An Innovative Approach to Delineate and Differentiate Clear and Turbid Water Ponds in Indian Sundarban Area Using Sentinel-2 MSI Data

Abstract

Pond water is the main source of irrigation in the Sundarbans area of West Bengal. The serious issues faced in agricultural lands and ponds are due to salinity. As salinity cannot be detected on high scales through satellite images, in this study the remotely sensed turbidity has been compared and studied with laboratory-measured salinity. The high-resolution Sentinel-2 MSI data with optimum bands for water quality detection have been used in this study. The small agricultural ponds in the study area were not detected by the conventionally used water indices, as the ponds were surrounded by homestead vegetation. Hence, a new index called Pond Probability Spectral Index (PPSI) was formulated and used in the decision tree algorithm to delineate all the ponds. Turbidity indices were used over the extracted ponds to classify them into three classes, clear, turbid and highly turbid. From the field accuracy assessment, one of the turbidity indices outperformed in segregating the ponds with an overall accuracy of 83.3%. It was also found that salinity was inversely proportional to turbidity only when the turbidity value was greater than 100 NTU. The total estimated pond area (average of 3 seasons) in the Sundarban region was 471.66 km2.