Spatial and Seasonal Change Detection in Vegetation Cover Using Time-Series Landsat Satellite Images and Machine Learning Methods

The present study used time-series Landsat-8 and 9 satellite datasets of June to February 2016–2017 and 2021–2022 to classify and detect the changes in vegetation covers. The studied Akole region of Ahmednagar district of Maharashtra, India, is vulnerable to drought conditions in a diverse environment. The spectral features based on the Normalized Difference Vegetation Index (NDVI) were calculated. Machine learning algorithms such as k-means clustering and Iterative Self-Organizing Data Analysis (ISODATA) clustering have also been applied to time-series NDVI images to classify the vegetation cover and detect the changes in vegetation. Furthermore, to identify different drought clusters. The results of the NDVI values ranged from 0 0.25 to 0.99 and 0.31 to 0.75 for 2016–2017 and 2021–2022, respectively. The classification results show that most of the areas were occupied by healthy vegetation in 2021–2022. In 2016–2017, the vegetations were less due to low rainfall. Regional planners and decision makers can use the present study to identify vegetation, assess and monitor drought severity, and predict future scenarios.