Using Markov chains to analyze changes in wetland trends in arid Yinchuan Plain, China

Three wetland distribution maps were drawn using Two Land-sat 5 Thematic Mapper (TM) images from 1991 and 1999, and a China–Brazil Earth Resources Satellite (CBERS)-02B image from 2006. A transition probability matrix was constructed using two wetland distribution maps, one from 1991 and one 1999 and GIS (Geographic Information System). The trends in changes in wetland types and the distribution area were predicted using a Markov model. The prediction model was then tested for relative accuracy and the feasibility of a x2x2 test. The prediction model’s relative accuracy was 98.5%. The x2x2 test results showed that both the simulated results and the actual wetland distribution area were in good agreement. Therefore, it is feasible to use the wetlands area transfer matrix to establish a transition probability matrix based on the Markov model and to predict the distribution pattern of the wetland in Yinchuan Plain. The results of this study may be helpful for local governments to develop wetland management policies.