Reversing the direction of the transportation planning process: Measuring transportation infrastructure constraints on land use with historical data.

摘要

Typically the "forward" transportation planning process is accomplished by first generating person or vehicle trips (trip generation), second distributing these trips to various transportation analysis zones (trip distribution), and third assigning these trips to the network (traffic assignment). In this case, however, a "backward" process was proposed, modified, and then applied to the Charlottesville area for the 1967 base year, culminating in a five-components modeling process. One of the components, however, was extremely weak and led the author to develop a different approach that replaced the first four steps. This new model was calibrated for 1967 and then applied for the 1979 and 1990 forecast years. Lessons learned were then used to calibrate the model for the 1979 base year and apply it for the 1990 forecast year, and to demonstrate it as an instrument for understanding land use limits that arise from traffic volumes.; In the revised approach, zonal trip ends are directly estimated from transportation system variables that are influenced by link volumes, roadway types, travel distances, and the geographical position of the zone. Additionally, the author regressed retail employment, nonretail employment, and population to zonal trip ends. After implementing this "direct" model for the base year, the calibrated model is then applied for the forecast year. The success of these forecast year predictions is evaluated by comparing the predicted values to known values for the base and forecast years. Errors on the order of 50% were obtained overall, with larger values for retail employment and smaller values for nonretail employment and population.; The "given" in this dissertation is the forecast year traffic volumes and the dependent variable is socioeconomic parameters, such as zonal population, for the forecast year. Suggestions about how this model formulation might be interpreted to yield land use limits as a function of traffic volumes are discussed. A simple but perhaps significant finding for how to achieve convergence with the iterative entropy maximization method is outlined. Data issues associated with linking 1967, 1979, and 1990 Charlottesville data are explained. The use of historical data to predict the present is emphasized.