Investigation of Markov Chain Model Efficiency for Estimating Land use and Land cover Changes Using Landsat Satellite Images

In recent decades, rapid changes of land use and land cover in the suburbs of major cities in Iran, especially Tehran as the capital, has been growing with important consequences such as the destruction of natural resources, environmental pollution and poor growth of cities. Land use changes and investigation of their causes and factors in a certain time period can be very vital for planners and managers. Purpose of this research is investigation of first grade Markov chain model efficiency for estimating land use changes and land cover for Tehran city and its suburbs during the period of 30 years. In this research multi-temporal LANDSAT satellite images of 1975, 1990 and 2005 were used. Initially, these images were classified in six classes and to the maximum likelihood method. Then, using land cover maps obtained for each period of 15 years, the transformation matrix for land cover classes were calculated between each both time periods. Finally, using the transformation matrix of the first period, the results of land use changes for 2005 was predicted by Markov chain model. Accuracy of the predict with Markov chain model were checked with the help of existing land use maps for 2005. The results showed that the highest percentage of difference is related to arable lands class (0/64622 percent) and the lowest percentage of difference is seen in green space class (0/00551 percent). It is worth noting that in all cases the magnitude of these differences was generally less than 7%, that this number is justified the useful and usable Markov model and its accuracy to predict changes well.