Application of remote sensing and regression learner in assessing the water quality of Tri An Reservoir, Vietnam

Abstract

Monitoring water quality in large reservoirs is essential for sustainable freshwater management. This study evaluates the applicability of Landsat satellite imagery, combined with spectral reflectance analysis, to assess the Vietnam Water Quality Index (VN-WQI) for Tri An Reservoir in 2024. Spectral reflectance values from Bands 1-9 were extracted and correlated with the VN-WQI. Regression analysis revealed that Band 1 and Band 5 were the most significant predictors of VN-WQI, with strong statistical significance (p < 0.0001). The optimal regression model achieved high accuracy on the training dataset, with R2 = 0.9774, Adjusted R2 = 0.9768, and RMSE = 12.82. Validation results showed consistent reliability, with RMSE = 13.99 and MAPE = 13.47%. The derived regression equation, VN-WQI = 711.748 × Band1 – 393.908 × Band5, demonstrates the strong ability of Landsat spectral data to estimate water quality conditions in the reservoir. The findings confirm that remote sensing, when integrated with statistical regression, provides an effective, low-cost approach for large-scale VN-WQI assessment and supports improved monitoring of water resources in Tri An Reservoir.