Potential of Open Source Remote Sensing Data for Improved Spatiotemporal Monitoring of Inland Water Quality in India: Case Study of Gujarat

Authors:

Abstract
Unison of in situ data with satellite remote sensing data has been widely used for water body quality monitoring; however, the developed synergized model is unique and thus needs to be recalibrated and revalidated before applying to other water bodies. In some Indian locations, water quality monitoring is conducted only once a year due to associated costs and time. To aid such instances, in this study, stepwise linear regression models were developed using in situ (annual) and Landsat 7 (biweekly) remote sensing data and validated for two inland water bodies (Sursagar and Nalsarovar lakes) in Gujarat state for dissolved oxygen, biochemical oxygen demand, electrical conductivity, pH, and nitrate. Results indicated that all models showed good to excellent performance metrics based on an r value (p < 0.01) ranging from 0.86 to 0.98 and 0.72 to 0.99 for Sursagar and Nalsarovar lakes, respectively. All models had root mean square errors less than 0.5, and residual predictive deviations greater than 2, which depicted good predictability. The models were able to increase the water quality assessment from annual resolution to biweekly resolution and provided insights on the dynamics of water quality parameters, improved understanding on key drivers for the change, and identified peak pollution leading to unfit conditions for domestic or agricultural consumption.