Land/use land /cover dynamics and future scenario of Mayurakshi river basin by random forest and CA–Markov model

Abstract

The study was focused on analyzing the land use and land cover status, change patterns, and future scenarios in the Mayurakshi basin in Jharkhand and West Bengal state of eastern India. The dataset collected for image classification included Landsat 5 (TM) (1991–2008) and Landsat 8 (OLI) (2020). Various sequential preprocessing steps such as atmospheric correction, image enhancement, mosaicking, masking, and clipping were performed using QGIS 3.16 and ArcGIS 10.8 software. The land use and land cover classes found in the study area were water, vegetation, bare land, agriculture, and built-up, and classification was executed by using the Random Forest machine learning algorithm. The accuracy of the classified land use and land cover was validated and accepted with Kappa agreements of 0.89, 0.85, and 0.88 for the years 1991, 2005, and 2020, respectively. Throughout the study period, agriculture emerged as the dominant land use class, followed by vegetation and bare land. The area under the land use and land cover categories of water, vegetation, and bare land continuously decreased between the years 1991–2005 and 2005–2020, while agriculture and built-up areas recorded an increase of 4.49%, 0.76%, 17.81%, and 2.04%, respectively. To project future land use and land cover status, the popular Cellular Automata Markov Chain Model was employed. The projected results indicate that agriculture will remain the dominant land cover with a share of 70.24%, followed by vegetation at 17.72% and built-up areas at 5.09%. However, a marginal decline is expected in both the agriculture and built-up classes.