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.
6 Advancing the State of the Practice I ncremental improvements can be made to the conventional travel models without changing their basic structure or approach to travel demand fore- casting. Some metropolitan planning organizations (MPOs) and other agen- cies, however, are experimenting with or have adopted fundamental changes in travel modeling that may significantly expand the applications of current models (VHB 2006). Because many of these advanced modeling practices have been implemented only recently, there is no consensus yet that they should be widely adopted. Since these practices are tied more closely to house- hold and traveler characteristics and behavior, they should in concept permit MPOs to address policy questions that cannot be treated with the conven- tional four-step models. Yet some practitioners remain unconvinced that their adoption is warranted in view of the perceived costs and difficulties associated with their implementation. This chapter addresses in turn improvements in four-step trip-based modeling; advanced modeling practices; the TRANSIMS system; experience with advanced practice; obstacles to model improvement; and model research, development, and implementation. IMPROVEMENTS IN FOUR-STEP TRIP-BASED MODELING Many improvements in the four-step process can be and have been made. Often these improved approaches become possible when application pro- cedures are implemented in one of the several commercially available soft- ware packages. These approaches may be conceptually appealing and should contribute to better forecasts. Indeed, some of the approaches reported by agencies do lead to better replication of observed patterns; however, few if any systematic studies have demonstrated that they lead to better fore- 90
Advancing the State of the Practice 91 casts. The following are some illustrative improvements to the four-step process: ⢠Improved measures of arterial congestion: The âBPR (Bureau of Public Roads) curveâ has been used for years to estimate congestion and delay. It yields good responses for freeways but has been viewed as lacking for arterial roadways, where intersection delay and queuing are major factors. Newer approaches now used by some MPOs estimate congestion on the basis of modeled delay at arterial intersections. ⢠Inclusion of both highway and transit travel in trip distribution: Trip dis- tribution, the second step in the four-step process, involves allocating travel among analysis zones. In areas with significant transit use, it is thought that trip distribution patterns should reflect not only highway but also transit travel times and costs. A number of agencies have implemented distribution models with this feature. ⢠Improved trip distribution models: âDestination-choiceâ models are an alternative to gravity models. They take into account characteristics of both trav- elers and their possible destinations in allocating travel among analysis zones and reduce the need to use arbitrary factors to match traffic counts. Such mod- els have been developed and applied by MPOs. In the early 1990s, destination- choice models were considered advanced practice; this remains true today. Deakin and Harvey (1994, 43) note that âthe aggregate gravity-type model remains deeply ingrained in practice despite its apparent disadvantages.â ⢠Improved modeling of nonmotorized travel: To incorporate bicycling and walking into the modeling scheme, some MPOs are introducing a high degree of spatial resolution into the model system since the measurement of small-scale accessibility is essential. One method that can be used for this purpose is to reduce zones to a size that can reflect meaningful walking dis- tances between zones. Walking distances should be no more than 0.5 mile between zone centroids in the urban portions of the modeling area, where the walking and bicycling modes are most likely to be used. Another method is to use geographic information systems to measure accessibility from a zone centroid (e.g., number of retail employees within 0.5 mile, number of house- holds within 10 minutes). With the ability to measure accessibility at a non- motorized level, variables that potentially influence the decision to walk and bike can be identified. Examples of typical variables are accessibility to jobs, shopping opportunities, and households. Other relevant variables are house- hold socioeconomic characteristics (e.g., automobile ownership, number
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 92 of workers) and intersection density (i.e., ease of crossing streets). If the city or region has household survey data that capture travel information for all modes, models that address the full spectrum of travel options can be specified. ⢠Improved sensitivity testing: Models are used to project the responses of travelers and the transportation system to changes but have often been vali- dated only on the basis of replication of observed conditions. Some MPOs, such as that of Las Vegas, have applied a technique that involves varying properties of the system (e.g., the population or employment in a zone, the capacity of a road) and examining the forecast response (Fehr & Peers 2005). While there is no way of ascertaining whether the forecast response is cor- rect, analysts can assess whether it is reasonable or explainable given what is known about traveler behavior. MPOs may undertake ambitious modeling improvement programs within the framework of their current methods. Tables 6-1 and 6-2 show a work program proposed by the Sacramento Area MPO to upgrade its land use and travel models to better represent user needs (DKS Associates 2001). ADVANCED MODELING PRACTICES It has been asserted that travel forecasting cannot be truly improved until the underlying paradigms reflect more fully the requirements and decision pat- terns of households, the interactions among the patterns of the various mem- bers of households, and household needs over more than a single day (McNally 1997; Boyce 2002). Travel models based on a more comprehensive under- standing of the activities of households would better reflect the full range of trade-offs that affect whether to make a trip, what time a trip is made, the destinations visited, the modes used, and the paths selected. Also needed is a more complete representation of the supply-side network to account for the details of congested operations throughout the day. No one new modeling approach can address all these needs. Rather, a suite of related approaches, taken together, shows promise for greatly improving modeling practice. These approaches are referred to here as âadvanced modeling practicesâ or advanced models. The readiness of advanced models for wider application is the subject of debate among travel forecasters. Some practitioners argue that the benefits to be derived from the apparently more complex and data-intensive procedures
TABLE 6-1 Example Land Use Model Elements and Upgrades to Address User Needs: Sacramento Area Council of Governments Element Current Versus Upgraded Practice Base-year Current practice: Track housing unit completions; apply vacancy population rates. Tallied by SACOG minor zone. To address user needs: More detail on household structure (size, workers, life cycle, etc.) and location. Base-year Current practice: Track job locations by situs address and SIC employment code. Tallied by SACOG minor zone. To address user needs: More detail on location. Ideally, more detail on employment types. Shifts in population Current practice: Allocate population growth to minor zone. demographics Rule-based cross-classification to persons, workers, and income. over time To address user needs: Forecast detailed household characteristics on the basis of known characteristics and trends. More geographic detail needed. Shifts in size and Current practice: Based on current development trends and land structure of use policy (general plans). Constrained by population growth. economy To address user needs: Tied to changes in labor supply and the over time ability of the transportation and land use system to serve the needs of various industries. Labor marketâ Current practice: Regional employment growth parallels (and is demand and constrained by) regional household growth. supply To address user needs: Changes in employment tied to employ- ment conditions (e.g., wages) and available labor in region. Household relocation Current practice: Not addressed. To address user needs: Minimally, allocations of new households should be based on household and area characteristics and on supply and demand by area. Ideally, âmoveâ or âstayâ decision for each household is based on household characteristics. Firm/business Current practice: Not addressed. relocation To address user needs: Minimally, aggregate allocation to zones, with floor space prices adjusted to clear the market. Ideally, âmoveâ or âstayâ decision based on firm characteristics. Floor space prices Current practice: Not addressed. To address user needs: Equilibrium with floor space demand by firms and households and area supplies. Development of Current practice: Implied development of acreage based on floor space acres/job rates. To address user needs: Simulation of development probability by parcel or grid cell, with consideration of floor space prices and vacancy. Goods movement/ Current practice: Simple truck model. shipment logistics To address user needs: Simulate shipment of goods at the firm/business level. Take account of industry characteristics. Note: SACOG = Sacramento Area Council of Governments; SIC = Standard Industrial Classification.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 94 TABLE 6-2 Example Travel Model Elements and Upgrades to Address User Needs: Sacramento Area Council of Governments Element Current Versus Upgraded Practice Automobile Current practice: Cross-sectional automobile ownership model. ownership To address user needs: Enhance current model with more detailed household data, linkages to other parts of model. Ideally, include vehicle type in model. Tour/trip Current practice: Trip-based. Limited use of accessibility variables. generation To address user needs: Day pattern model with logsum feedback from lower models. Some accounting for household characteristics. Destination Current practice: Trip-based destination choice, integrated with choice mode-choice model. To address user needs: Tour-based destination choice, with intermedi- ate stops. Mode choice Current practice: Trip-based, with nonmotorized modes. To address user needs: Tour-based mode choice, with mixed-mode tours. Time of travel Current practice: Fixed factors. To address user needs: Time-choice model, sensitive to household characteristics and travel conditions. Level of spatial Current practice: Zone level for all. detail To address user needs: Some block-face level of detail needed (especially for nonmotorized travel). Network Current practice: Multiclass equilibrium for highway; shortest-path simulation/ AON for transit. Nonmotorized travel not assigned. route choice To address user needs: More classes needed, especially for transit. Ability to assign nonmotorized trips. Ideally, network microsimulation. Application Current practice: Zone-based enumeration by originâdestination, framework mode, purpose, and time of day. To address user needs: Person-based and firm-based enumeration, to track demographic characteristics with travel. External and Current practice: Fixed matrices. special trips To address user needs: Airport access model needed. Interregional travel keyed to growth in neighboring regions. Note: AON = all-or-nothing assignment. have not yet been demonstrated and may not be worth the effort. On the other hand, many members of academia and some others assert that advanced models have been implemented, that the major barriers to implementation have been resolved, and that the use of such models should permit agencies to develop better forecasts.
