Monitoring Water Quality Parameters in Trivandrum Coast, Kerala using Sentinel Images

Remote sensing has been played an important role in water quality monitoring. The Physiochemical parameters of water like Suspended sediments and Turbidity are an important determinants of water quality in coastal zones. Total suspended solids will increase the turbidity of water, which can affect the water quality. Turbidity is that just a cloudiness of fluid. Various parameters causing turbidity are phytoplankton, sediments from erosion, resuspended sediments, waste discharge, algae growth, urban runoff etc.Then,the naturally occurring events like storms, river flows, waves ,tides and activities like fishing, shipping and dredging operation also cause turbidity. This study aims to develop and implement regression models for estimating and mapping TSS and Turbidity concentrations from Sentinel 2A and 2B satellite images over the Trivandrum coast,Kerala.Concurrent complementary in-situ data for Turbidity were collected based on different latitude and longitude. By the collected turbidity values, TSS values can be determined. Using the secondary satellite data, giving latitude and longitude of field data and extracted reflectance values of each band and then regressed with their respective field data of TSS and Turbity.An algorithm was proposed to obtain the regression coefficient. This algorithm can be used to estimate the TSS and Turbidity concentration with a high correlation coefficient(R²=0.86) for Turbidity and (R²=0.82) for TSS. Finally, The map of Turbidity and TSS concentration was generated by using the proposed algorithm. This study found that TSS and Turbidity mapping can be conducted by using sentinel 2A and 2B images over the coast of Trivandrum ,Kerala. The predicted result from Sentinel 2A and 2B imagery exhibit that the predicted TSS and turbidity has a close agreement with the field data with the high R² values. Then, TSS and turbidity mapping has been done for validating the results with the in-situ data of turbidity (October 21, 2018 and November 10 and 15, 2018) and got 94% accuracy with mean difference error value of - 0.12903.