An analysis of change detection in land use land cover area of remotely sensed data using supervised classifier

Abstract

In the present work, change detection in land use and land cover (LULC) area of Chikkamagaluru district were assessed using remote sensing data and supervised classifier. Chikkamagaluru district is known for the green cover; therefore an analysis of the land use land cover of the district is the main objective of this work. The change detection of an entire Chikkamagaluru district has been carried out for the period between 2017 and 2021 by using Sentinel-2 multispectral remote sensing data. Supervised classification-based support vector machines (SVM) have been applied to assess the LULC of the study area. An experimental result shows the positive changes in vegetation cover, water bodies, and negative changes observed in bare ground and rangeland. Overall classification accuracy of the SVM was estimated to be 86.30% for 2017 and 85.36% for 2021. The performance of SVM is also compared with the other supervised classifiers such as neural networks, maximum likelihood classifier (MLC), minimum-distance-to-means, and Mahalanobis distance. The comparison results show that SVMs provide better classification results as compared to other supervised classifiers.