Chapter 2 - Assessment of land use land cover change detection in multitemporal satellite images using machine learning algorithms

Abstract

The world is encountering fast land use land cover (LULC) changes because of several socioeconomic activities and natural phenomena. Detecting and monitoring changes on the Earth’s surface are vital to predict and avoid land degradation and improve the land rehabilitation process. The objective of this work is to gain a quantitative analysis of LULC changes in Madurai (3742km2), Tamil Nadu, India, using the normalized difference vegetation index (NDVI) and maximum likelihood classification (MLC). Multitemporal satellite data of years 2013, 2015, 2017, and 2019 were used to decide the variations in the land cover, including agriculture, vegetation dynamics, urban, and habitat disturbance of the study area. Accuracy assessment was performed by gathering ground truth points of 2019. The results indicated that the LULC classes were accurately recognized in MLC with an accuracy of 91% with a Kappa value of 0.89 and NDVI with an accuracy of 92% with a Kappa value of 0.90. Thus, this technique would be convenient to plan and execute critical management decisions to preserve biodiversity.