Travel Demand Forecasting: Parameters and Techniques (2012)

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.

A-1 A.1 Environmental Protection Agency . . . . . . . . . . . . . . A-1 A.2 Federal Highway Administration . . . . . . . . . . . . . . . A-3 A.3 Federal Transit Administration. . . . . . . . . . . . . . . . . A-4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-7 This appendix discusses federal agency requirements for transportation planning and travel models in urban areas. The requirements for three agencies are presented—the Environ- mental Protection Agency, the Federal Highway Administra- tion, and the Federal Transit Administration—and are up to date as of the time of the writing but are subject to change based on updated legislative and rulemaking actions. A.1 Environmental Protection Agency The most specific federal agency requirements for travel demand forecasting are found in the Transportation Con- formity Rule, promulgated by the Environmental Protection Agency (EPA) under the Clean Air Act (CAA) [42 U.S. Code (USC) 85 § 7401 et seq.]. A.1.1 Background The EPA is the federal agency charged with implementing the requirements of the CAA, a comprehensive federal law that regulates air pollutant emissions from areawide, stationary, and mobile sources. Under the CAA, EPA established National Ambient Air Quality Standards (NAAQS), which set limits on concentrations of specific air pollutants throughout the United States. Each state is responsible for monitoring the concentrations of air pollutants within its borders and reducing emissions of those pollutants that exceed the NAAQS. Areas within each state that currently exceed the NAAQS for specific pollutants are designated as nonattainment areas. Each nonattainment area is classified according to the amount by which it exceeds the NAAQS for each type of pollutant. The CAA establishes timetables (depending on the nonattainment classification) by which the area must reduce its pollutant concentrations in order to meet the NAAQS. When a non- attainment area reduces its pollutant concentrations below the NAAQS, it is redesignated as a maintenance area. Main- tenance areas must continue to monitor their air pollutants and maintain NAAQS for a period of 20 years after their redesignation. Each state must develop a state implementation plan (SIP) that explains how it will reduce air pollutant emissions in each nonattainment area to meet and maintain the NAAQS. Every SIP includes an emissions budget, which sets limits on the amount of pollutants each nonattainment area in the state can emit. Transportation conformity is required under the CAA to ensure that federally funded and approved highway and transit activities in nonattainment and maintenance areas are consistent with (i.e., “conform to”) the SIP. According to the CAA, a conforming transportation activity must not: • Create any new air quality violations; • Increase the frequency or severity of existing violations; or • Delay timely attainment of NAAQS. The Transportation Conformity Rule [40 Code of Federal Regulations (CFR) Parts 51 and 93], which establishes crite- ria and procedures for determining whether transportation activities conform to the SIP, was first promulgated under the authority of the 1990 CAA amendments in November 1993. Current conformity regulations reflect a comprehensive revi- sion of the 1993 rule and were published on August 15, 1997 (Federal Register, 62, p. 43780). The Federal Highway Administration (FHWA), Federal Transit Administration (FTA), and metropolitan planning organizations (MPOs) are responsible for making conformity determinations, based on criteria and procedures described in A p p e n d i x A Federal Planning and Modeling Requirements

A-2 the conformity rule. Transportation activities that require a conformity determination include long-range transportation plans (LRTP), transportation improvement programs (TIP), and federally funded or approved transportation projects. To demonstrate conformity, forecasts of regional emissions resulting from a LRTP or TIP must not exceed the motor vehicle emissions budgets for each specified pollutant, as defined in the SIP. Regional motor vehicle emissions must be estimated using EPA-approved emission factor models (e.g., MOBILE, MOVES, or EMFAC), per 40 CFR 93.111. These emission factor models, in turn, require estimates of vehicle speeds and travel volumes [in vehicle miles traveled], which are derived from the travel models used by transportation planning agencies to forecast travel demand under alternative transportation scenarios. The 1997 conformity rule amendments, among other changes, mandated the use of network-based travel models to support conformity determinations in certain nonattainment areas, and included other requirements relating to model structure, input assumptions, included variables, and vali- dation procedures. These requirements are described in the next section. A.1.2 Travel Model Requirements in the CAA Transportation Conformity Rule The specific requirements for travel models are described in Section 122 of the Transportation Conformity Rule [40 CFR 93.122 (b)]. However, these requirements apply only to serious, severe, and extreme ozone nonattainment areas or serious carbon monoxide nonattainment areas whose metro- politan planning area contains an urbanized area population over 200,000 (based on the most recent decennial census