Prediction of spatio-temporal (2030 and 2050) land-use and land-cover changes in Koch Bihar urban agglomeration (West Bengal), India, using artificial neural network-based Markov chain model

The land is one of the most exigent natural resources and is dynamic in nature. Land use and land cover refer to the categorization and classification of human activities and natural elements on the landscape within a specific time frame. Thus, land cover is the physical material on the Earth's surface, whereas land use describes how people utilize the land. The change in land use and land cover (LULC) is a significant issue from a global perspective. This article aims to evaluate changes in the LULC in the past (1990-2020) and to predict the future LULC (2030 and 2050) in Koch Bihar urban agglomeration (West Bengal). The present study uses Landsat 5 (TM) and Landsat 8 (OLI/TIRS) remote sensing data, ASTER DEM, open street map, and the Census of India data. The Maximum Likelihood Classifier (MLC) algorithm is used for the supervised classification of land, and an Artificial Neural Network (ANN)-based Cellular Automata-Markov Chain (CA-Markov) model has been used to predict the future LULC pattern in 2030 and 2050. The overall accuracy of the classified images (1990, 2000, 2010, and 2020) is 90%, 92%, 93%, and 91%, respectively. The Kappa coefficient is more than 0.80 (0.873 (1990), 0.899 (2000), 0.911 (2010), and 0.887 (2020). The result shows that the amount of agricultural land would decrease to a great extent, from 50.89 km(2) in 1990 to 22.30 km(2) in 2050. At the same time, the built-up area would increase significantly from 5.91 km(2) in 1990 to 32.91 km(2) in 2050. The findings of the present study guide planners and resource managers in designing a roadmap for long-term sustainable land-use and land-cover management in the Koch Bihar urban agglomeration.