-
Notifications
You must be signed in to change notification settings - Fork 5
Travel Model One Overview
MTC and ABAG use a travel model (also known as a travel demand model or travel forecasting model) to describe travelers' reactions to transportation projects and policies and quantify the impact of individual decisions on the Bay Area’s transportation networks and environment. The travel modeling suite includes three main analytical tools: a population synthesizer, a travel model, and a vehicle emission model. Each tool is described below. While the travel model represents most strategies and policy interventions in the plan, the travel model cannot represent some transportation strategies. The analysis for these policy outcomes are detailed in the Off-Model Calculations section.
MTC and ABAG’s travel model is an agent-based simulation where the "agents" are individual households and their members. The model simulates the behavior of individuals and households as they carry out daily activities, influenced by input land development patterns and transportation projects and policies. Each agent must be characterized in detail for this simulation.
Population synthesizers are software programs that create lists of households and persons for travel model simulations. For Plan Bay Area 2050+, MTC and ABAG used PopulationSim. This synthesizer samples households from the 2017-2021 Census Public Micro-sample (PUMS) data to match population dimensions spatially, ensuring aggregate sums align with other Census summary tables or land use projections from the land use model. For example, if BAUS2 forecasts that 60 households with 100 workers and 45 children will live in spatial unit X in 2035, PopulationSim will locate 60 PUMS households in spatial unit X, selecting households so the number of workers is close to 100 and the number of children is close to 45.
The population synthesizer “controls” (i.e., minimizes the discrepancy between the synthetic population results and the historical Census results or the land use forecasts) at the travel analysis zone (TAZ) along the following dimensions:
- Number of total households (individuals living in non-institutionalized group quarters, e.g. college dorms, are counted as single-person households);
- Number of total households by size (four categories: 1, 2, 3 or 4+);
- Number of households by income quantile (four income quantiles as defined in Table 7);
- Number of households by number of workers (four categories: 0, 1, 2, 3+);
- Number of persons by age (five categories: 0-4, 5-19, 20-44; 45-64; 65+) and,
- Number of persons living in non-institutionalized group quarters by type (three categories: college dorm, military, and other non-institutional group quarters)
Travel Model One codebase: https://github.com/BayAreaMetro/travel-model-one.
Travel models are frequently updated, so specifying the version used for analysis is useful. The current analysis uses MTC/ABAG’s Travel Model 1.6 (version 1.6.1), released in May 2025, calibrated to 2015 and 2023 conditions (with the latter being a targeted calibration and not full), and validated against 2015 and 2023 conditions. Travel Model 1.6 is a minor update to Travel Model 1.5, which was used for Plan Bay Area 2050. Developed to support the needs of Plan Bay Area 2050+, Travel Model 1.6 added a sub-model to represent working from home (telecommuting) and included a targeted mode choice recalibration for 2023 to capture post-COVID mode choice preference changes. Please see Development for more information on Travel Model One version history.
TM1.6 is an "activity-based" model. It is a partial agent-based simulation where the agents are households and individuals residing in the Bay Area. It does not simulate individual behavior of passenger, commercial, and transit vehicles on roadways and transit facilities but does simulate the behavior of aggregated vehicles and transit riders. The travel model simulates a typical weekday when school is in session, the weather is pleasant, and no major incidents disrupt the transportation system.
The model operates on a synthetic population representing each actual household and person in the nine-county Bay Area, both historically and prospectively. Travelers move through travel analysis zones (TAZs; map of TAZs) and use the transportation system. The model simulates a series of travel-related choices for each household and person within each household, organized sequentially:
- Usual workplace and school location – Each worker, student, and working student in the synthetic population selects a travel analysis zone in which to work or attend school (or, for working students, one zone to work and another in which to attend school).
- Household automobile ownership – Each household, given its location and socio-demographics, as well as each member’s work and/or school locations (i.e., given the preceding simulation results), decides how many vehicles to own.
- Daily activity pattern – Each household chooses the daily activity pattern of each household member, the choices being (a) “mandatory” travel, defined as going to work or school, (b) "non-mandatory" travel, defined as other activities outside the home, or (c) activities at home . Note that people whose daily activity pattern includes work or school tours may also make tours for other purposes as well.
- Work/school tour frequency and scheduling – Each worker, student, and working student decides how many round trips they will make to work and/or school and then schedules a time to leave for, as well as return home from, work and/or school.
- Joint non-mandatory tour frequency, party size, participation, destination, and scheduling – Each household selects the number and type (e.g., to eat, to visit friends) of “joint” (defined as two or more members of the same household traveling together for the duration of the tour) non-mandatory (for purposes other than work or school) round trips in which to engage, then determines which members of the household will participate, where, and at what time the tour (i.e., the time leaving and the time returning home) will occur.
- Non-mandatory tour frequency, destination, and scheduling – Each person determines the number and type of non-mandatory (e.g., to eat, to shop) round trips to engage in during the model day, where to engage in these tours, and at what time to leave and return home.
- Tour travel mode – The tour-level travel mode choice (e.g., drive alone, walk, take transit) decision is simulated separately for each tour and represents the best mode of travel for the round trip.
- Stop frequency and location – Each traveler or group of travelers (for joint travel) decides whether to make a stop on an outbound (from home) or inbound (to home) leg of a travel tour, and if a stop is to be made, where the stop is made, all given the round trip tour mode choice decision.
- Trip travel model – A trip is a portion of a tour, either from the tour origin to the tour destination, the tour origin to a stop, a stop to another stop, or a stop to a tour destination. A separate mode choice decision is simulated for each trip; this decision is made with awareness of the prior tour mode choice decision.
- Assignment – Vehicle trips for each synthetic traveler are aggregated into time-of-day-specific matrices (i.e., tables of trips segmented by origin and destination) that are assigned via the standard static user equilibrium procedures to the highway network. Transit trips are assigned to time-of-day-specific transit networks.
See Model Schematic for a graphic representation of the choices and sequence.
Additional information on travel model system design, UECs, key inputs, properties & configs, and outputs, refer to https://github.com/BayAreaMetro/modeling-website/wiki/UsersGuide.
The MTC/ABAG travel model generates spatially and temporally specific estimates of vehicle usage and speed for a typical weekday. This data is then input into an emissions model to estimate on-road mobile source criteria pollutants and carbon dioxide emissions (used as a proxy for all greenhouse gases). For the current plan air quality analyses, MTC and ABAG used the California Air Resource Board’s EMissions FACtor (EMFAC) 2014 for SB 375 calculations, EMFAC 2021 for off-model strategies’ emission calculations and Plan Bay Area 2050+ Equity Analysis and Performance Analysis calculations, CT-EMFAC 2021 - adjusted to EMFAC 2025 particulate matter and exhaust emissions-for Plan Bay Area 2050+ EIR mobile source air toxic emission inventory estimation, and EMFAC 2025 for Plan Bay Area 2050+ EIR criteria pollutant emission inventory estimation.
See https://github.com/BayAreaMetro/modeling-website/wiki/Emissions-Modeling-at-MTC for more information on emissions modeling.
Travel Model One has been applied to various scenario planning efforts, transportation project assessments, and policy studies. Below are major application projects; this is not an exhaustive list.
Previously the MTC/ABAG Analytical Modeling Wiki (http://analytics.mtc.ca.gov/foswiki/Main/WebHome)
Please email [email protected] if you find anything missing here.