An Artificial Neural Network Approach using Remote Sensing Satellite data for Land Use Land Cover LULC Changes Classification in Dausa District Rajasthan India

Abstract

Abstract newlineThe present work is an attempt to the Land Use Land Cover (LULC) changes newlineclassification, monitoring, and spatiotemporal prediction using Artificial Neural newlineNetwork Multilayer Perceptron (ANN-MLP) and MLP-Markov Chain (MC) models. newlineDausa district (Dausa city and its surrounding area) of Rajasthan, India has been newlineselected for this study for several reasons including arid climatic setting being a newlinesensitive precursor to the climate change scenarios and the huge population pressure newlineexperienced by the area. After a thorough literature review it has been found that very newlinelimited studies have been studied on the selected study area for present work. The newlineMLP based supervised classification for two periods 2001 and 2018 have been newlineanalyzed using Landsat 7 Thematic Mapper (TM) and Landsat 8 Operational Land newlineImager (OLI) satellite images. The images were classified into six Land Use/Land newlineCover (LU/LC) categories viz. Built-up (Settlements), Cultivated Lands newline(Agricultural/Cropland), Water Body, Uncultivated/Fallow Lands, Barren Lands, and newlineForest/Vegetation Cover. The accuracy assessment for both classified images was newlineperformed using confusion matrix led Kappa Coefficient (K) technique. Reasonable newlineaccuracies, K=0.82 (2001) and K=0.91 (2018), have been achieved for datasets selected newlinefor both periods of time. The MLP-MC model based spatiotemporal LULC prediction newlinefor the year 2045, using the trends in the classified LULC results for the period 2001- newline2018, prophecies that the Built-up Land would increase to reach 76.10 (sq. km) newline(67.60% increase) in 2045 with the reference year 2001 whereas the increase in this newlineclass of LULC would only be 39.34% during the period 2018-2045.