Advancing the State of the Practice 95 Following is a discussion of advances that go beyond the prevailing four- step modeling paradigm. Improved Land Use Modeling Planning agencies have been considering for years how best to reflect the inter- actions between transportation investment decisions and land development pat- terns. For a number of MPOs, various forms of land use models are now part of the routine process for analysis of growth, allocation of growth, and study of the land use impacts of alternative transportation investment programs. Miller et al. (1999) suggest that MPOs wishing to analyze land useâtransportation interactions should consider adopting a land use model for their analyses. Land use models have a long history of evolution and application in the United States and elsewhere. A recent innovation is the acceptance and use of âintegrated urban modelsâ that combine advanced land use and transportation models to better represent the interactions between transportation and land use. A variety of land use models are in operational use. While differing in their details and their relative strengths and weaknesses, they demonstrate that land use models can be applied successfully in practice. The models do, however, require significant investment in data assembly, model development, and tech- nical support staff. Given the diversity of urban regions and associated planning needs, it is unlikely that a single standardized modeling methodology will emerge. The more likely scenario is that diverse methods will be employed that share common objectives (credible projection of future land uses) and principles (e.g., sensitivity to transportation system effects, appropriate treatment of real estate market processes). Miller et al. (1999) provide guidance for how to imple- ment a land use modeling capability within an MPO or other agency concerned with undertaking integrated analysis of transportation and land use policies. Tour-Based Models Tour-based modeling recognizes that travelers may have multiple purposes and multiple stops within each tripâthus a âtour.â This is a significant advance over the four-step trip-based approach, which aggregates trips from zone to zone according to such purposes as âhome to work.â Tour-based modeling has been applied by a few MPOs and can be an important step toward full activity-based modeling (VHB 2006).
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 96 Activity-Based Models Activity-based models differ from previous travel forecasting methods in con- cept and structure. The approach recognizes the complex interactions between activity and travel behavior. The conceptual appeal is that the need and desire to participate in activities form the basis of the model. By emphasizing partic- ipation in activities and focusing on sequences or patterns of activity, such an approach can address complex issues (Bhat and Koppelman 2003). The dif- ferences between activity-based models and the current four-step approach include a consistent and continuous representation of time, a detailed repre- sentation of persons and households, time-dependent routing, and micro- simulation of person travel and traffic. Activity-based models require more detailed information about population demographics than is available from surveys or the Census Bureau. âPopulation synthesizersâ have been developed so that available data can be used to extrapolate synthetic populations that are statistically equivalent to actual populations. They can also apply land use data to locate all households relative to the transportation network (Hobeika 2005). Regional-scale traffic microsimulation is an end product and major con- tribution of the federal TRANSIMS project (discussed below) and other activity-based models as well. The static assignment of current MPO models is replaced by a process that addresses such traffic effects as queuing and upstream effects of congested links. Motor vehicle emissions, for example, cannot be adequately estimated by static assignment outputs; microsimulation or dynamic network loading is needed. Discrete-Choice Modeling Travel decisions are made by individuals, not by traffic analysis zones (Domencich and McFadden 1975). While there can be benefits to aggrega- tion when all aspects of decision processes cannot be accounted for, model results will be improved to the extent that model sets can more clearly repre- sent both choices available to travelers and decision factors relevant to indi- vidual travelers. Discrete-choice methods have been used for many years for the development of mode-choice models and are increasingly used for the development of destination-choice models. Discrete-choice methods have not been widely used for the application of models. As synthetic populations are increasingly used for forecasting households, the use of discrete choice for model application will become more attractive.
