group (50 groups including empty and waste), miles driven by hazardous materials class and type of service. VIUS data may be obtained from the U.S. Census Bureau. More information about VIUS may be obtained from http://www.census.gov/svsd/www/tiusview.html. Transborder Surface Freight Data The Transborder Surface Freight Data set is a large sample of shipments between the United States and Canada and the United States and Mexico. Freight flow data in dollars and tons are provided by destination state or origin state, by point of entry or exit, by commodity and by mode (mail, highway, rail, vessel, and pipeline). Data are updated monthly. Indi- vidual shipment records may be obtained. More information on the Transborder Surface Freight Data may be obtained from http://www.bts.gov/transborder/. Freight Analysis Framework The FAF, developed by FHWA, is a modeling system that forecasts the amount of freight traveling on modal (truck, wa- ter, and rail) networks throughout the United States. It is pri- marily a policy tool for the federal government. Forecasts for 2010 and 2020 have been made. Results for commodities are reported at the two-digit Standard Transportation Commodity Code (STCC) level. The model itself and much of the input data are not available for state use. However, the FAF provides the following results for its base year (1998) and forecast years that can be of use to statewide travel forecasting models: • Tons of freight shipped in the United States by state or international gateway, type of commodity, and mode of transportation; • Flows of freight along major routes by range of tonnage and mode; and • Number of trucks using road segments. The FAF is currently undergoing major revisions to provide additional detail and to make its results more useful. Results are downloadable from the FHWA website. More information on the FAF may be obtained at http://ops.fhwa.dot.gov/freight/ freight_analysis/faf/. This site explains planned revisions to the

9FAF. The FAF has recently been reviewed by the TRB Com- mittee on the Future of the FHWA’s Freight Analysis Frame- work (Meyburg 2004). Census Transportation Planning Package The Census Transportation Planning Package (CTPP) is a special tabulation of the decennial census that reports data by traffic analysis zone (TAZ), both rural and urban. By analy- sis of the journey-to-work questions, the CTPP provides in- formation on home-to-work flows, modes of travel to work, ridesharing to work, vehicle availability, commute times, and employment counts at the workplace. Demographic data are tabulated by both place of work and place of residence. The CTPP is a valuable source of information about employment in the workplace by industrial sector and is available on CD- ROM from the BTS. CTPP is usually available approxi- mately 4 years into the decade. More information may be found at http://www.fhwa.dot.gov/ctpp/. Public Use Microdata Sample The Public Use Microdata Sample (PUMS) is an output of the decennial U.S. Census. The sample contains 5% of all house- hold records from the long form that have been cleaned of any identification and geocoded coarsely to special zones called PUMAs (Public Use Microdata Areas), which are areas with at least 100,000 persons. PUMS allows special tabulations that are not normally available for a metropolitan area, county, or state. Many planning agencies use PUMS to understand household structure when building trip generation models. ES-202 ES-202 is a cooperative program where states report infor- mation on employment that is derived from the states’ un- employment insurance programs to the federal government. ES-202 data can provide employment at the workplace by in- dustrial category; however, federal rules dictate that the con- fidentiality of the data must be respected. The quality of the data and the ways in which it is administered differ across states. As with all secondary sources of employment data, the geocoding of work locations requires considerable cleaning and verification. The most serious problem for travel fore- casting is that the mailing addresses of employers do not nec- essarily agree with the addresses of the actual workplaces. ES-202 data may be available from a state’s labor or em- ployment agency. The data set is continuously updated. National Networks The National Highway Planning Network is available as geo- graphic information system (GIS) layers and contains a topological description of 450,000 mi of arterial highways in the United States. A similarly comprehensive GIS database is available for the National Rail Network. Networks are avail- able by state, and are obtainable from the BTS on the National Transportation Atlas Database CD-ROM. This CD-ROM also contains the U.S. Army Corps of Engineers Navigable Waterway Network, hydrographic features, fixed-guideway transit networks, runways, seaports, Amtrak stations, airports, intermodal terminals, and jurisdictional boundaries. Railroad Carload Waybill Sample Railroads doing business of more than 4,500 carloads per year are required to submit a sample of their waybills to the Surface Transportation Board. Waybills contain information on origin and destination points, type of commodity, number of cars, tons, revenue, length of haul, participating railroads, interchange locations, and cost. Publicly available data from the sample are geocoded to Bureau of Economic Analysis (BEA) regions; however, commodities are reported to five- digit STCC. The 2002 sample contains information on nearly 600,000 shipments from 66 railroads. The confidential data are available to a single point of contact with a state govern- ment, often an agency that regulates railroads; therefore, it is possible for a state department of transportation (DOT) to gain access to data with precise geocoding. Strict rules apply to disseminating data outside of state government. Data within the waybill about revenue from a shipment are con- sidered to be inaccurate by the Surface Transportation Board. COMMERCIAL DATABASES AND FORECASTS IN USE BY STATES The following commercial databases and forecasting ser- vices were specifically mentioned by states when answering questions about their models. This section is included to am- plify on state responses and is not intended to be a compre- hensive listing or review of such databases. Reebie TRANSEARCH Reebie can supply data on multimodal commodity flows be- tween locations in the United States at a greater level of spa- tial detail than the CFS. Reebie integrates data from both public and proprietary data sources. Commercial Demographic Forecasts Companies such as Global Insight (formerly WEFA and DRI) and Woods & Poole can provide economic and demo- graphic forecasts by county. Similar forecasts may be avail- able for certain states through universities or state agencies. Commercial Employment Databases D&B (formerly Dun and Bradstreet) maintains a compre- hensive database of U.S. companies and their characteristics.