con- ducted by the U.S. Census Bureau). In those areas meeting the above criteria, estimates of regional transportation-related emissions used to support conformity determinations must be made at a minimum using network-based travel models according to procedures and methods that are available and in practice and supported by current and available documentation. Agencies must dis- cuss these modeling procedures and practices through the interagency consultation process, as described elsewhere in the Transportation Conformity Rule [40 CFR 93.105 (c) (1) (i)]. Network-based travel models must, at a minimum, satisfy the following requirements: • Network-based travel models must be validated against observed counts (peak and off-peak, if possible) for a base year that is not more than 10 years prior to the date of the conformity determination. Model forecasts must be ana- lyzed for reasonableness and compared to historical trends and other factors, and the results must be documented. • Land use, population, employment, and other network- based travel model assumptions must be documented and based on the best available information. • Scenarios of land development and use must be consis- tent with the future transportation system alternatives for which emissions are being estimated. The distribution of employment and residences for different transportation options must be reasonable. • A capacity-sensitive assignment methodology must be used, and emissions estimates must be based on a methodology which differentiates between peak and off-peak link volumes and speeds and uses speeds based on final assigned volumes. • Zone-to-zone travel impedances used to distribute trips between origin and destination pairs must be in reasonable agreement with the travel times that are estimated from final assigned traffic volumes. Where use of transit currently is anticipated to be a significant factor in satisfying transporta- tion demand, these times also should be used for modeling mode splits. • Network-based travel models must be reasonably sensitive to changes in the time(s), cost(s), and other factors affecting travel choices. Additionally, reasonable methods in accordance with good practice must be used to estimate traffic speeds and delays in a manner that is sensitive to the estimated volume of travel on each roadway segment represented in the network-based travel model. Highway Performance Monitoring System (HPMS) esti- mates of vehicle miles traveled (VMT) shall be considered the primary measure of VMT within the nonattainment or maintenance area and for the functional classes of roadways included in HPMS, for urban areas that are sampled on a separate urban area basis. For areas with network-based travel models, a factor (or factors) may be developed to reconcile and calibrate the network-based travel model estimates of VMT in the base year of its validation to the HPMS estimates for the same period. These factors may then be applied to model estimates of future VMT. In this factoring process, consideration will be given to differences between HPMS and network-based travel models, such as differences in the facility coverage of the HPMS and the modeled network description. Locally developed count-based programs and other departures from these procedures are permitted subject to the interagency consultation procedures described elsewhere in the rule. In all areas not otherwise subject to network-based modeling requirements, regional emissions analyses must continue to use such models and procedures if the use of those procedures has been the previous practice of the MPO. Otherwise, areas may estimate regional emissions using any appropriate methods that account for VMT growth by, for example, extrapolating

A-3 historical VMT or projecting future VMT by considering growth in population and historical growth trends for VMT per person. These methods also must consider future eco- nomic activity, transit alternatives, and transportation system policies. A.2 Federal Highway Administration The FHWA has very few explicit regulations related to the use of travel demand forecasting. The joint FHWA/FTA Statewide and Metropolitan Transportation Planning Regu- lations (23 CFR Parts 450 and 500) include only one specific reference to travel demand forecasts. That single reference, cited below, is included in the section of the metropolitan planning regulations dealing with the development and con- tent of the metropolitan transportation plan: (f) The metropolitan transportation plan shall, at a minimum, include: (1) The projected transportation demand of persons and goods in the metropolitan planning area over the period of the transportation plan [23 CFR 450.322 (f)(1)] Every designated MPO is required, as part of the metro- politan transportation planning process, to prepare a metro- politan LRTP that considers at least a 20-year planning horizon: The MPO shall review and update the transportation plan at least every 4 years in air quality nonattainment and maintenance areas and at least every 5 years in attainment areas to confirm the transportation plan’s validity and consistency with current and forecasted transportation and land use conditions and trends, and to extend the forecast period to at least a 20-year planning horizon [23 CFR 450.322 (b)]. The joint planning regulations provide no other specific requirements or guidance as to how future transportation demand shall