Simulation of urban expansion patterns by integrating auto-logistic regression, Markov chain and cellular automata models

This research analyses urban expansion patterns and their eco-risks in the Poyang Lake region in China. A hybrid model consisting of auto-logistic regression, Markov chain and cellular automata (CA) is designed to improve the performance of the standard logistic regression model. An eco-risk assessment (ERA) index by integrating landscape fragmentation index and area weighted eco-service value index is established to promote the effectiveness for dynamically evaluating the environment and eco-security in watersheds. Scenario predictions are introduced to better understand the relationship between urban dynamics and their eco-risks. Three urban development scenarios such as historical development trend (HDT), environment protection priority (EPP) and goal-oriented restriction (GOR) are designed and transplanted into the CA model through the parameter self-modification method. The quantitative analysis results showed that in the period of the past five years, the urban growth primarily concentrated in the metropolitans. The simulations show that under the HDT scenario the urban growth will mainly emerge in the metropolitans, while under the EPP and GOR scenarios the urban growth will expand along with the metropolitans or the road networks and highways, respectively. Moreover, the ERA demonstrated that the GOR scenario was more effective in meeting the goal of environment protection and urban sustainable development for the study area.