Prediction of spatial land use changes based on LCM in a GIS environment (A case study of Sarabeleh (Ilam), Iran

This study aims at monitoring and predicting land use changes using LCM module in the Sarabeleh area. Land sat images of year 1988، 2001 and 2011 were employed to produce digital land use maps. The images were classified into five classes including forest، rangeland، barren land، agriculture and residential areas. LCM module in Idrisi GIS software was used to analyze the land use changes and predict the land uses status in 2011، based on artificial neural network (ANN) and Markov Chain analysis. ANN was trained with various influencing factors include distance from road، distance from residential areas، distance from forest edge، fragmentation index، elevation، slope and aspect. The results indicated that 14691 hectares of the forest cover have been degraded during the period 1988-2011. Furthermore، the barren land areas have been increased 9874 ha in comparison to initial situation. The results of transition power modeling using artificial neural network showed high accuracy in most of the sub modules (60-86%). The total error in modeling for the year 1390 image obtained as12/84% which represents conformity between the predicting of mode land ground truth image and acceptability of the model. Also، the result shows that forest area will decrease in the year 1400 compare to 1390 while، barren lands increase.