Land use and land cover classification for change detection studies using convolutional neural network

Abstract

Efficient land use land cover (LULC) classification is crucial for environmental monitoring, urban planning, and resource management. This study investigates LULC changes in Nanjangud taluk, Mysuru district, Karnataka, India, using remote sensing (RS) and geographic information systems (GIS). This paper mainly focuses on the classification and change detection analysis of LULC in 2010 and 2020 using linear imaging self-scanning sensor-III (LISS-III) remote sensing images. Traditional methods for LULC classification involve manual interpretation of satellite images, which provides lower accuracy. Therefore, this paper proposed the Convolutional Neural Network (CNN)-based deep learning method for LULC classification. The main objective of the research work is to perform an efficient LULC classification for the change detection study of the Nanjagud taluk using the classified maps of the years 2010 and 2020. The experimental results indicate that the proposed classification method is outperformed, with an overall accuracy of 94.08% for the 2010 data and 95.30% for the 2020 data. Further, change detection analysis has been carried out using classified maps and the results show that built-up areas increased by 8.34 sq. km (0.83%), agricultural land expanded by 2.21 sq. km (0.23%), and water bodies grew by 3.31 sq. km (0.35%). Conversely, forest cover declined by 1.49 sq. km (0.15%), and other land uses reduced by 11.93 sq. km (1.22%) over the decade.