Integrating Satellite Remote Sensing and Field Measurements for Assessing Nitrogen and Phosphorus Dynamics in Tropical Freshwater Ecosystems of Thailand

Abstract

Eutrophication increasingly threatens tropical freshwater systems, where nutrient enrichment drives harmful algal blooms and rapid water-quality decline. This study presents a validated satellite-based approach for retrieving total nitrogen (TN) and total phosphorus (TP) in Thai inland waters using Sentinel-2 imagery. A stratified sampling campaign collected 264 water samples from 50 lentic water bodies during April–May 2024. Results showed that key environmental predictors were identified through correlation analysis and integrated into multiple linear regression (MLR) and polynomial regression (PO) algorithms, with performance evaluated using 10-fold cross-validation. PO consistently outperformed MLR for both nutrients. For TN, PO achieved higher calibration accuracy (R2 = 0.63; RMSE = 4.74 mg/L) than MLR (R2 = 0.47; RMSE = 5.73 mg/L) and maintained strong validation performance (R2 = 0.76). For TP, PO likewise yielded superior cross-validation accuracy (R2 = 0.69; RMSE = 0.07 mg/L; IoA = 0.89) compared with MLR (R2 = 0.38; RMSE = 0.10 mg/L; IoA = 0.70). Spatial distributions of TN and TP derived from the imagery reinforced these findings. The PO-derived TN maps effectively captured nutrient hotspots associated with agricultural runoff, whereas the PO-based TP maps produced more stable and consistent spatial patterns. The findings demonstrate that Sentinel-2 can reliably retrieve TN and TP in tropical waters and offer a scalable pathway for improving nutrient surveillance and supporting science-based freshwater management in regions experiencing accelerating eutrophication.