Depiction of Water Characteristics from Multispectral Imagery Using Image Processing

The evolution of remote sensing techniques enforces to acquire veracious and continual updates of surface water characteristic on Earth. To discernment water bodies from multispectral imagery one of the primarily employed techniques is Indices method with general threshold value. This general threshold value is fixed for most of the cases, but when minuscule water bodies surrounded by plenty of dense forests, clouds, shadows and constructed area, they convey results with some error. This study assess the accomplishment of the primarily employed water indices techniques NDWI35, NDWI36, NDWI37 (Normalized Difference Water Index) and NWI (New Water Index) in Ayodhya region of Uttar Pradesh, India. The assessment result obtained from confusion matrix was not able to delineate surface water bodies with superior accuracy. To improve the accuracy in delineation of water bodies, enhanced water indices techniques were proposed with prime threshold and clustering algorithm. The results obtained from enhanced water indices techniques shows consequential enrichment in the overall accuracy and kappa coefficient. All the multispectral imagery data accumulated by Landsat-8 (OLI) satellite on 22nd March 2019 was used for obtaining results.