Predicting land use changes in northern China using logistic regression, cellular automata, and a Markov model

Abstract

Land use changes are complex processes affected by both natural and human-induced driving factors. This research is focused on simulating land use changes in southern Shenyang in northern China using an integration of logistic regression, cellular automata, and a Markov model and the use of fine resolution land use data to assess potential environmental impacts and provide a scientific basis for environmental management. A set of environmental and socio-economic driving factors was used to generate transition potential maps for land use change simulations in 2010 and 2020 using logistic regression. An average relative operating characteristic value of 0.824 was obtained, indicating the validity of the logistic regression model. The logistic–cellular automata (CA)–Markov model was calibrated by comparing the actual and simulated land use maps of 2010. A match of 83.7% was achieved between the predicted and actual maps of 2010, which represented a satisfactory calibration. This indicated that the integration of the logistic regression, CA, and Markov model has a high potential for simulating land use change in northern China. The calibrated hybrid model was implemented to obtain a land use map for 2020. The results showed a new wave of suburban development in the southwestern, west, and northwestern parts of the study area during 2010–2020. In addition, urban expansion has been accelerating, which is very likely to exacerbate the extensive environmental pollution currently existing in this area. Moreover, rapid urban expansion has resulted in significant decreases in forest areas and agricultural lands.