be forecast, leaving the determining of such forecasts up to the discretion of each MPO. A transportation management area (TMA) is defined as an urbanized area with a population over 200,000, as defined by the Census Bureau and designated by the Secretary of Trans- portation, or any additional area where TMA designation is requested by the Governor and the MPO and designated by the Secretary of Transportation. An MPO with less than 200,000 may be designated a TMA if it contains any part of an adjacent TMA. Those MPOs that do not represent a designated TMA and not in an air quality nonattainment or maintenance area may request approval from FHWA and FTA to develop a simplified transportation plan, subject to the complexity of the transportation problems in the metropolitan planning area. No further elaboration is included in the regulations on what elements of the transportation plan may be simplified, but this element of the regulations has generally been interpreted to allow smaller MPOs with no significant plans for major transportation improvements (i.e., no capital investments in new highway or transit capacity) to continue to receive federal funding for system maintenance, etc. MPOs that are in air quality nonattainment or maintenance areas for ozone or carbon monoxide must make a conformity determination for any updated or amended transportation plan in accordance with EPA’s transportation conformity regulations [23 CFR 450.322 (l)]. EPA’s Transportation Con- formity Regulations [40 CFR 93.122 (b) and (c)], described elsewhere in this section, do include specific requirements for travel forecasting models. Although the FHWA has few specific regulatory require- ments pertaining to travel forecasting models, the agency has a long history of supporting research and providing technical assistance to state departments of transportation (DOTs) and MPOs in travel demand estimation and forecasting. Currently, most research and technical assistance on travel demand fore- casting funded by FHWA is coordinated through the Travel Model Improvement Program (TMIP), administered out of the Office of Planning. A recently established companion program focusing on freight models is administered out of the FHWA’s Office of Freight Management and Operations. FHWA and FTA oversight of the metropolitan transporta- tion planning process is handled through a formal certification review, conducted jointly by FHWA and FTA field planners in each TMA at least every 4 years. MPOs representing urbanized areas that are not designated as TMAs are allowed to self-certify that they are meeting all federal transportation planning requirements. Historically, the TMA certification process focused on process requirements (e.g., existence of a metropolitan trans- portation plan and public participation plan; composition of the MPO policy board(s); coordination agreements with key stakeholders) and rarely addressed technical issues such as the travel models used in forecasting future passenger and freight demand. In an effort to encourage its field planners to increase awareness of the importance of travel models at MPOs, the FHWA developed a “certification checklist for travel fore- casting methods” (Federal Highway Administration, 2009), to be used in certification reviews. The checklist does not include questions on the specific modeling components used at the MPO but rather focuses on three, generally nontechnical, categories of questions: (1) issues or proposed projects for which forecasts will be used as indicators of model scrutiny by external organizations; (2) key indicators of the MPO’s technical capabilities; and (3) availability of documentation on current conditions, planning/modeling assumptions, and forecasting methods. The certification checklist is intended to act as a rough first filter to help identify those MPOs that

A-4 may require additional technical assistance in forecasting, or whose forecasting approach may not be suitable for intended applications. A.3 Federal Transit Administration The FTA conducts periodic workshops on travel forecasting for transit New Starts applications. The goal of these workshops is to share with project sponsors and their model consultants how FTA evaluates travel forecasts. Furthermore, the work- shops serve as a forum for FTA to establish acceptable model- ing procedures, inputs, and outputs essential for producing reliable forecasts that are sensitive to socioeconomic and level-of-service changes. The material presented in this section is a synthesis of the information that the FTA provided during the September 2007 travel forecasting workshop in St. Louis, Missouri (Federal Transit Administration, 2007). A.3.1 FTA Requirements The FTA provides guidance on the following key aspects of travel forecasting for New Starts: • Properties of travel models; • Rider surveys; and • Calibration and validation. The subsections that follow discuss the FTA’s requirements for each of these items. Properties of Travel Models The FTA’s requirements for the properties of travel models are fairly broad. The FTA supports a localized approach to travel modeling and forecasting. The rationale for such a requirement is that there are no standard or “correct” methods that are universally applicable to all regions. Models will need to reflect the fact that each metropolitan area has unique condi- tions and must be responsive to local decision making. Because