Determination of a Desirable Time Step for Quasi-Dynamic Urban Model with Sapporo Test Bed

Abstract

This paper presents an application of the Sapporo GIS based Test Bed, which has been built to support urban model testing and development. It briefly describes TRANUS Sapporo model, which has been calibrated both at one time point and for a time interval. The model is quasi-dynamic, which runs its land-use and transportation submodels sequentially with time lag. In this study, Sapporo Test Bed was an efficient tool to identify the inputs and parameters that are the main cause of uncertainty of the model results. The uncertainty is accumulated by the quasi-dynamic structure, which is a usual structure of the most existing urban models. Moreover, this paper has proposed a heuristic approach to determine a desirable time step for quasi-dynamic urban model. This is achieved by a simplified Monte Carlo simulation technique. Uncertainty in the model output is evaluated from various statistical indexes. It is found that a time step of 3.3 years is most desirable by trading off between uncertainty and accuracy. However, a 3.3-year time step may not be desirable from a practical viewpoint, we evaluated that a time step of 5 years is also acceptable compared with 3.3 years based on our simulation results.