A Spatial Disaggregation Model to Improve Long-Term Land Use Forecasting with Transport Models Based on Zonal Data

This paper examines the application of spatial disaggregation model in forecasting long-term land use change. It discusses an approach for disaggregating the predicted results on the basis of zonal data. The zonal data are disaggregated for each cell and the aggregated cell data, obtained based on the building units in the base year, are then employed for land use forecasting. The requirements of the proposed model are addressed by conducting empirical analyses to answer two questions: ‘Which cells will be redeveloped before the target year?’, and ‘How much will these cells be redeveloped?’ To answer the first question, the proposed model calculates the probability of redevelopment of a cell using a binary logistic regression model. This approach is proven to be capable of identifying the cells to be redeveloped (with a success rate of 80.4%) based on the rank of the accessibility and density values for each cell. To answer the second question, the proposed model estimates the expected redeveloped floor space in the cell in terms of the location utility. The R^2 value is found to be approximately 69.5%, which is sufficiently high for a forecasting model. The results of this study are expected to be used to develop a more detailed disaggregation model to forecast changes in land use in urban regions.