To study the dynamic changes of land use and predict the future land use scenarios based on the current land use, this paper uses Cellular Automata-Markov (CA-Markov) model to simulate the landscape pattern in 2030. The results show that in the study area during the period 1980-2005, grassland and construction land increased, and woodland increased slightly; waters and unused land decreased, and arable land underwent dramatic changes. The simulation precision of CA-Markov model is 87.28%, indicating that the use of it for simulation is reliable. The land use of the study area will be changed greatly in the future. This method provides a reference for the regions to carry out land prediction, and the research results can provide a basis for the study of optimization of land.