A Hybrid Remote Sensing and Statistical Modelling Framework for River Water-Quality Forecasting During Extreme Hydrologic Events

Extreme hydrologic events such as floods and intense rainfall significantly disrupt river water quality by accelerating sediment transport, nutrient loading, and pollutant dispersion, creating major challenges for environmental monitoring and forecasting. This study presents a hybrid remote sensing and statistical modelling framework designed to improve river water-quality prediction during extreme hydrologic disturbances. Multispectral satellite data from Sentinel2 MSI and Landsat-8 OLI were processed to derive water-quality indicators including turbidity, chlorophyll-a, and Total Suspended Matter (TSM), while ground-based hydrologic observations and meteorological datasets provided complementary temporal information. Statistical forecasting models, including Multiple Linear Regression (MLR), ARIMA/SARIMA, and Generalized Additive Models (GAM), were integrated within a data fusion architecture to combine spatial and temporal predictors. Model performance was evaluated using RMSE, MAE, coefficient of determination, and Nash–Sutcliffe Efficiency metrics under event-based validation conditions. Results demonstrate that the hybrid framework substantially outperforms standalone remote sensing and statistical approaches, reducing prediction error by more than 25% and improving forecasting reliability during flood peaks. The system successfully detected pollution pulses, captured spatial heterogeneity across the river basin, and maintained stable prediction accuracy during pre-event, peak, and post-event phases. Findings highlight the value of satellite observations for monitoring inaccessible regions and confirm that integrated modelling enhances early-warning capability for water-quality degradation. The study provides a scalable methodological foundation for operational environmental monitoring systems, supporting water utilities, environmental protection agencies, and disaster-response planning in river basins increasingly affected by climate-driven hydrologic extremes.