Tests of a Family of Trip Table Refinements for Long-Range, Quick-Response Travel Forecasting

This paper addresses the problem of using traffic counts to ascertain zonal trip generation characteristics when performing quick-response travel forecasts. A family of origin-destination (O-D) trip table estimation methods containing three unexplored members (biproportional, uniproportional, and dynamic biproportional) is proposed to solve this problem. The family is tested on static planning networks for Tallahassee, Florida; Northfield, Minnesota; and Fredericton, New Brunswick, Canada. Results indicate that travel forecasting models can be made to match ground counts better by a simple factoring of origins, destinations, or both. The three methods that directly solve for origin or destination factors have computational and statistical advantages over full-matrix O-D trip table estimation procedures, and the results are qualitatively and quantitatively interpretable.