Simulating the Future Land Use and Land Cover by Implementing Machine Learning Knowledge on the South-Western Zone of Rajshahi District, Bangladesh

Abstract

Urbanization causes major urban pressures and changes in habitat, ecology, and environmental diversity. Particularly in developing countries, they are often characterized with strong urban sprawl. This research aimed to detect the temporal LULC change with the driving forces of LULC and forecast the future LULC via simulation in Rajshahi, Bangladesh. The LULC was classified through supervised classification using Erdas Imagine. To predict future land-use/ land cover change, the Land Change Modeler (LCM) of Terrset application was used with the implementation of CA and Markov chain analysis. “Built-up area” and “bare land” increased whereas “water bodies” and “vegetation” landforms decreased from 1991 to 2011. Although, “builtup area” and “vegetation” increased as “water bodies” and “bare land” decreased from 2001 to 2021. The driving forces of the LULC change were identified through correlation. The conventional driving forces were found to be ‘population,’ ‘literacy rate,’ ‘population density,’ ‘household income’ and ‘average household size’. When the outcome of the coefficient is positive, the driving force works proportionally. Average accuracy of the prediction model is approximately 85% and the predicted image was validated with the classified image of 2021 and the average accuracy was 92.03%. LULC for the years 2001, 2011 and 2021 along with the driving forces served as the basis to predict land-use/land cover for 2031, and 2041. This research helps to analyse urban planning policies and environmental management, and to take proper measure for proper control of LULC change

Journal of Engineering Science 13(1), 2022, 61-67