Advancing the State of the Practice 97 Supply-Side Models Advanced computerized traffic models that provide greater temporal and operational detail have been developed. They have the potential to be com- bined with conventional or advanced travel demand models, although properly integrating such advanced supply models with demand models may require coding a more detailed highway network that includes facili- ties carrying local traffic and intersection control information. Integrating transit supply and transit demand models poses a more challenging task because of the temporal variations in transit routes and schedules and the unavailability of transit at certain times of the day. Following are descrip- tions of two supply models that hold promise for integration with travel demand modeling. Traffic microsimulation is the modeling of individual vehicle movements on a second or subsecond basis for the purpose of assessing the traffic perform- ance of highway and street systems, transit, and pedestrians. Microsimulation can provide the analyst with valuable information on the performance of the existing transportation system and potential improvements. The past few years have seen a rapid evolution in the sophistication of microsimulation models and a major expansion of their use in transportation engineering and planning practices (Dowling et al. 2004). Traffic microsimulation can be combined with an activity-based travel demand model to provide a power- ful tool for forecasting and analyzing supply-side transportation system and facility performance. In addition to traffic microsimulation, methods for regional- or local-scale network dynamic traffic assignment applications have been developed. These software systems have the potential to predict where and when drivers will travel on the road network. They have great potential for operational planning, such as real-time intelligent transportation system applications. Issues exist in terms of how best to use the more aggregate, static outputs from the four-step equilibrium assignment as inputs to the more dynamic/micro models. While dynamic assignment and traffic microsimulation are more realis- tic than current static equilibrium methods, they are also computationally far more expensive. Indeed, these models generally still cannot feasibly be applied at the full urban region level with a reasonable expenditure of computation time and resources. As progress is made toward greater use of activity-based travel models, as cost-effective computing power continues to increase, and as dynamic assignment methods that run more rapidly are developed, the
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 98 gradual introduction of these methods into operational regional modeling is likely.1 TRANSIMS Starting in 1992, the federal government undertook a pioneering model development project to advance the state of the practice of travel forecasting. The initial ground-breaking work on TRANSIMS was performed at Los Alamos National Laboratory. TRANSIMS is a computer-based system for simulating the second-by-second movements of every person and every vehi- cle throughout the transportation network of a large metropolitan area. It consists of multiple integrated simulations, models, and databases. By employ- ing advanced computational and analytical techniques, it creates an inte- grated environment for analysis of regional transportation systems (Los Alamos National Laboratory 2007). TRANSIMS incorporates and integrates some of the advanced modeling practices detailed above, in particular population synthesis, activity-based modeling, and traffic microsimulation. TRANSIMS was funded primarily by congressional appropriation and administered through the Federal Highway Administrationâs (FHWAâs) Travel Model Improvement Program (TMIP). From 1992 to 2003, $38 million was spent on TRANSIMS, about three-quarters of which went to Los Alamos for basic research and development. After 2003, a 3-year hiatus occurred during which no funding was available for TRANSIMS development or imple- mentation. The Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU) allocates $2 million annually to TRANSIMS, some of which is to support implementation by MPOs and other operating agencies and some of which is to support TRANSIMS- related development activities. TRANSIMS was originally field-tested in DallasâFort Worth, Texas, and Portland, Oregon. The work in Portland stopped during the hiatus in fund- ing but is now being continued under SAFETEA-LU. SAFETEA-LU will also support two to three new deployments a year. Already funded are simulations of hurricane evacuation plans in New Orleans ($300,000, in cooperation with the Louisiana Department of Transportation and Development); a planning 1 Personal communication, E. J. Miller to J. Williams, March 4, 2007.
Advancing the State of the Practice 99 study in Burlington, Vermont ($300,000, in cooperation with the Chittenden County MPO); and simulation of freight border crossings in Buffalo, New York ($500,000). TRANSIMS technology is also being used for projects not funded through SAFETEA-LU. These include the following: ⢠Evacuation planning for Chicago, sponsored by the City of Chicago and Illinois Department of Transportation ($1.28 million); ⢠A congestion study for central New Jersey, sponsored by Rutgers University ($500,000); ⢠A study of street closings in Washington, D.C., near the White House, sponsored by FHWA ($1.5 million); ⢠A feasibility study for TRANSIMS in Atlanta, Georgia, sponsored by the Georgia Regional Transportation Authority ($50,000); and ⢠Linking of TRANSIMS with the UrbanSim land use and policy model, sponsored by the University of Vermont ($800,000). TRANSIMS has not yet been implemented by any MPOs for use in their core travel forecasting activities. There are a number of reasons for this. First, the original software evolved in a research and development setting at a gov- ernment laboratory. While suitable for use in that setting, it was not well adapted for general deployment. In addition, early versions required high- performance computers and the Linux operating system, which many agen- cies did not own or have access to. The user interface and system documentation were deficient and did not easily support applications. In addition, the capa- bility to handle transit assignment through a time-sensitive network model has not been developed. As TRANSIMS has evolved from a research concept, public perceptions have been shaped by the problems associated with the ini- tial start-up of this complex new technology. There was a perception among many practitioners that implementing TRANSIMS (or other activity-based models) might be an overwhelming task. There have been some misconceptions about TRANSIMS as well. The extent and cost of necessary data collection, computer hardware requirements, and the complexity of implementation have been exaggerated. Implementing such a new model set does require more data, staff resources, and computing power than continuing to use existing technology, but it is demonstrably achievable (see âExperience with Advanced Practiceâ below). A number
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 100 of improvements have made TRANSIMS more accessible and ready for implementation: ⢠Availability in a Windows environment; ⢠Hardware advancesâthe requisite computers can be purchased for $15,000 as of 2007; ⢠Improved documentation; ⢠Removal of restrictive licensing agreements and the move to an âopen- sourceâ environment; and ⢠Easier transition from the old to the newâin the Portland case study, ways of layering TRANSIMS methods over existing methods and data were discovered, thus creating a more tractable deployment path. On the basis of its experience and knowledge, the committee believes that TRANSIMS provides an important bridge from the current practice of static, trip-based modeling to a future practice that better represents personal activ- ity and dynamic traffic flow throughout the day. The groundwork provided by TRANSIMS research and development has materially assisted other model developers in moving toward highly disaggregate tour-based models and in particular has demonstrated the importance of fully representing the tempo- ral dimension for both demand and supply. The committee believes that the federal government should continue TRANSIMS and other initiatives with the aim of developing advanced modeling methods that, once proven effective, can be transferred to practice by the most efficient means. EXPERIENCE WITH ADVANCED PRACTICE Questions remain about the wisdom of investing in advanced modeling prac- tices. For example, is the advanced practice more than the agency really needs? Are the forecasts reasonable? Can the agency maintain the model set? The current state of knowledge is such that there can be no definitive answer to these questions, but the following discussion of field experience with advanced practice models should shed some light. The following three agencies in North America have implemented advanced activity-based travel models and are using them in practice (VHB 2006): ⢠Mid-Ohio Regional Planning Commission (MORPC), Columbus; ⢠New York Metropolitan Transportation Council (New York City); and ⢠San Francisco County Transportation Authority.
