Temporal Pattern Change in Water and Vegetation Using Geospatial Data and Thematic Indices

This paper conducts a per-pixel change detection analysis of Bansagar Lake's water and vegetation, situated in the Ganges Basin, Madhya Pradesh, India (coordinates: 24°05'59.8“N, 81°04'59.5“E). The study employs the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Water Index (NDWI) as remote sensing indices to detect changes in the area of interest. Data set was obtained from IRS LISS III satellite, obtained from the NRSC Bhuvan portal, were used for this analysis over a four-year period from 2014 to 2017. Our approach shows strong alignment with ground truth which has been generated using Sentinel Application Platform (SNAP) tool, with NDWI relative error between 0.072% and 4.918%, and NDVI error ranging from 2.895% to 9.339%. Higher NDVI errors are due to distinct sparse and dense vegetation classes. The temporal analysis of multispectral images from 2014 to 2017 also reveals a decline in vegetation and water bodies, with NDWI pixel values decreasing by −2.822% and NDVI pixel values decreasing by −3.5%.