Info USA is a mailing list company that tries to maintain a complete list of businesses in the United States. Claritas pro- vides demographic data. All three companies are possible sources of data on employment at the workplace. Regional Economic Model, Inc. Regional Economic Model, Inc. (REMI), essentially has two products, Policy Insight and TranSight. Policy Insight is designed to forecast the economic impacts of major govern- mental policy initiatives. TranSight specifically forecasts the economic impacts of transportation projects. DEFINITIONS OF COMMON TECHNICAL TERMS USED TO DESCRIBE STATEWIDE TRAVEL MODELS Technical Concepts All-or-nothing traffic assignment—A model step where all traffic between an origin and destination is assigned to the shortest path between that origin and destination and no traffic is assigned to any other path. An all-or-nothing traffic assignment is unresponsive to delays caused by traf- fic. Historically, many statewide models have used the all- or-nothing traffic assignment because volume-to-capacity ratios were difficult to determine for 24-h forecasts and networks in urban areas were sketchy. Many freight compo- nents still use all-or-nothing assignment to preload trucks to a highway network. BPR curve—A simple expression that computes travel time as a function of volume, originally developed at the Bureau of Public Roads (BPR). A BPR curve has two para- meters, α and β, that can be varied by functional class: where t0 is free flow travel time, v is the assigned volume, and c is the capacity. The BPR curve is used within a traffic as- signment step to provide loaded travel times so that traffic can be placed on the shortest path. Commodity group—A grouping of similar commodities that can be analyzed and forecasted together. The groupings are often based on the Standard Classification of Transported Goods or the older STCC. Standard Classification of Trans- ported Goods codes are of up to five digits, organized such that adding a digit increases the precision of the commodity description. Composite impedance (or composite disutility)—A measure of the separation between an origin and a destination (often as a function of travel time, travel cost, and convenience) that takes into consideration the accessibility of more than one mode between the origin and destination. Composite impedances are t t v c = + ⎛⎝ ⎞⎠⎡⎣⎢ ⎤ ⎦⎥0 1 α β 10 often used along with gravity expressions. The following equa- tion shows a composite impedance expression, tij, for two modes (1 and 2) between origin zone i and destination zone j. The empirical constant, θ, is usually of similar size to the in-vehicle time coefficient from a logit mode-split expression, provided the impedance has units of minutes. Composite impedances are especially important for statewide and intercity models, where the travel times by various modes can differ radically, but trip distribution must be accom- plished ahead of (before knowing) mode split. Dynamic all-or-nothing assignment—See “all-or-nothing traffic assignment.” Trips are assigned within small intervals of time so as to track the progress of packets of vehicles over time between their origins and destinations. The principle ad- vantage of dynamic traffic assignment for statewide models is an ability to determine the amount of traffic that occurs during peak hours within urban areas. Dynamic equilibrium traffic assignment—An application of equilibrium principles (see static equilibrium traffic as- signment) where trips are also assigned within small inter- vals of time, so as to track the progress of packets of vehicles over time between their origins and destinations. A single dynamic equilibrium traffic assignment requires several dynamic all-or-nothing assignments (see “dynamic all-or- nothing assignment”). Four-step model—A modeling paradigm that has become standard practice in urban areas and involves the major steps of trip generation, trip distribution, mode-split, and traffic as- signment. A common variation is a three-step model that eliminates the mode-split step. A four-step model may in- volve minor steps, including time-of-day and automobile-oc- cupancy calculations. Fratar factoring—A popular empirical technique for fore- casting origin-to-destination trip patterns by applying row and column factors to an existing origin-destination (OD) table. Fratar factoring can also be applied to a production-to- attraction trip table. GPS (Global Positioning System)-based survey—Use of the GPS to trace the location of a traveler or vehicle over time, which would be linked to a travel diary. Gravity expression—Sometimes called a “gravity model,” which determines the production-to-attraction trip pattern as a function of the number of productions and attractions in each zone and measures of proximity between zones. Grav- ity expressions can be either singly or doubly constrained. A singly constrained gravity expression holds productions by zone constant, allocating trips to other zones on the basis of a measure of their zonal attractiveness. A doubly constrained t e eij tg tg = +[ ]1 1 2θ θ θln

