Travel Forecasting for the Capital Investment Grant Program

For project evaluation in the Capital Investment Grant Program, FTA allows the use of region-wide travel models, incremental data-driven models, or FTA’s Simplified Trips-on-Project Software (STOPS). The choice is at the project sponsor’s option.

Region-wide travel models are for an entire city or metropolitan area. For more information on these types of models, please see TFResource's urban and metropolitan model page.

Incremental data-driven models are compelled by data rather than by models. They provide a straightforward approach to representing predictable project situations. The types of transit travel data used for these types of models are boarding-to-alighting, origin/destination, or station-to-station flows. Service level changes are applied to the flow data. These changes reflect the improvements and impacts of the proposed project. The changes in service levels are then applied to flow data via elasticities, arithmetic representations of how sensitive transit trips are to changes in the service level variables. The elasticities are transferable across different urban areas if the type of data is also consistent. Data-driven models can be developed in a spreadsheet program, or applied in regular travel modeling software. They generally require few resources to set up and execute.

FTA's Simplified Trips-on-Project Software (STOPS) was released in September 2013. It is a limited implementation of the conventional “4-step” travel model[1]. More information on STOPS can be found here (opens new window).

# FTA Considerations

For region-wide travel models and incremental data-driven methods, FTA will consider five aspects[2]:

  1. The adequacy of the current transit survey and data,
  2. The forecasting method properties (its structure and design, consistency with good practice),
  3. How well the methods understands existing travel markets and transit travel patterns,
  4. The reasonableness of inputs (demographics, service changes) used in the forecasts, and
  5. The plausibility of the forecasts for the proposed project, especially if the forecasts are reasonable and commiserate with the proposed changes to the transit system and consistent with similar projects in the region or around the country.

If the project sponsor selects STOPS, then only the last two aspects are considered by FTA.

# Adequacy of current transit survey and data

The most recent transit survey and data should sufficiently reflect current riders and their travel patterns. Some important characteristics of the survey are:

  1. Age - How recent was the survey taken? Surveys taken more than 5 years ago may not reflect current travel patterns and rider characteristics.The age of auxiliary data, such as passenger and parked vehicle counts, is also important. The applicability of older data depends on the magnitude and types of changes in the region or corridor, changes to the transit system or ridership.
  2. Variables - Does the transit survey include the variables that are most important? For example, transit surveys are generally expected to record at least the geo-coded origin and destination, boarding and alighting stops/stations, trip purposes, access and egress modes, the transit path (i.e., route and modal sequence), and basic rider socio-demographic characteristics.
  3. Sampling plan - Has the sampling plan been designed to produce a sample representive of actual riders, and address known non-response biases?
  4. Data quality - What levels of quality assurance scrutiny have been undertaken to ensure that data collected is consistent across all data items within a survey record?
  5. Expansion - Is the expansion process sufficiently dis-aggregated enough to compensate for known non-response biases?

# Forecasting Method Properties

Regardless of the forecasting method chosen, it should meet the following objectives:

  1. Be consistent with good travel modeling practice,
  2. Understand existing travel markets and transit travel patterns,
  3. Be able to produce reasonable forecasts for new services (if applicable), and
  4. Quantify FTA evaluation measures (as needed).

Please note that the being consistent with good travel modeling practice does not mean the model has to conform to the newest or latest modeling trends.Activity-based, traditional trip-based, data-driven/simplified models, or spreadsheet models may meet the above four properties depending on the project. Some issues that are inconsistent with good modeling practice are: over-specification of utility constants in the mode choice model, unrealistically low/erratic highway and transit speeds, inconsistent weighting of travel components throughout each modeling phase and insufficient network coding.

# Understanding Existing Travel Markets and Travel Patterns

The forecasting method should be tested to ensure that it fully grasps the existing travel markets and travel patterns. Generally this is performed by comparing the results of a transit on-board survey to the forecasting method's outputs across key travel markets.Travel markets can be defined in any number of ways: trips made by transit dependent populations, suburban-to-core work travel, and intra-downtown travel are just a few common ones. Travel markets depend on whom the project is trying to serve.

# Reasonableness of Inputs

Inputs are not limited to model inputs, but more precisely should be thought of inputs to the mode choice models:

  1. Population and employment data - do the population and employment data conform to existing land uses (for current year forecasts) and proposed land uses (for horizon year forecasts)?
  2. Travel flows, especially with respect to the key travel markets - do the general (i.e., amodal) travel flows reflect known conditions, or do they reasonably reflect expected changes to the proposed land use, demographic and transportation system?
  3. Transportation system network - does the network properly reflect the speeds, capacity and accessibility?

# Plausibility of Forecasts

The plausibility of the forecasts can be confirmed if they are found to be similar in nature with similar projects in the region (if they exist) or around the country. For truly unique projects, the plausibility can be ascertained by comparing the expected ridership changes for key travel markets against the proposed changes in the transit system.

# Other pages in development

  1. Travel Forecasting Results Report

# References

  1. FTA's page on STOPS, (opens new window) ↩︎

  2. FTA's New Starts page on Travel Forecasts, (opens new window) ↩︎

This site uses cookies to learn which topics interest our readers.