Time Series Analysis of Water Feature Extraction using Water Index Techniques from Landsat Remote Sensing Images

Recently, the remote sensing data is widely used for the extraction of water body from the satellite images. The accuracy assessment of the extracted water features from the satellite images is highly correlated with the real time data. Spatiotemporal changes in nagarjunasagar reservoir, located in India in a period of 2014 to 2019 time series and analysis using multi temporal Landsat-8 (OLI) images. Unsupervised classification (Isodata) and spectral water indexing methods, including NDVI, NDWI, MNDWI and AWEI were evaluated for surface water body extraction and change detection. The overall accuracy and kappa coefficients were evaluated for water indexing methods. The statistical parameters of the accuracy results show that AWEI achieved 96.26% overall accuracy, 0.94 kappa coefficient and MNDWI achieved 96.94% overall accuracy, 0.95 kappa coefficient. The AWEI and MNDWI water indexes performed better results as compared to other water indexing methods.