models are used to forecast transit ridership, it is essential that they explain the current transit conditions and capture the tradeoffs between travel times and costs. These favorable properties are heavily dependent on the model cali- bration and validation procedures (discussed in the subsec- tion after next). In addition to capturing current conditions, the models will need to fulfill their ultimate objective of yielding reasonable forecasts. Specifically, FTA requires rea- sonable “deltas” (changes in ridership between a base year and forecast year) for ridership that are consistent with the underlying socioeconomic growth as well as level-of-service improvements. Unreasonably high or low ridership forecasts are clear indications that the model parameters may need further examination. The evaluation of a proposed New Starts transit proj- ect relies on the cost-effectiveness ratio of the project. The cost-effectiveness ratio relates the cost of the project to the expected benefits, usually expressed as time savings, from the project. Obviously, the estimated cost of the project is independent of the travel modeling procedures; however, the expected user benefits are inextricably linked to the model- ing procedures and inputs. A major component of the FTA’s guidance on model properties, therefore, relates to the user benefits implied by the model. The FTA requires that models adequately support the case for a new transit project by cap- turing appropriate user benefits for various market segments. Further, the models should be amenable to an analysis of the primary causes of the benefits. The FTA recognizes that a range of modeling approaches can be used to obtain the desired model properties. These approaches could include either the traditional trip-based models or the more advanced tour and activity-based models, as long as due attention is paid to the model properties and the implied user benefits. In summary, the FTA recognizes good models based on coherent forecasts. Careful calibration and validation coupled with rigorous quality assurance checks will help achieve the ultimate objective of developing models to gain insights into performance and benefits of the alternatives. Rider Surveys Rider surveys are an important source of current transit information and are crucial to calibrating models that reflect the current conditions accurately. Where possible, the FTA recommends surveys before and after project opening to get a time-varying picture of ridership patterns and also to evaluate the model predictions. In cases where only the older survey data are available, the usefulness of the data in explaining current patterns depends to a large extent on the rate of growth in the metropolitan area as well as on any major transit system changes in the area. To the extent that these changes are minimal, the FTA deems the older data acceptable for current day predictions. The success of rider surveys in capturing the current transit travel patterns depends on the design of the surveys in terms of the sampling plan, the questionnaire, and the data items included in the questionnaire. The FTA recommends that the sampling plan be designed with the transit markets in mind. The transit markets are determined not only by the socioeconomic attributes but also by the geographic attributes such as the area type of the origin and/or destination of the trip. Because these markets have

A-5 different response rates and different travel patterns, the FTA urges sample allocation and survey methods that account for these differences and improve overall response rates. The FTA’s guidance on questionnaire design relates to the visual and interpretational aspects of the survey. Specifically, the FTA recommends that the surveys be simple in terms of layout, readability, and wording. Attention to these three aspects can help avoid round-trip reporting and can provide better data on trip origins and destinations. Successful surveys are succinct. Recognizing this, the FTA has identified several key data items that must be included in the surveys and several others that either require the use of discretion or are simply unnecessary. Figures A.1, A.2, and A.3 show the FTA’s comments on the usefulness of various commonly included traveler, trip, and other characteristics, respectively, in rider surveys. In addition to the rider surveys, the FTA recommends the use of other ridership data, where available, to inform the modeling process. These data could include on-off counts and park-and-ride utilization counts. Calibration and Validation As indicated previously, the FTA emphasizes that forecasts should be based on models that are tested rigorously against current transit ridership patterns. The FTA requires that the model forecasts serve as a useful basis for quantifying and understanding user benefits from the proposed New Starts projects. The implications of a careful calibration and vali- dation methodology are threefold: first, it necessitates better current data; second, it calls for a better focus on transit markets; and third, it requires better tests and standards. Source: Session 4: Data Collection, Slide 46 (Federal Transit Administration, 2007). Figure A.1. FTA comments on frequently included traveler characteristics. Source: Session 4: Data Collection, Slide 45 (Federal Transit Administration, 2007). Figure A.2. FTA comments on frequently included trip characteristics.

