Developing a high-fidelity GIS-based travel demand model framework for improved network-wide traffic estimation

Abstract

The 4-step Travel Demand Model is usually based on roadway networks, which exclude roads of lower functional classification and are comprised of coarse traffic analysis zones (TAZs). Coarse zones (mostly in rural areas) tend to yield a higher percentage of intrazonal trips, which are not accounted for because the travel demand model can only capture those trips across the zone boundaries, resulting in fairly high model estimation errors. The special attention given to rural zones in this dissertation is motivated by the fact that they significantly influence the modeling accuracy and network-wide traffic volume assignability, especially to lower functional classification roads (locals and collectors). In an effort to overcome the shortcomings of the traditional model, this dissertation developed a GIS-based high-fidelity travel demand model (HFTDM) zonal enhancement framework capable of generating finer-grained spatial resolution TAZs. The resulting TAZs are more uniform in size and capable of determining both trip productions and attractions and can be used to achieve network-wide traffic volume estimation with improved accuracy to include local and collector roads. The developed (HFTDM) zonal enhancement framework was not expected to produce accurate traffic estimates on all links in the road network due to the lack of local travel surveys used to calibrate at each stage of the modeling process. To develop the TAZ structure enhancement procedure, the dissertation presents a methodological, systematic, GIS-based framework by integrating the travel demand modeling software platform, remotely-sensed images, parcel-based digital property maps, AZTool aggregation algorithm, and areal interpolation technique. This dissertation developed an HFTDM zonal enhancement framework for the Greater Fredericton Area (GFA) in the province of New Brunswick based on well-designed 10 individual-TAZ structure resolutions. The developed HFTDM zonal enhancement framework showed a general trend of network-wide incremental improvement in both modeling accuracy and assignment coverage capability along with increasing study area zonal resolution from coarse-grained to finer-grained zones. The case study showed that increasing the GFA spatial resolution from the coarsiest TAZ structure at Census Tract (CT) level (27 CT TAZs) to the finest TAZ structure at 4252 “fine” TAZs resulted in an improvement to modeling accuracy represented by an increase in the coefficient of determination, R², by 0.4092 (from 0.2490 to 0.6582) and an improvement in traffic assignment coverage by 46% (from 29% to 75%).