Estimation of Water Quality Parameters for Deepor Beel Using Landsat 8 Data

Monitoring the water quality of a wetland is very essential as it forms a major ecosystem in the environment. Although in situ wa-ter quality measurements are precise, high costs limit their applications. Remote sensing has demonstrated high potential as a cost-effective alter-native to traditional water quality monitoring techniques. In this study, the remotely sensed Lansat 8 dataset was used for estimating the water quality parameters of Deepor Beel, Assam. Using in situ data, regression models were created to establish the relation of water quality parameters—pH, total dissolved solids (TDS), and turbidity; with the bands of the Landsat data. For linear regression, R2 values for all the three parameters were more than 0.70. With R2 values over 0.90 for all parameters, the decision tree regression model produced more promising results. The analysis also showed that the pH value had a substantial correlation with the NIR band of the Landsat dataset. TDS, on the other hand, had a significant impact on the reflectance of the SWIR 1 and the green bands. For estimation of turbidity, the modified normalized difference water index (MNDWI) and red bands were found to be vital.