| description abstract | Currently, land use and land cover change (LULCC) dynamics are critical in the sustainability of global environmental change. In this study, we performed a comprehensive intercomparison of different hybrid models, employing several performance metrics. This study aims to determine the best modeling approach for predicting future land use and land cover changes. The methodology is demonstrated in the Rengali catchment of the Brahmani basin in India, using Landsat imagery from 1990, 2005, and 2020. The analysis of temporal mapping from 1990 to 2020 showed a substantial decrease in the forest area (from 40.96% to 33.78%) and a significant increase in both agricultural land (31.39% to 41.77%) and built-up area (3.63% to 16.96%). Three hybrid models, namely artificial neural network-cellular automata (ANN-CA), multilayer perceptron-cellular automata-Markov chain (MLP-CA-MC), and logistic regression-cellular automata-Markov chain (LR-CA-MC) were assessed to determine the effectiveness in predicting future land use and land cover (LULC) changes in the basin. Among the three models, the MLP-CA-MC model showed superior performance following the validation phase. Later, the MLP-CA-MC model was employed to estimate the future LULC projections for 2050 and 2075 by considering the nonstationary relationship between selected driving variables and LULC. The model forecasts that by 2075, the forest area, water bodies, and barren land may decrease by 27.20%, 3.82%, and 2.38%, respectively, compared to the observed LULC in 2020. On the other hand, agricultural land and built-up areas are expected to increase by 45.21% and 21.39%, respectively. The results of this study can provide land use planners, environmentalists, and policymakers with essential information to formulate effective management strategies and laws that will serve the public better. | |
