Spatial diffusion model for simulation of urban land cover change

Spatial diffusion is a stochastic spatio-temporal process. Spatial diffusion phenomena occur because of the existence of a concentration gradient of information or material flow. At a regional level, most human-induced land-use and land-cover change, including urban land-cover change, can be considered as a spatial diffusion process. In this research, a spatial diffusion (SDIF) model has been developed to simulate urban land cover change at a regional level. This research aims to establish a theoretical framework for spatial diffusion modeling and build a spatial diffusion modeling system for simulation of urban land-cover change. The Chicago metropolitan region was selected as the study area. The land-cover data derived from Landsat images of 1972, 1985 and 1997 were used as seed data sets for the SDIF model to simulate and predict urban land-cover change. US Census information, biophysical and socioeconomic factors were incorporated into the SDIF model. Spatial restrictions such as protected areas were used to restrain future urban development. Data sensitivity analysis indicates that the SDIF model output is sensitive to census data. Based on the population, household and employment projections for 2020, SDIF simulation of urban land cover through the year of 2020 has been achieved. The SDIF model evaluation indicates that the overall modeling simulation accuracies for the years of 1985, 1997 and 2002 are 65.4%, 66.8%, and 72.2% respectively. The SDIF model performs well in comparison with the Dynamic Landscape Simulation (DLS) models for the simulation of landscape change for the same study area