LAUSR

Urban expansion is a hot topic in land use/land cover change (LUCC) researches. In this paper, maximum entropy model and cellular automata (CA) model are coupled into a new CA model (Maxent-CA) for urban expansion. This model can help to obtain transition rules from single-period dataset. Moreover, it can be constructed and calibrated easily with several steps. Firstly, Maxent-CA model was built by using remote sensing data of China in 2000 (basic data) and spatial variables (such as population density and Euclidean distance to cities). Secondly, the proposed model was calibrated by analyzing training samples, neighborhood structure and spatial scale. Finally, this model was verified by comparing logistic regression CA model and their simulation results. Experiments showed that suitable sampling ratio (sampling ratio equals the proportion of urban land in the whole region) and von Neumann neighborhood structure will help to yield better results. Spatial structure of simulation results becomes simple as spatial resolution decreases. Besides, simulation accuracy is significantly affected by spatial resolution. Compared to simulation results of logistic regression CA model, Maxent-CA model can avoid clusters phenomenon and obtain better results matching actual situation. It is found that the proposed model performs well in simulating urban expansion of China. It will be helpful for simulating even larger study area in the background of global environment changes.