Advancing the State of the Practice 101 Eight others are currently in the process of designing and implementing such models (Cervenka 2007): ⢠Atlanta Regional Commission, ⢠Denver Regional Council of Governments (DRCOG), ⢠Metropolitan Transportation Commission (MTC) (San Francisco Bay Area), ⢠North Central Texas Council of Governments (DallasâFort Worth), ⢠Portland Metro (Oregon), ⢠Sacramento Area Council of Governments, ⢠St. Louis East-West Gateway Council of Governments, and ⢠Tahoe Regional Planning Agency (Lake Tahoe, California and Nevada). The growing interest in advanced modeling reflects an understanding that the current trip-based models are not well suited to analyzing the com- plex range of policy alternatives that are of interest to many urban areas (Meyer and Miller 2001). Activity-based models, in contrast, offer full incor- poration of the time-of-day dimension, which permits modeling of differen- tial time-specific tolling and parking policies and flexible working hours, as well as production of improved inputs needed for the Environmental Protection Agencyâs MOBILE model. Activity-based models also allow for detailed representation of segments of the travel market and portrayal of value of time for population segments. Travel response to demographic changes can also be accounted for. Finally, pairing an activity-based model with a traffic microsimulation model permits detailed analysis of improvements in traffic operations (Vovsha et al. 2005). Following are four case studies of the implementation of advanced models. Mid-Ohio Region Travel Demand Model This new set of regional travel forecasting models for MORPC was com- pleted in 2004. It is described as an advanced, multistep tour-based micro- simulation model (Anderson and Donnelly 2005). The model features explicit modeling of intra-household interactions and joint travel that is of crucial importance for realistic modeling of the individual decisions made in the household framework and in particular for choice of the high occu- pancy vehicle (HOV) as travel mode. The original concept of a âfull indi- vidual daily patternâ that constituted a core of the previously proposed activity-based model systems has been extended in the MORPC system to incorporate various intra-household impacts of different household members
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 102 on each other, joint participation in activities and travel, and intra-household allocation mechanisms for maintenance activities. The model also features enhanced temporal resolution of 1 hour with explicit tracking of available time windows for generation and scheduling of tours instead of the 4â5 broad time-of-day periods applied in most of the conventional models and activity-based models previously developed. (PB Consult 2005, 1) As of January 2007, the prior, conventional model was no longer in use; the MORPC activity-based model had been estimated and validated and was in use for the long-range plan, air quality conformity, and transit alternatives. The work on transit alternatives included the North Corridor Transit Project, a likely candidate for the Federal Transit Administrationâs (FTAâs) New Starts program. Because of the high standards set by FTA for travel demand modeling for New Starts, the performance of the MORPC model for this transit study was evaluated with some care (Schmitt 2006). The following are some findings concerning the modelâs performance for this study: ⢠Overall, the modeled trip distribution for work purposes appears to be as good as or better than that of comparable models used elsewhere in the United States. ⢠The model produced reasonable results for user benefits. ⢠The maps from the model were very good at explaining the benefits and disbenefits of the project. MORPC was found not to be taking advantage of the increased functional- ity of the new model because of a need to catch up with a backlog of routine work, but it was reported that with the new model, the range of applications that could be addressed was considerably expanded (Anderson 2007). New York Best Practices Model Planning and data collection for this model were conducted in the 1990s, and the model was implemented in 2002. This is described as an activity- based model employing microsimulation to replicate the travel patterns of each person in the region using all modes of travel, including nonmotorized. The model covers 28 counties and has 3,600 transportation analysis zones. During 2002â2006, the model was used for air quality conformity analy- sis, major investment studies, analysis of the Transportation Improvement
Advancing the State of the Practice 103 Program and regional transportation plan, and the Manhattan area pricing study (Chiao et al. 2006). Those using the model results for particular studies (Tappan Zee Bridge/ I-270 Alternatives Analysis and Kosciuszko Bridge) reported either having no problems or being highly satisfied with the model results, which appeared to be intuitive and to provide an improved level of detail as compared with other models (VHB 2006). San Francisco County Transportation Authority The San Francisco County activity-based model was developed to provide more detailed and accurate information on traveler behavior with respect to destination choices, modal options, and time of day. The model focuses on travel in San Francisco County and combines input from the regional metro- politan commission for a complete portrayal of travel (Outwater and Charlton 2006). The model was used to provide forecasts for the New Central Subway light rail transit project and the alternatives analysis for the Geary Study. For the Central Subway project, the model was used to calculate user benefits for an FTA New Starts application; staff who worked on the application reported satisfaction with the model (VHB 2006). For another application, the San Francisco model was linked with traffic microsimulation software to estimate and portray network impacts of a bus rapid transit project (Charlton 2007). Finally, in the development of the countywide transportation plan, the San Francisco model was applied to an equity analysis to estimate impacts on mobility and accessibility for different populations. Equity analyses performed by traditional models suffer from aggregation biases and limited data. The San Francisco microsimulation model makes it possible to estimate impacts on different communities according to gender, income, automobile avail- ability, and household structure (Outwater and Charlton 2006). DRCOG Activity-Based Model This model is in the planning stages. It is of particular interest as DRCOG, the Denver MPO, conducted an extensive regional visioning process (Metro- vision), after which the model features needed to support regional planning for the elements of Metrovision were determined. DRCOG concluded that, while activity-based modeling could not fully address all issues, it would be clearly superior to four-step modeling in many respects. Among the issues for which
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 104 activity-based modeling was judged to be superior were the following (Sabina and Rossi 2006): ⢠Pricing and tolling analysis, ⢠Policies sensitive to time of day, ⢠Urban centers and transit-oriented development, ⢠Transportation project analysis, and ⢠Induced travel. OBSTACLES TO MODEL IMPROVEMENTS Despite recent demonstrated advances, the pace of change in travel demand forecasting practice through the years has not been fast. MPO staffs want to use travel forecasting tools that are consistent with the state of the practice and are appropriate for the issues the MPO must address. At the same time, they work within the constraints of time and budget, both of which must be directed to meeting current project planning needs as well as conducting any research activities. Following are some salient obstacles to adopting advanced modeling practices. Cost Cost is one potential barrier to the implementation of advanced modeling practices. The cost of implementing an activity-based model depends on a number of variables, including the size of the network, the extent of transit service, and the availability of activity information from a recent home inter- view survey. Another key issue affecting cost is the extent to which there is a continuous representation of time for traffic assignment. Information on implementation cost was sought informally from three agencies (MORPC, DRCOG, and MTC) and an experienced consulting firm. Respondents expressed costs primarily in a range representing both consultant and staff costs. The average of these total costs was $1 million to $1.4 million. Technical Issues In addition to cost, some agencies may have technical reasons for being reluc- tant to adopt advanced models. These include the following (Vovsha et al. 2005):
Advancing the State of the Practice 105 ⢠Activity-based models provide probabilistic forecast results; different model runs with the same inputs produce different outputs. This has impli- cations for meeting regulations that require point estimates of travel. ⢠Data are required from a large-sample home-interview travel survey (typically 4,000 to 5,000 households). ⢠It may be difficult to achieve reasonable computer run times given the complexity of the model. Staffing and Training As noted in Chapter 4, in many agencies, staff members with the skills required to develop and apply advanced practices are limited. Most small and medium MPOs have few staff members assigned to travel forecasting. These employees may have skills in applying the existing model but often lack train- ing or experience in model development. Unless special efforts are made, many of these employees will not have exposure to or interest in new meth- ods. For all MPOs, the transition from the old model to the new may be dif- ficult to achieve given the demands on MPOsâ technical staff for production and continuity of model results. Institutional Issues Another obstacle to model improvement activities by MPOs is aversion to changing the status quo. The committee believes there is institutional reluc- tance to suggest problems with existing models since projects planned using those models may be challenged not only in the public arena but also in law- suits. Implementing a new modeling procedure may be viewed as an implicit admission that there were problems with the models previously used. Where planned projects exist over which some controversy remains, implementing a new procedure may open up the possibility that previous decisions will be challenged and that completed analyses will need to be reassessed in light of new forecasts. Procedures established for analyzing the conformity of an adopted transportation plan with air quality programs are another salient issue in considering the development of new models. Given the work involved in revising forecasts, agencies may be reluctant to change models once the model emissions budget within the state implementation plan has been established.