Fusing long and short distance travel in the Colorado Statewide Model
Corresponding Author: Jeffrey Newman, Cambridge Systematics, Inc.
Presented By: Jeffrey Newman, Cambridge Systematics
Statewide travel modeling for larger states has in most instances been undertaken as a fusion of two travel models: a “typical” daily travel model that includes shorter and more frequent trips, plus an "atypical" travel model that includes long distance trips that are undertaken infrequently. For example, this approach has been used for statewide models in Ohio and California. In developing a statewide model for Colorado, we have adopted an alternative approach: fusing both typical and atypical travel into a single activity-based model of travel behaviors. This approach was adopted for Colorado for several reasons, including the fact that imposing any reasonable distance-based cutoff to differentiate these two models would move substantial portions of the daily commuting and tours between metropolitan areas, and even some intraregional tours, into the long-distance realm. Two important challenges arise from building this kind of model using a traditional household travel survey: the proper handling of non-closed tours (e.g. overnight travel) in the daily activity diary survey, and the fusion of single-day diary household survey data with longer period long distance survey data.
As is common for household travel surveys, the Front Range Travel Counts survey that is being used for the Colorado Statewide model was conducted in two parts: a detailed travel diary containing all trips made in a 24 hour period, plus a long distance travel log, containing only longer trips (nominally, those over 50 miles) across a 14 day period. Because the single day diary is truncated at 24 hours, for many respondents it contains trips in non-closed tours, because they were not at home at the start or end of day, or both. These instances may include “typical” activities that happen to occur over night (e.g. worker on the night shift) or “atypical” activities related to other overnight travel. Because the Colorado model explicitly includes both of these within the single model framework, it was important to include these tours in model estimation. This paper describes the methodology adopted for explicitly modeling non-closed tours, and for fusing daily diary and long distance log data into a single modeling framework.