Modeling Trip-generation and Distribution using Census, Partially Correct Household Data, and GIS

The efficiencies of urban transport systems in several cities are drastically affected due to difficulties imposed by rapid urbanization and the proliferation of private modes of transport. The conventional four-stage travel demand modeling approach provides an ideal platform to formulate strategies to rectify problems in urban transport. Trip generation is the first stage in this exercise (where trip production and trip attractions are modelled), followed by trip distribution in the second stage. The present work related to the development of models for trip generation and trip distribution necessitated the use of census data related to the number of households in each zone since the available revealed preference (RP) data compiled based on household interview surveys was partially incorrect. A review of the literature indicated that studies on the use of sparsely available and partially inaccurate data such as revealed preference and zone-specific secondary data on trip generation and trip distribution were limited. In the present study, the use of the initial trip generation regression models developed based on existing household survey data resulted in prediction errors ranging between 26% and 32%. Modeling efforts after applying corrections to zone-specific characteristics based on secondary data and the use of trip rate per household later resulted in prediction errors of less than +/- 5%. In the latter phase of work related to trip distribution modeling, a log-linear regression model was developed based on a smaller refined set of the revealed preference data obtained by eliminating erroneous data in a stage-wise manner. The use of the calibrated and validated model ensured that the errors in predicted trip frequencies were less than 0.6%. Here, the information on the inter-zonal aerial distances that formed part of the trip distribution model was obtained using GIS approaches that employed the moment area method, which considered the intensity of land use at the sub-zone level. The combined strategy incorporates the use of GIS-based approaches to determine inter-zonal aerial distances, and the use of the refined relationship between trip interchanges and the inter-zonal aerial distances in the development of a reliable log-linear regression model for trip distribution contributed towards attaining higher accuracies in travel demand estimation. The modeling approaches described herein do not rely on the use of sophisticated technology, and time-consuming data processing. The study will provide the basic framework for transport planners to formulate better strategies for travel demand modeling where available data is noisy and less reliable.