Comparative analysis of land use changes modeling based-on new hybrid models and CA-Markov in the Urmia lake basin

Abstract

Land use change is one of the most significant environmental issues in many countries and is the main reason for ecosystem changes. This issue is one of the most important environmental crises in Iran due to the lack of planning and inappropriate human use of the natural environment. Many parts of Iran, especially the Urmia Lake basin, are involved in challenging land use changes. The trend of land use changes extracted from Landsat images in the Urmia lake basin with overall accuracy and Kappa coefficient of more than 85 % and 80 % respectively shows that during the 32-year period (1989–2021) agricultural lands, built-up lands and barren lands respectively 3330.83 km2, 466.86 km2 and 5325.8 km2 have increased in area. On the other hand, the areas of vegetation and water bodies decreased by 6625.79 km2 and 2497.7 km2, respectively. The current trend of land use changes in Urmia basin based on all models of Markov chain-Multi Criteria Evolution-Cellular Automata (MC-MCE-CA), Markov Chain-Multilayer Perceptron-Cellular Automata (MC-MLP-CA), Markov Chain-Sim Weight-Cellular Automata (MC-SW-CA), Markov Chain–Logistic Regression-Cellular Automata (MC-LR-CA) indicate the continuation of changes until 2037. Meanwhile, based on the most accurate models such as MC-MLP-CA and MC-LR-CA models, agricultural lands will increase by 1436.60 km2 and 1258.33 km2, built-up lands by 75.23 km2 and 65.52 km2, and barren lands by 808.94 km2 and 856.90 km2 respectively. Also, the vegetation lands will be reduced by 1890.64 km2 and 1755.31 km2 and water bodies by 430.13 km2 and 425.44 km2. In general, the results showed that MC-MLP-CA, MC-LR-CA, MC-SW-CA, MC-MCE-CA models with Overall Kappa 89.89, 87.09, 84.52 % and 84.84 and Percent of Correctness 93.65, 93.09, 91.52 and 91.77 respectively, with higher accuracy than the CA-Markov model (Overall Kappa = 81.49 % and Percent of Correctness = 87.49 %) are more effective for land use changes modeling.