Spatial Disaggregation of California Freight Demand

In recent years the role of statewide freighting forecasting models has been expanded to much finer levels of analysis than regional or even county levels, those being the most disaggregate spatial levels for which public freight data sources are typically available. In partnership with other state agencies and Metropolitan Planning Organizations (MPOs), Intelligent Transportation System Irvine (ITSIrvine) has completed development for the California Department of Transportation (Caltrans) of the California Statewide Freight Forecasting Model (CSFFM). A critical challenge was to provide a framework for organic integration between the CSFFM and a finer spatial level such as that in the new California Statewide Travel Demand Model (CSTDM) to meet Caltrans and MPO needs. Factoring methods are currently largely used for disaggregating freight demand. Such methods cannot adequately capture the complex structure and behavior of freight movements, advances in logistics, information technology, and relocating infrastructure at the MPO level. One advantage of the CSFFM, modal path-based origin-destination (OD) representation, cannot be fully utilized by MPOs because factoring methods tend to break the chains of modal path-based information in the conversion to tripbased information. This research initially sought to explore and develop truck tour-based models for disaggregating CSFFM from an aggregate Freight Analysis Zone (FAZ) level to the more disaggregate Traffic Analysis Zone (TAZ) level in CSTDM, by using truck GPS data from the American Transportation Research Institute (ATRI). Expected results included new and improved insights into the spatial and temporal operations of trucks at the urban and MPO level, contribution to the statewide-related component of urban freight modeling, and an evaluation of traffic and environmental impacts of state-level policies and air pollution mitigation strategies. However, after detailed investigation of the ATRI GPS data it was concluded that several problems with the data made it inadequate for disaggregating a CSFFM truck matrix for about 200 FAZs to the CSTDM 5000 TAZ level. Therefore, a new approach was developed. It involved estimation of a direct demand model at the CSFFM FAZ level using as inputs only independent variables readily available at CSTDM's level of aggregation and, as dependent variables, the final truck matrices estimated by CSFFM. CSFFM outputs can then be applied to CSTDM's zoning system with the resulting estimates being appropriately scaled.