LULC Mapping For Identifying Change In Agricultural Land Cover using Satellite Images

Abstract

Remotely sensed images or satellite images has opened a new window for researchers to extract and analyze various land cover classes spread over the earth surface. This study proposes an enhanced Machine Learning (ML) based change detection method for the analysis of change in agricultural land cover along with other land cover classes such as, water bodies, built-up area and barren land. Machine learning algorithm employs classification of multi-spectral remote sensing data into different classes created during training of model. The study has been carried out by analyzing Geographical features of Sangli city and its surrounding area by using LISS III multispectral satellite images of year 2014 and year 2019, collected from ISRO Bhuvan web-portal. Satellite images are processed using GIS tools and spatial modelling has been implemented to automate the manual process of classification using ML algorithm. Further, the extracted classes of the classified images have been compared to analyze change in the land cover classes. After analysis it has been observed that vegetation which includes most of the agricultural land has been decreased by 13.40% and Barren land has been decreased by 25.36%. However, built up has been increased by 23.99% due to growing urbanization.