The impact of zoning structure on statewide traffic demand modeling

Spatial analysis and modeling frequently involves the use of aggregated zonal data. The literature suggests that the definition of zonal systems can impact the analysis and modeling results. Traffic analysis zones used to represent the spatial structure of the study area are basic to traffic demand models. Little is known about the extent and degree of the impact of a zonal system on statewide traffic demand modeling. With the assistance of a Geographic Information System (GIS), this study designed eleven zoning structures, combining three levels of census geography and four types of centroid location. The purpose was to explore the effects of different zoning structures on statewide traffic demand modeling. Research scenarios were examined and compared using several measures, such as relative difference, correlation coefficient and percent Root Mean Square Error between estimated volumes and observed ground counts. Statistical analysis was also conducted to identify the possible differences among different zoning structures. The model performances varied significantly at different levels of TAZs as smaller zones had better modeling results. Additionally, the results showed the effect of centroid location, which tended to increase with an increase in TAZ sizes, and the interaction effect between the zoning structure and the detail of network. In summary, the study has confirmed the impact of zoning structure on statewide traffic demand modeling and the important role of GIS in transportation planning.