11 gravity expression, often used in statewide models, allocates trips between zones while also holding trip productions and trip attractions constant. A typical gravity expression finds the number of trips, Tij, between production zone i and attraction zone j: Tij = PiXiAjYj f(tij) where Pi is the number of productions in zone i, Aj is the number of attractions in zone j, and f(tij) is a measure of proximity between zones i and j, as a function of impedance, tij, between zones. The measure of proximity, f(tij), is often called a friction factor. Xi and Yj are balancing factors that are set such that the numbers of productions and attractions, re- spectively, are conserved in each zone. Some implementa- tions of the gravity expression have a term, kij, which are called “k-factors” or “socioeconomic adjustment factors.” k-factors are empirical adjustments to the gravity expression based on household travel surveys or screenline counts to provide a better fit between the model and base-year data. Household sectors—Groups of households within an eco- nomic or land use model, usually organized by economic or life-cycle status. Industrial sectors—Groups of similar businesses, usually organized by type of product or service. Industrial sectors are often defined according to North American Industry Classi- fication System or the older Standard Industrial Classifica- tion codes. North American Industry Classification System codes are of up to six digits, organized such that adding a digit increases the precision of the industry description. Input-output (IO) model—A type of economic model that tracks flows of revenue (or sales) between industries and households in a national or regional economy. An IO model is organized by industrial sectors. A single cell in an IO table would list the amount of revenue gained by a producing sec- tor from sales to a consuming sector. Logit expression—Sometimes called a “logit model,” this is a method for determining the number of people who will make a particular choice (such as mode or destination) given the “utilities” of each alternative. A logit expression deter- mines the proportion of people, pi, who choose an option, i: where Ui is the utility of option i, and an option can be either a mode or a destination, or both. Utility is usually taken to be a linear combination of travel time, travel costs, and mea- sures of convenience, such that Ui becomes more negative or less positive as trip lengths increase. When used for destination choice, a logit expression is a form of a singly constrained gravity expression. Logit p e e i U Uk k = ∑ 1 expressions are preferred for activity allocation within land use components of integrated models. Logit expressions can replace gravity expressions for trip distribution within a tra- ditional four-step model. When doing so, the number of trip attractions in a zone is calculated by the expression rather than given as an input. Therefore, logit expressions are more sensitive to changes in policies and infrastructure. There is no consensus as to when logit expressions are preferred over gravity expressions for trip distribution. Microscale traffic simulation—Sometimes called traffic microsimulation, this is traffic simulation that tracks the lo- cation and performance of individual vehicles. Microscale traffic simulations can be used as post-processors for output from a statewide travel forecasting model, so as to provide better estimates of delay from the assigned traffic. Monte Carlo simulation—A technique that is used within microsimulation that can generate random events, such as households of given characteristics, trips, start times, modes, and vehicles. The probability of an event is taken from his- torical information or from theory, such as a logit expression. Multiclass assignment—A method of traffic assignment that separately accounts for different vehicle classes. Differ- ent vehicle classes may be assigned to different routes if the link impedances vary across vehicle types. Multiclass assign- ments may take many different forms, static or dynamic and all-or-nothing or equilibrium. Multiclass assignments also account for the differential impact heavy vehicles have on the traffic stream. Multiclass assignment can be used to distin- guish automobiles from trucks and buses, single-occupant automobiles from multiple-occupant automobiles or low- income drivers from high-income drivers. Nested logit algorithm—The use of two or more logit ex- pressions to determine the number of people who will make a particular choice when the decision process is assumed to consist of a sequence of preliminary choices. Nested logit al- gorithms are organized as a hierarchy, such that modes be- come more specialized in the lower parts of the hierarchy. Similar modes tend to be grouped together into “nests.” Travelers are assumed to make decisions between nests be- fore making decisions about the individual modes within nests. A utility for a nest is created as a composite of utilities of all modes within a nest (see composite impedance). Origin–destination (OD) table estimation from ground counts—A method of determining the OD patterns of vehi- cles by primarily using observations of ground counts. OD table estimation usually requires a good guess as to the OD patterns, often referred to as a “seed” or “prior” table. The es- timation algorithm tries to make limited improvements to the seed table, so that the assigned volumes will be closer to ground counts. There are many mathematical formulations to the OD estimation problem, and the various formulations will result in different OD tables using the same data. For