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 106 The committee believes that the interagency structure of planning within a metropolitan region may also be a barrier to change. In many metropolitan areas, local planning agencies, transportation providers, and state agencies may maintain their own travel forecasting models that use outputs from the regional model, or they may borrow the MPOâs models for their own use. In such instances, significant MPO modeling enhancements might be viewed as a hindrance to ongoing work by other agencies, which are likely to be repre- sented on regional transportation technical committees and the MPO policy board. This is not an insurmountable problem, but the need to build a con- sensus among all users of the MPO model and its outputs can be a significant complicating factor in efforts to introduce new modeling approaches. Need for Tangible Evidence The need for evidence has two facets. First, agencies may believe that their cur- rent models are adequate for current uses and have no evidence to the con- trary. MPOs have rarely investigated the extent to which forecasts produced by their models have been valid. Time and funds for retrospective analysis are lacking. Periodic validation of a model set will reveal surface problems such as differences between assigned volumes and counts but will give no indication of where within the model set problems may reside. A true reassessment of the existing model set, from generation through distribution and mode choice to assignment, requires as many data as are required for model development, or more. Lacking such retrospective analysis that demonstrates a failure of current forecasting procedures, agencies are under little pressure to devote resources to the exploration or development of new procedures. Second, proof that the advanced modeling practices are better than current practices is needed. Before undertaking major investments in new models, MPOs want tangible evidence that the new procedures will yield forecasts that are notably better than those produced with currently accepted procedures. Overcoming Obstacles to Model Improvements The committeeâs web-based survey showed that 70 percent of large and medium MPOs identified features of their models needing improvement. In the web- based survey, about 20 percent of small and medium MPOs and almost 40 per- cent of large MPOs reported that they are exploring replacing their existing
Advancing the State of the Practice 107 model with an activity- or tour-based model. Three U.S. cities are known to have implemented such advanced models, and eight others are in the design process. While some MPOs are satisfied with the status quo, it is apparent that there is a growing willingness to adopt or at least explore advanced practices that may better serve MPOs with more complex needs. Some lead agencies clearly have found ways to overcome obstacles to improvement, and it is likely that with increased experience, better home interview techniques, and faster computers, these difficulties may be mitigated. Presumably with greater experience, the ini- tial cost of model development will fall. A strong case can be made for the pool- ing of resources among MPOs for joint development, and for continued or increased federal support and leadership in advanced model development. MODEL RESEARCH, DEVELOPMENT, AND IMPLEMENTATION Activities aimed at advancing the state of practice through research and devel- opment take place at each level of government and through nongovernmental efforts as well. There is great potential for expansion and better coordination of this work. Federal Initiatives As noted in Chapter 3, the federal government has a strong interest in robust metropolitan travel forecasting to ensure that federal funds are being used to support top-priority needs for maintenance and improvement of the national transportation system and to meet the requirements of federal laws, in partic- ular the Clean Air Act, the Clean Water Act, and the National Environmental Policy Act. As also noted in Chapter 3, FTA has taken a strong role in improv- ing modeling practice. TMIP has been sponsored by FHWA since 1992. Its mission is to âsup- port and empower planning agencies, through leadership, innovation and support of travel analysis improvements, to provide better information to support transportation and planning decisionsâ (tmip.fhwa.dot.gov/about/ mission.stm). The program has three goals: ⢠Help planning agencies build their institutional capacity to develop and deliver travel modelârelated information to support transportation and planning decisions;
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 108 ⢠Develop and improve analytical methods (including TRANSIMS) that respond to the needs of planning and environmental decision making; and ⢠Support mechanisms designed to ensure the quality of technical analy- sis used to support decision making and to meet local, state, and federal pro- gram requirements (tmip.fhwa.dot.gov/about/goals_activities.stm). A 2003 performance assessment found that TMIP had had a positive influence on short-term model improvements, leaving transportation agen- cies in a better position to address federal and state planning requirements. Specific activities have included the following: ⢠Enhancements to current models, ⢠Topical conferences and workshops, ⢠A newsletter (1,300 subscribers), ⢠A website with a library of literature on modeling topics (visited on average 1,500 times a day), ⢠An e-mail list that reaches a national and international audience (almost 900 members and 50 postings per month), and ⢠A travel model peer review program for which more than 20 agency reviews had been completed as of 2006. Long-term model development has been accomplished through TRANSIMS, discussed above. The evaluation report notes: âWhile much has been accomplished, continuing outreach and additional research are needed to help advance the state-of-the-art with travel forecasting modelsâ (Shunk and Turnbull 2003, 27). In 2007, FHWA is providing TMIP staff support and, through the agencyâs research program, the primary funding for TMIP activities. TRANSIMS is funded separately by specific allocations [as it was previously under the Transportation Equity Act for the 21st Century (TEA-21)]. In the latter days of TEA-21, TMIP was funded at approximately $500,000 annu- ally for all activities other than TRANSIMS. The same approximate level of funding has continued under SAFETEA-LU. Given the stated purposes of the program and the apparent need for such a national program to advance the state of practice in travel modeling, the committee finds this level of funding to be inadequate. In the late 1970s and early 1980s, FHWA and the Urban Mass Transportation Administration (later FTA) spent about $5 million a year on travel model development and implementation, equivalent to about $15 mil-
Advancing the State of the Practice 109 lion in current dollars. A strong federal role is needed to provide models and data development, assistance with implementation, training, and documen- tation. The resources currently being provided are insufficient to allow the federal government to assume this role in a meaningful way. The current authorized FHWA and FTA capital program totals about $40 billion. It would appear appropriate to make an annual investment of 0.05 percent, or $20 million, of this amount for the development and implementation of improved travel forecasting models. State Initiatives The states have their own national research program, the National Cooperative Highway Research Program (NCHRP), sponsored by individual state trans- portation agencies and the American Association of State Highway and Transportation Officials (AASHTO) in cooperation with FHWA. NCHRP was created in 1962 as a means to conduct research of interest to the states in acute problem areas that affect highway planning, design, construction, operation, and maintenance nationwide. Funding for the program is con- tributed by each state, drawing from federal State Planning and Research funds. Research topics are chosen annually by the AASHTO Standing Committee on Research. NCHRP conducts research on topics related directly to metropolitan travel forecasting. Examples are the completed NCHRP Report 388: A Guidebook for Forecasting Freight Transportation Demand and two efforts currently under way: NCHRP Projects 8-37, Standardized Procedures for Personal Travel Surveys, and 8-61, Travel Demand Forecasting, Parameters and Techniques.2 In the past, NCHRP funding has been programmed to sup- port specific TMIP activities. Other Research and Development Initiatives Other sources of funding and research to advance the state of practice in travel modeling include the national Transit Cooperative Research Program (TCRP), established with FTA sponsorship and funding in July 1992. The pro- gram has an independent governing board representing the transit industryâ 2 Updates NCHRP Report 187: Quick-Response Urban Travel Estimation Techniques and Transferable Parameters: Userâs Guide.