A-6 The FTA recommends that project sponsors take advantage of the funding and guidance opportunities available from the FTA to collect good quality “before” and “after” survey data. The issue of better focus on transit markets can be achieved through an evaluation of model performance by each trip pur- pose, socioeconomic group, production-attraction area types, and transit access modes. The FTA deems the matching of overall target totals as an insufficient measure of model calibra- tion. The standards for model calibration must rely as much on behavioral significance as they do on statistical significance. The FTA defines validation as a valid description of travel behavior as well as plausible forecasts of “deltas” for the future year. The FTA recommends careful documentation of key transit markets, current transit modes, and calibration forecasts to help evaluate the overall effectiveness of the model for New Starts analysis. The FTA provides guidelines on the allocation of resources to the three important tasks of model development, cali- bration, and validation. Because of the critical importance of model validation, the FTA recommends that estimation be conducted only where necessary and that the testing (calibration and validation) task be fully funded. In model estimation, statistical procedures are used to develop values for model parameters that will provide a best fit with observed travel data. The FTA’s guidance here indicates that it may be acceptable in many cases to transfer previously estimated parameters from another area’s model and then calibrate and validate them to local data in the new area. The FTA has provided guidance on specific properties of travel models to ensure proper calibration and validation. The FTA has found that many travel models have one or more of the following problems: • Unusual coefficients in mode choice models; • Bizarre alternative-specific constants; • Path/mode choice inconsistencies; • Inaccurate bus running times; and • Unstable highway-assignment results. Since naïve calibration leads to bad alternative-specific constants and has the cascading effect of producing errors in trips and benefits, the FTA suggests that modelers ask them- selves if patterns across market segments are explainable. The FTA also suggests that there be conformity between parameters used in transit path selection and mode choice utility expressions for transit choices. That is, the path-building process must weigh the various travel time and cost compo- nents in a manner that is consistent with the relative values of the mode choice coefficients. The consequences of inconsisten- cies include the following: • Better paths may look worse in mode choice; and • Build alternatives may lose some trips and benefits. The FTA requires that level-of-service estimates for transit (and highway) must: • Replicate current conditions reasonably well; • Predict defensible deltas by comparing conditions today versus the future; and • Predict defensible deltas when comparing conditions across alternatives. The FTA recommends a careful analysis of highway and transit travel times between carefully selected origins and destinations to understand the quality of the model networks. Spurious values of travel time can distort the magnitude as well as the pattern of predicted trip making and can adversely affect the quality of project user benefits. A.3.2 Summary of FTA Guidelines The FTA’s requirements are geared toward reasonably accounting for current patterns and predicting reasonable future ridership for the proposed New Starts projects. The FTA does not provide rigid targets for parameters in travel Source: Session 4: Data Collection, Slide 47 (Federal Transit Administration, 2007). Figure A.3. FTA comments on frequently included other characteristics.

A-7 models. Rather, the FTA recommends methods that can be used to ensure that models reflect current travel behavior and predict reasonable future patterns. The FTA’s expectations from travel models and the New Starts process can be summarized as follows: • Coherent narrative of the model parameters, inputs, and outputs; • Regular and early communication regarding model param- eters and forecasts to ensure that the agency/sponsor is proceeding in the proper direction; • Reasonable model forecasts in light of the expected land use growth, service characteristics, and other project-related attributes; and • Proper documentation and uncertainty analysis, which is directly related to the requirement of the Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU) of 2005 that asks the FTA to provide the U.S. Congress with an assessment of contractor performance. The FTA will rate contractors based on the following measures: – Comparison of predicted and actual ridership; – Quality of documentation; – Uncertainty analysis, including magnitude of impact; and – Before and after studies for various stages, including alternatives analysis, preliminary engineering, pre-project construction, and 2 years after opening. References Federal Highway Administration (2009). “Certification Checklist for Travel Forecasting Methods.” http://www.fhwa.dot.gov/planning/ certcheck.htm (As of September 15, 2011.) Federal Transit Administration (2007). Travel Forecasting for New Starts Workshop, St. Louis, Missouri. http://www.fta.dot.gov/documents/ Sessions_01-04.pdf (As of February 8, 2012.)