example, a simple generalized least-squares approach at- tempts to minimize this expression: where Va is a ground count for link direction a, Tij is the number of trips between origin i and destination j to be estimated, Tij* is the seed trip table, paij is the proportion of trips between zones i and j that use link direction a (as determined by an equilibrium traffic as- signment), N is the number of zones, wa are link weights, z is the trip table weight, and s is a single factor that is either set to 1 or selected by soft- ware to scale the trip table to produce the correct average traffic count. Each direction of a two-way link, a, may have a separate ground count. Tij is constrained to be no smaller than zero (i.e., cannot be negative), otherwise Tij is unbounded. It is readily apparent that there are as many variables in the constrained minimization problem as there are cells in the OD table. Thus, computation times can be long for large net- works. Many formulations and algorithms have been pro- posed to accelerate computation times. Special generator—A business or other activity site that is so large or so specialized that it should not be included in standard trip generation calculations for a traffic analysis zone. A special generator may have a separate zone in the model or its trips may be added to those coming from more general land uses in a zone. Static equilibrium traffic assignment—A method by which traffic is assigned such that travel times on links are consis- tent with volumes and volumes are consistent with travel times. A “user-optimal” equilibrium traffic assignment method, which has been implemented in statewide models, also routes each vehicle on its shortest path between an origin and a destination. The most common algorithms for static equilibrium traffic assignment require that the assigned vol- umes from several all-or-nothing assignments be averaged. Stochastic multipath assignment—Traffic between an ori- gin and destination is divided across many paths between that min P w V s p T za a ija ij j N i N a A = − ⎛ ⎝⎜ ⎞ ⎠⎟ + === ∑∑∑ 111 2 T sTij ij j N i N * −( ) == ∑∑ 2 11 12 origin and destination, with the shortest path usually getting the largest share. Transshipment—Goods shipment with multiple legs of the journey, with short-term storage between the legs, either in warehouses or at terminals. Alternatively, transshipment can refer to importing goods from one country (e.g., China) that pass through another country (e.g., Canada) on its way to the destination (e.g., United States). Notes About Terminology Used in This Synthesis There is a tendency among those involved in building travel forecasting models to use the word “model” to describe var- ious pieces of a model as well as the whole modeling frame- work. For example, planners often refer to a trip distribution technique as a “gravity model” or a mode-split technique as a “logit model.” To help distinguish between various parts of a model, the following terms are used herein. Algorithm—A series of expressions or computational processes that produces a specific result within a step. An example of an algorithm is path building. Component—A collection of steps that leads to a particu- lar result. Most statewide models have two components: passenger and freight. Expression—A single equation that yields a single answer. For example, mode-split steps might be built around a logit expression, which itself contains a utility expres- sion. Model—The whole modeling framework, including soft- ware, databases, components, steps, algorithms, and expressions. A model excludes the personnel neces- sary to operate it or to interpret its results. These personnel would be included into the “modeling process.” Software—Models require software for their implementa- tion. There are three major classes of software: statistical estimation software, travel forecasting modeling software, and GIS. Although different com- mercial software packages have distinguishing fea- tures, many are sufficiently general and flexible to meet the needs of statewide travel forecasting. Thus, this syn- thesis avoids mentioning or endorsing specific software products. Software can also be custom written for a model. Step—A series of expressions or algorithms that represents a behavioral process within a component. An example of a step is mode split.

Next: Chapter Two - Survey of Statewide Travel Forecasting Practice »
Statewide Travel Forecasting Models Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Synthesis 358: Statewide Travel Forecasting Models examines statewide travel forecasting models designed to address planning needs and provide forecasts for statewide transportation, including passenger vehicle and freight movements. The report explores the types and purposes of models being used, integration of state and urban models, data requirements, computer needs, resources (including time, funding, training, and staff), limitations, and overall benefits. The report includes five case studies, two that focus on passenger components, two on freight components, and one on both passenger and freight.

  1. ×

    Welcome to OpenBook!

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

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

    No Thanks Take a Tour »
  2. ×

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

    « Back Next »
  3. ×

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

    « Back Next »
  4. ×

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

    « Back Next »
  5. ×

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

    « Back Next »
  6. ×

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

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

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

    « Back Next »
Stay Connected!