Landsat-Derived NMSI Index for Estimating Total Suspended Solids Using Linear Regression Models

Remote sensing provides an efficient and cost-effective approach for monitoring water quality at the watershed scale. This study assessed the potential of Landsat-derived indices for estimating total suspended solids (TSS) in the Day River Basin, Vietnam. Field-based TSS data collected during the dry (April 2025) and rainy (June 2025) seasons were integrated with the Normalized Multi-band Suspended Index (NMSI) extracted from Landsat imagery. Linear regression models were developed to link in-situ TSS measurements with NMSI values. It was found that the results demonstrated strong predictive performance. The dry-season model achieved R² = 0.801, RMSE = 2.277 mg/L, and a highly significant regression slope (p < 0.001). The rainy-season model performed even better, with R² = 0.8447, adjusted R² = 0.8377, RMSE = 1.889 mg/L, and MAE = 1.121 mg/L. Spatial maps of predicted TSS revealed clear seasonal contrasts and downstream amplification of suspended sediment loads, with higher concentrations observed in the middle and lower reaches during the rainy season. These findings confirm that NMSI is a reliable spectral proxy for suspended sediment, and that Landsat imagery combined with regression analysis can provide accurate, spatially explicit estimates of TSS. This approach supports the development of basin-scale monitoring frameworks and contributes to sustainable sediment and water-quality management in the Day River Basin. Keywords: Total suspended solids, Linear Regression, remotely sensed images, Day River Basin, Vietnam.