Modeling land-use change using partitioned vector cellular automata while considering urban spatial structure

Abstract

While many published studies have explored the impact of spatial heterogeneity on land-use change, few have focused on regional differences in land-use transition rules caused by urban spatial structure. In this paper, we measured urban land-use diversity by developing self-adaptive kernel density estimation and entropy weight methods and determine the urban spatial structure (composed of urban regions, inner and outer urban-rural fringes, and a rural hinterland) by applying a spectral clustering method. Combining local neighborhood effects and environmental effects, the land-use transition rules of different types of regions were mined to construct a partitioned vector cellular automata (CA) model that zonally simulates urban land-use change. The proposed model was applied to the simulation of the land-use change process in Jiangyin City, China, from 2007 to 2017. The resulting simulation accuracy was higher than that of other well-accepted CA models that do not consider urban spatial structure, and the conventional neighborhood assimilation rule was found not to be applicable to the conversion of construction land. The results and findings demonstrate that the proposed model is an effective means for urban planners to simulate and analyze urban evolution processes of cities with urban spatial structures that fit a concentric circle model.