Abstract:
The development of travel demand model networks from geographic information system (GIS) datasets is described, based on work done for multi-state metropolitan planning agencies, project studies in small city settings, and a statewide traffic forecasting model. Various public access multi-jurisdictional GIS street layers were used to create such networks, based on their availability and any enhancement by local agencies, and they are compared based on various criteria for “model readiness.” These include how “data-rich” they are, how well they overlay (or otherwise join) other GIS datasets with necessary data, how accurately/properly they depict traffic access patterns, if the density of endpoints/nodes is appropriate to the task, and street segment connectivity through such nodes. Data were joined to networks from a variety of sources, some of which are not traditionally used in travel forecasting including railroad crossings and roadway curve and grade files. The modeling philosophy utilized in all these settings follows that developed for Cedar Rapids, Iowa in the 1990’s, and comparable levels of success are documented for base year model validation along with proper sensitivity to roadway and traffic control features in developing project design year forecasts.
