Predicting Land Use Change Affected by Population Growth by Integrating Logistic Regression, Markov Chain and Cellular Automata Models

Demographic change was considered to be the most major driver of land use change although there were several interacting factors involved, especially in the developing countries. This paper presents an approach to predict the future land use change using a hybrid model. A hybrid model consisting of logistic regression model, Markov chain (MC), and cellular automata (CA) was designed to improve the performance of the standard logistic regression model. Experiment was conducted in Giao Thuy district, Nam Dinh Province, Vietnam. Demography and socio-economic variables dealing with urban sprawl were used to create a probability surface of spatio-temporal states of built-up land use for the years 2009, 2019, and 2029. The predicted land use maps for the years 2019 and 2029 show substantial urban development in the area, much of which are located in areas sensitive to source protections. It also showed that aquacultural land changes substantially in areas where are in the vicinity of estuary or near the sea dike. There was considerable variation between the communes; notably, communes with higher household density and higher proportion of people in working age have larger increases in aquacultural areas. The results of the analysis can provide valuable information for local planners and policy makers, assisting their efforts in constructing alternative sustainable urban development schemes and environmental management strategies.