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 110 the TCRP Oversight and Project Selection Committee, which also selects research topics. TCRP has performed research that has contributed to advancing the state of practice in travel forecasting. Examples are the com- pleted TCRP Report 48: Integrated Urban Models for Simulation of Transit and Land Use Policies: Guidelines for Implementation and Use and the in- progress TCRP Project H-37, Improving Travel Forecast Models for New StartsâMode Specific Constants. University researchers can also make substantial contributions to research and practice, working with MPOs and states. One example is the joint ini- tiative of the University of Texas at Austin and the DallasâFort Worth MPO to demonstrate the Comprehensive Econometric Microsimulator for Daily Activity-Travel Patterns, an econometric activity-based modeling system. This work is being funded by the Texas Department of Transportation (Bhat et al. 2006). Another example is the University of California at Davisâ Caltrans Air Quality Project, which since 1997 has been developing and implementing transportation-related air quality analysis tools and procedures that help regional, state, and federal agencies achieve improved air quality (aqp.engr.ucdavis.edu/). In Florida, there is a statewide Florida Model Task Force that commissions research projects from the stateâs universities to ben- efit all Florida MPOs (Florida Model Task Force 2007). Consultants play a key role in technology transfer and application devel- opment. Notably, the three implementations of activity-based metropolitan models (San Francisco, New York, and Columbus, Ohio) have depended heavily on consultant leadership, and TRANSIMS also relies on consultant assistance for its current development and implementation activities. Metropolitan Opportunities The principal consumers of research and development in models for metro- politan travel forecasting are the MPOs (and states that perform model devel- opment and forecasting on behalf of MPOs). These operating entities are responsible for providing validated regional models for use in analyzing and forecasting changes in travel for alternative transportation investments and policies. As noted in Chapter 2, they are also faced with meeting expanded requirements for their planning programs. Evaluation of which potential model enhancements can usefully be imple- mented is ultimately the MPOsâ responsibility, funding to support improved
Advancing the State of the Practice 111 or new models must be sought by individual MPOs, and implementation of new modeling practices must take place at the metropolitan level. Despite these considerable responsibilities, the MPOs currently have no national, col- lective means of identifying and directing the most appropriate research and development that would serve their needs or of funding such activities. Each MPO must find its own funding, data, consultant assistance, and trained staff for model development. To the extent that metropolitan areas have their own unique conditions, this may be appropriate. But there is also a strong case to be made for the economies of a pool-funded approach to modeling research and development that could benefit many or all MPOs. Figure 6-1 shows federal funding from FHWA and FTA available from 1992 to 2006 to support the planning activities of all 384 MPOs. Funding levels are shown in both current and constant dollars, indexed to 1992. Since 1992, funding in current dollars has grown from $161 million to $366 mil- lion, an increase of 127 percent. If inflation is taken into account, the increase is to $287 million, or 78 percent. Concurrent with this increase in MPO funding was an increase in the scope of MPO responsibilities, due mainly to expanded federal requirements. There was also steady growth in the number of MPOs as more urban areas reached 400 350 300 250 $ Millions 200 150 100 50 Current $ Constant $ 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Year FIGURE 6-1 Federal MPO funding, 1992â2006. (Sources: FHWA 2006; FTA 5303 Apportionment Table, personal communication from Ken Johnson to J. Williams, April 24, 2006.)
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 112 the 50,000 population threshold. Nonetheless, $366 million is a substantial figure, and a modest takedown from this figure could support a national MPO research program, controlled by MPOs and dedicated to their research needs. The New York State Association of Metropolitan Planning Organiza- tions (NYSAMPO) has shown how such an initiative can work on a state- wide basis. There are also national models for how entities with common research interests can benefit from pool-funded research: NCHRP for state transportation agencies, TCRP for transit agencies, the Airport Cooperative Research Program (ACRP) for airports, the National Cooperative Freight Research Program (NCFRP), and the Hazardous Materials Cooperative Research Program (HMCRP). A Metropolitan Planning Cooperative Research Program would give MPOs the lead in developing their own national research agenda and carrying out that agenda through a research program. The scale of such a program can be roughly estimated from that of the cited national programs, for which approximate annual funding is as follows: ⢠NCHRP, $33 million; ⢠TCRP, $9 million; and ⢠ACRP, $10 million; ⢠NCFRP, $3 million; and ⢠HMCRP, $1 million. The administrative costs of these programs are roughly 25 percent. Assuming a Metropolitan Planning Cooperative Research Program wished to start 12 research and development projects a year and that these projects averaged $300,000 each, the cost of the program would be (12) ($300,000) (1.25) = $4.5 million. This $4.5 million would represent 1.2 percent of total federal (FHWA and FTA) funding for MPOs ($366 million in 2006). Following the exam- ple of NYSAMPO, the smaller MPOs (those with populations of under 200,000) might be exempted from financially supporting the program, in which case the takedown would be greater for the larger MPOs. This fund could be created through the state transportation agencies that receive MPO funds or through the federal government. Another approach would be for MPOs with common needs to join together for research and development studies of mutual interest. State transportation agencies often join together for such pool-funded research on topics of com- mon interest. FHWA has a program to facilitate this type of pooled research, and MPOs are mentioned as possible participants (www.pooledfund.org/).
Advancing the State of the Practice 113 The project-by-project approach does not lend itself to creating an ongoing research program but may answer the needs of a group of MPOs with a com- mon problem to address. Regardless of the specific operating mechanism, pooling of research and development funds offers an efficient means of meeting MPO needs for model enhancement, development, and implementation. Another advantage is the possibility of leveraging funds through joint ventures with federal, state, transit, and other research programs. MPOs could be in charge of substantial ongoing funds, which could be used to satisfy their own model research and development needs or for other research and development purposes, according to their wishes. The following are examples of what such pooled research might accom- plish (Cervenka 2005): ⢠Rigorous examination of implemented (or estimated) advanced models, with sensitivity and validation tests; ⢠Exploration of data and parameters transferable from region to region; ⢠Development of universally estimated, locally calibrated models; ⢠Pooled acquisition of computer software and hardware; and ⢠Documentation of practice for shared-use applications. An Integrated Approach to Research, Development, and Implementation of Advanced Models Currently, elements of research, development, and implementation for travel models are diffused among local, state, and federal governments and other entities. Each of these entities has a definite and discernible role. The federal government takes the lead and bears the risk for high-payoff research that will benefit the nation and facilitates diffusion of advanced practices. Through their research programs, states sponsor advances that meet state and MPO needs and facilitate training and technology transfer through statewide model user groups. The transit industry has its research program in support of tran- sit agencies, as well as means for technology transfer through such groups as the American Public Transportation Association. MPOs bear the responsi- bility for transferring travel model research into practice, a role that might be facilitated through a national program of application-oriented research funded and directed by MPOs. These various elements of research, development, and implementation could be better integrated for the mutual benefit of all parties and achievement
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 114 of the ultimate goal of improved travel forecasting models. This integra- tion could be effected through a national travel forecasting steering com- mittee. This committee could meet regularly to set goals and an agenda for joint activities to improve travel models and modeling practice, avoid- ing duplication of effort and ensuring that resources would be directed toward top priorities. An activity associated with the national steering committee could be the development and production of a national travel forecasting handbook. Currently, no single source of information describes current or evolving prac- tices in travel modeling and forecasting. Such a handbook could identify alter- native best practices for addressing various travel markets and metropolitan needs. It would be an informational and evolving document, with no pre- scriptive or regulatory implications, and would reflect recognition that differ- ent approaches are needed according to the metropolitan context. Creation of the handbook might be directed by the travel forecasting steering committee and accomplished through a national organization that would bring together practitioners and researchers from government agencies, consulting firms, and academia. The primary stakeholders would be those responsible for conduct- ing metropolitan travel forecasting. Resources to support the handbook might be derived from NCHRP, TCRP, the recommended Metropolitan Cooperative Research Program, and the federal government. SUMMARY FINDINGS AND RECOMMENDATIONS This chapter has addressed improvements in four-step trip-based modeling; advanced modeling practices; TRANSIMS; experience with advanced prac- tice; obstacles to model improvement; and model research, development, and implementation. Improvements in Four-Step Trip-Based Modeling MPOs may undertake ambitious modeling improvement programs within the framework of their current models. Typical results are improved measures of arterial congestion, accounting for highway and transit in trip distribution, improved trip distribution models, improved modeling of nonmotorized travel, and improved sensitivity testing.
