by Charles CHEUNG, John BLACK, and Paul VAN DEN BOS
Researchers have clearly demonstrated the weaknesses in the gravity model specification. Yet the model remains today at the heart of the four-step modeling (for example, TransCAD) used in practice in the urban transportation planning process. There is an array of suitable statistical measures to test model goodness-of-fit against survey origin-destination (O-D) data that allow calibrated model specifications to be evaluated and the best model selected; but the implications of inaccuracies in trip distribution models are avoided by practitioners. The aggregated gravity model, one stratified by industry and occupation, and an intervening opportunities model are calibrated on journey-to-work Census data for Sydney. O-D residuals are assigned to the transport network to check for spatial bias using the TransCAD software to pinpoint where investment decisions may have been based on either over- or under-estimation of traffic flows. The implications of these findings for transportation policy and infrastructure investment are articulated. The conclusions point towards a need for research and development into improved spatial interaction models.
more about urban travel behavior:
No comments:
Post a Comment