Prediction of Land Cover Changes in Horizon of 2028 through a Hybrid Model of Markov Chain and Cellular Automata;Catchment Area around Bazangan Lake Case Study

Abstract

Introduction and Background Detection and prediction of changes are necessary for maintenance of an ecosystem particularly in rapidly-changing and often unplanned regions in developing countries. Aims This study predicts the land use changes in catchment area around Bazangan Lake for the year of 2028 with the aim of investigating the development in Bazangan Lake ecosystem based on the observed degradation from 2002 to 2015. Methodology The classification of studied area was carried out based on five categories of irrigated agriculture, rainfed agriculture, rangeland, water zones and residential areas through TM, ETM and OLI sensors and utilization of independent component analysis (ICA) with an overall accuracy of 92.23% and kappa coefficient of 0.89% for the years of 1999, 2002 and 2015. Afterwards, the land use changes were predicted by a hybrid model of Markov chain and cellular automata. The overall accuracy and kappa coefficient were determined in IDRISI software by the help of ERRMAT Module to verify mode. Conclusion According to error matrix, the overall accuracy of performance was 71 percent and kappa coefficient 0.87 percent which proved Markov chain and cellular automaton (CA-Markov) for predicting the land use classes in upcoming 13 years. According to results, the continued current process of land use changes in this region will change Bazangan Lake area to 12.81 hectares, the irrigated agriculture land area to 495.91 hectares, rainfed agriculture land area to 5764.42 hectares, rangelands to 4592.15 hectares, and residential land area to 94.74 hectares in the next 13 years.