Accessibility measures represent an important component of an activity-based model (ABM) that expresses closeness of the modeled individual to potential locations where the activity “supply” (corresponding land-use type, employment, etc) is present. While there is little doubt about the accessibility is a strong determinant of individual activity patterns, there is no consensus about how to measure and represent it functionally. Accessibility measures to support ABMs should be user-centric, i.e., these measures must be able to quantify what people perceive and what affects their behavior. They also should be context-specific, i.e., particular to a certain choice dimension such as household auto-ownership, mode, time-of-day (TOD), etc., to support multiple sub-models in an advanced ABM. In most cases, temporal variation of accessibility is incorporated implicitly using TOD-specific level-of-service measures. However, there is also a dimension of spatial accessibility to activity opportunities which varies by TOD due to opening and closing hours for different business that is highly differential across different areas in the region.
In this study, we develop TOD-specific gravity type accessibility measures that account for both time-varying skims and land-use (activity-supply) effects. The developed context-specific accessibility measures were used as explanatory variables in various behavioral choice models of the ABM system developed for the Southern California region. Each of these measures corresponds to a given activity purpose, person type, and are sometimes further segmented by travel arrangement type (individual or joint), auto availability, and/or mode depending on what specific choice dimension is modeled. Special effort was made to make these accessibility measures properly differentiated by hour of day so that they can be linked to the corresponding time-of-day specific choices.
The paper includes a detailed analysis of accessibility measures themselves as well as their impacts on travel choices. With this component in place, the entire ABM becomes more sensitive to land-use and other related policies that target spatio-temporal distribution of activities and trips. In particular, it opens a way to evaluate transit-oriented and pedestrian-oriented land-use development scenarios from all perspectives (number of trips, mode shifts, time-of-day shifts, trip chaining patterns, propensity to carpool, etc).