A common forecasting question that arises on most projects involving the California Environmental Quality Act (CEQA) is, “What travel demand model or forecasting methodology should be used?” To conduct transportation analysis that meets environmental regulatory conditions and provides a high level of confidence in the analysis results, analysts should follow state‐of‐the‐practice or best practice methodology. Most regulations applicable to transportation impact analysis, including CEQA, do not recommend a specific forecasting method or model. However, supporting guidance materials from responsible agencies such as Caltrans and court rulings on cases involving technical methodology suggest that following state‐of‐the‐practice or best practice methodology is expected. This expectation is grounded in the basic purpose behind environmental regulations like CEQA that attempt to accurately identify and disclose potential impacts and to develop effective mitigation. Having accurate and reliable travel forecasts is essential for meeting these expectations.
A key challenge in following the state‐of‐the‐practice is that it can vary depending on many factors; some of which are listed below:
- Complexity of the transportation network and number of operating modes
- Planned changes in the transportation network (particularly to major roads or transit systems)
- Availability of resources to develop and apply travel demand models
- Population and employment levels
- Congestion levels
- Regulatory requirements
- Types of technical and policy questions posed by decision makers
- Desired level of confidence in the analysis findings
- Anticipated level of legal scrutiny
In California, travel demand forecasts are generated using various forms of models that range from simple spreadsheets based on historic traffic growth trends to complex computer models that account for numerous factors that influence travel demand. According to Transportation and Land Development, 2nd Edition, ITE, 2002, the appropriate model depends on the size of the development project and its ability to affect the surrounding area. As projects increase in size, the likelihood of needing a complex model (such as a four‐step model) increases because of the number of variables that influence travel demand and transportation network operations. The study area can also influence the type of model needed especially if congestion occurs or if multiple transportation modes operate in the study area. Either of these conditions requires robust models that can account for the myriad of travel demand responses that can occur from land use or transportation network changes.
When using a forecasting model, the following four fundamental criteria should be met to ensure compliance with state of the practice expectations:
- The scale of the model should match that of the project. Most projects are local in nature meaning that they involve specific intersections, interchanges, highway or freeway segments, or corridors. Therefore, locally valid travel demand models should be used to develop traffic volume forecasts for these types of projects. Using regional travel demand models without modification to address the scale of the project is not appropriate.
- The model should be calibrated and validated within the study area. The model’s validation in the study area should be verified for each time period being forecast (i.e.., daily, AM peak hour, PM peak hour, etc.) and for each mode that is being analyzed.
- The model validation should include static and dynamic tests. Static and dynamic validation tests should include those specified in California 2010 Regional Transportation Plan Guidelines, California Transportation Commission, 2010 and Model Validation and Reasonableness Checking Manual, FHWA, 1997. Static tests verify that the model can match base year traffic counts closely while dynamic tests verify that the model contains an appropriate level of sensitivity
related to the types of transportation network or land use changes associated with the project. These tests are particularly important for projects involving tolling or multiple modes.
- The model’s land use or socioeconomic forecasts should be tested for reasonableness. Models are used to forecast travel demand for a specific horizon year (i.e., 20 years after construction). The land use and socioeconomic forecasts need to match the horizon year and be based on reasonable market conditions that reflect past and future development trends for the specific study area.