Monitoring and modeling land use/cover changes in Arasbaran protected Area using and integrated Markov chain and artificial neural network

Abstract

Land cover is rapidly changing across the globe. An understanding of how land use/cover changes (LUCCs) are made can lead to appropriate management measures to assess environmental changes and to develop policies and plans for sustainable land cover in the future. In this study, we used Land Change Modeler (LCM) to predict LULC by 2030. LUCC was detected using Landsat satellite images, TM, ETM+, and OLI sensors for the years 1987, 2002, and 2017. Transmission potential was modeled using an artificial neural network (multilayer perceptron) and seven variables. LUCCs were modeled for 2017 using the Markov chain with the calibration period 1987–2002, and then, we validated the LULC map that was predicted through LCM by comparing it with the ground reality map of 2017. Finally, LULCs were predicted for 2030 using the calibration period 2002–2017. The largest change in all periods is associated with the change of forest lands into other land uses. Prediction for 2030 suggest that approximately 6000 ha of forest land will change into pastures and 248 ha into agricultural lands.