Advancing the State of the Practice 115 Advanced Modeling Practices Travel models can be improved by being based on a more comprehensive understanding of the activities of households. Also needed is a more complete representation of the supply-side network to account for the details of con- gested operations throughout the day. No one new modeling approach can address these and other needs. Rather, a suite of related approaches, taken together, shows promise for greatly improving modeling practice. These approaches include improved land use modeling, tour-based models, activity- based models, discrete-choice modeling, traffic microsimulation, and dynamic traffic assignment. There remain questions about the wisdom of the investment in advanced modeling practices. For example, are they more than the agency really needs? Are the forecasts reasonable? Can the agency maintain the model set? The current state of knowledge is such that there can be no definitive answer to these questions. For this reason, the committee believes that MPOs experimenting with or fully implementing advanced modeling practices should document their experiences, including costs, advantages, draw- backs, and any transferable data or model components. In addition, the committee recommends that studies be performed to compare the per- formance of conventional and advanced models and to evaluate how well-implemented advanced models handle complex planning issues that are beyond the scope of current models. TRANSIMS TRANSIMS is a computer-based system capable of simulating the second-by- second movements of every person and every vehicle through the transporta- tion network of a large metropolitan area. It incorporates and integrates some of the advanced modeling practices detailed in this chapter, in particular pop- ulation synthesis, activity-based modeling, and traffic microsimulation. From 1992 to 2003, $38 million was spent on TRANSIMS, about three- quarters of which went to Los Alamos National Laboratory for basic research and development. TRANSIMS was originally field-tested in DallasâFort Worth, Texas, and Portland, Oregon. SAFETEA-LU will support two to three new deployments a year. TRANSIMS technology is also being used for projects not funded through SAFETEA-LU.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 116 TRANSIMS has not yet been implemented by any MPOs for use in core travel forecasting activities. Some reasons for this include the softwareâs hav- ing been developed in a research and development setting, its not being well adapted for general deployment, early requirements for high-performance computers and the Linux operating system, a poor user interface, and docu- mentation that did not easily support applications. There has been a percep- tion that implementing TRANSIMS (and other activity-based models) may be an overwhelming task. Yet a number of improvements have made TRANSIMS more accessible and ready for implementation. On the basis of its knowledge and experience, the committee believes TRANSIMS has provided an important bridge from the current practice of static, trip- based modeling to an improved future practice. The federal government should continue funding TRANSIMS development and implementation at appropriate levels. Experience with Advanced Practice Three agencies in the United States have implemented advanced, activity- based travel models and used them successfully for typical transportation planning applications. Users report satisfaction with the model results, and where analysis has been done, the results are described as reasonable and comparable with those from the prior, trip-based models. At least eight addi- tional U.S. cities are actively planning for the introduction of advanced models. Obstacles to Model Improvement Obstacles to the adoption of advanced modeling practices include the following: ⢠Cost of implementation, ⢠Limited staff skills, ⢠Reluctance to suggest problems with existing models since doing so could cause projects planned on the basis of those models to be challenged, ⢠Reluctance to change models once the model emissions budget within the state implementation plan has been established, ⢠No analysis demonstrating a weakness of current forecasting procedures,
Advancing the State of the Practice 117 ⢠The need for evidence that new procedures will perform better than the current ones, and ⢠The belief of some MPOs that their current models are doing an ade- quate job. Model Research, Development, and Implementation Activities to advance the state of practice through research and develop- ment take place at each level of government and through nongovernmen- tal efforts as well. There is great potential for expansion and better coordination of this work. Federal Initiatives TMIP has been sponsored by FHWA since 1992. A 2003 performance assessment found that TMIP had a strong positive influence on short-term model improvements. Successes have included enhancements to current models, topical conferences and workshops, a newsletter, a website, an e-mail list that reaches a national and international audience, and a travel model peer review program. Long-term model development has been accomplished through TRANSIMS, discussed above. The committee finds the current annual funding for TMIP ($500,000) to be inadequate. The committee calls on the U.S. Department of Transportation, FHWA, and FTA to take the following steps to facilitate the needed improve- ments in both models and practice: ⢠Support and provide funding for incremental improvements to existing four-step (or three-step) trip-based models, in settings appro- priate for their use. ⢠Support and provide funding for the continued development, demonstration, and implementation of advanced modeling approaches, including activity-based models. ⢠Continue to rely on TMIP as an appropriate mechanism for ad- vancing model improvement. ⢠Increase funding to an appropriate level to support the federal gov- ernmentâs role as a partner with MPOs and state transportation agencies in the development and implementation of improved modelsâan annual investment of approximately $20 million.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 118 State Initiatives The states have their own national research program, NCHRP, sponsored by individual state transportation agencies. Funding for this program is con- tributed by each state. NCHRP conducts research on topics directly related to metropolitan travel forecasting. Other Research and Development Initiatives Other sources of funding and research support efforts to advance the state of practice in travel modeling. One example is TCRP. Consultants also play a key role in technology transfer and applications development. University researchers can make substantial contributions to research and practice, working with MPOs and states. Individual MPOs and universities could form partnerships to foster travel model research and implementation of advanced modeling practice. Metropolitan Opportunities The principal consumers of research and development in metropolitan travel forecasting models are the MPOs (and states that perform model development and forecasting on behalf of MPOs). Despite their consid- erable responsibilities, the MPOs currently have no national research pro- gram of their own. The committee believes the MPOs would benefit from establishing a national metropolitan cooperative research pro- gram. Because model applications must fit local needs and context, it is important for MPOs to take a leadership role in model selection, devel- opment, application, testing, and verification. Large costs are involved in both improving current models and developing more advanced models. Rather than duplicating these costs at each MPO, it would be beneficial to pool resources for such activities as model enhancement, new model development, implementation procedures, and staff training programs. MPOs nationally receive annual funding of $366 million. A takedown of 1.2 percent from this total would produce a program with a $4.5 million annual budget, which should be sufficient to start 10 to 12 research proj- ects a year. An Integrated Approach to Research, Development, and Implementation Currently, elements of research, development, and implementation in travel modeling are diffused among local, state, and federal governments
Advancing the State of the Practice 119 and other entities. These levels of government should work cooperatively to establish appropriate goals, responsibilities, and means of improving travel forecasting practice. This cooperation could be accomplished through a national travel forecasting steering committee. This commit- tee could set goals and an agenda for joint activities aimed at improving travel models and modeling practice. An activity associated with the national steering committee should be the development and production of a national travel forecasting handbook. This would be an informational and evolving document, with no prescriptive or regulatory implications. REFERENCES Abbreviation FHWA Federal Highway Administration Anderson, R., and B. Donnelly. 2005. Comparison of the Prior and New MORPC Travel Forecasting Models. Mid-Ohio Regional Planning Commission, Columbus. Anderson, R. 2007. Presentation at Workshop 164, 86th Annual Meeting of the Transportation Research Board, Washington, D.C., 2007. Bhat, C., and F. Koppelman. 2003. Activity-Based Modeling of Travel Demand. In Handbook of Transportation Science (R. W. Hall, ed.), Kluwer Academic Publishers, Boston, Mass., Chapter 3. Bhat, C., J. Guo, S. Srinivasan, A. Pinjari, N. Eluru, I. Sener, R. Copperman, and P. Ghosh. 2006. Comprehensive Econometric Microsimulator for Daily Activity- Travel Patterns: Recent Developments and Sensitivity Testing Results. Presented at Conference on Innovations in Travel Demand Modeling, Austin, Tex., May. www. trb.org/Conferences/TDM/papers/BS2A%20-%20CEMDAP.pdf. Boyce, D. 2002. Is the Sequential Travel Forecasting Paradigm Counterproductive? Journal of Urban Planning and Development, Dec. Cervenka, K. 2005. Adopting Innovative Methods for Planning. Presentation at Workshop 111, 84th Annual Meeting of the Transportation Research Board, Washington, D.C. Cervenka, K. 2007. An Update on Advanced Model Development. Presented at 11th National Planning Applications Conference, Daytona Beach, Fla., May 8. Charlton, B. 2007. The San Francisco Model . . . in Fifteen Minutes. Presentation at Workshop 116, 86th Annual Meeting of the Transportation Research Board, Washington, D.C. Chiao, K., B. Bhowmick, and A. Mohseni. 2006. Lessons Learned from the Implementation of NY Activity-Based Travel Models. Presented at Conference on Innovations in Travel Modeling, Austin, Tex., May. www.trb.org/Conferences/TDM/papers/BS4A%20-%20 Final%20White%20Paper%20NYBPM%20experience%204-28.pdf.
METROPOLITAN TRAVEL FORECASTING Current Practice and Future Direction 120 Deakin, E., and G. Harvey. 1994. A Manual of Regional Transportation Modeling Practice for Air Quality Analysis, Chapter 3. National Association of Regional Councils, Washington, D.C. tmip.fhwa.dot.gov/clearinghouse/docs/airquality/mrtm/ch3.stm. DKS Associates. 2001. Final Land Use and Transport Modeling Design Report and Addenda to SACOG Model Design Report. Sacramento Area Council of Governments, Calif. Domencich, T. A., and D. McFadden. 1975. Urban Travel Demand: A Behavioral Analysis. North-Holland, New York. Dowling, R., A. Skabardonis, and V. Alexiadis. 2004. Traffic Analysis Toolbox. Volume III: Guidelines for Applying Traffic Microsimulation Modeling Software. FHWA, Washington, D.C. Fehr & Peers. 2005. Las Vegas Travel Demand Model Guidelines for Estimation, Calibration and Validation. Regional Transportation Commission of Southern Nevada. FHWA. 2006. Apportionment of Metropolitan Planning Funds. 4510.273, 288, 307, 328, 379, 394, 410, 423, 445, 474, 503, 541, 560, and 599. www.fhwa.dot.gov/legsregs/ directives/notices.htm. Florida Model Task Force. 2007. Research Projects. www.fsutmsonline.net/index.php?/ site/directory/modeling_research. Hobeika, A. 2005. TRANSIM Fundamentals. Virginia Polytechnic University. tmip.fhwa. dot.gov/transims/transims_fundamentals/. Los Alamos National Laboratory. 2007. Why TRANSIMS? Los Alamos, N.Mex. www.ccs.lanl.gov/transims/index.shtml. McNally, M. G. 1997. The Potential of Integrating GIS in Activity-Based Forecasting Models. Center for Activity Systems Analysis, University of California, Irvine. Meyer, M. D., and E. J. Miller. 2001. Urban Transportation Planning: A Decision-Oriented Approach. McGraw-Hill, Boston, Mass. Miller, E. J., D. S. Kriger, and J. D. Hunt. 1999. TCRP Report 48: Integrated Urban Models for Simulation of Transit and Land Use Polices: Guidelines for Implementation and Use. Transportation Research Board, National Research Council, Washington, D.C. Outwater, M., and B. Charlton. 2006. The San Francisco Model in Practice: Validation, Testing, and Application. Presented at Conference on Innovations in Travel Modeling, Austin, Tex., May. PB Consult. 2005. Mid-Ohio Regional Planning Commission Transportation Modeling System Overview and Summary. www.drcog.org/documents/MORPC.pdf. Sabina, E., and T. Rossi. 2006. Using Activity-Based Models for Policy Decision Making. Presented at Conference on Innovations in Travel Demand Modeling, Austin, Tex., May. Schmitt, D. 2006. Application of the MORPC Microsimulation Model: New Starts Review. Presented at Conference on Innovations in Travel Demand Modeling, Austin, Tex., May. www.trb.org/Conferences/TDM/papers/BS1A%20-%20Activity-Based%20Model%20 Application.pdf. Shunk, G., and K. Turnbull. 2003. Product Delivery of New and Improved Travel Forecasting Procedures. Draft final report, NCHRP Project 8-36, Task 6. Texas Transportation Institute, College Station. www.transportation.org/sites/planning/docs/nchrp6.doc.
Advancing the State of the Practice 121 VHB. 2006. Results of FY2006 Travel Forecasting Research, Task 5: Review of Current Use of Activity-Based Modeling. Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, Washington, D.C. Vovsha, P., M. Bradley, and J. Bowman. 2005. Activity-Based Forecasting Models in the United States: Progress Since 1995 and Prospects for the Future. In Progress in Activity- Based Analysis (H. Timmermans, ed.), Elsevier Press, Amsterdam, Netherlands.
