Assessing the feasibility of integrating Sentinel-2 Chlorophyll-a data into a water quality monitoring system for early warning purposes in the Thach Han River

Abstract

Water-quality models are important tools for monitoring, assessing, and supporting the management of surface-water quality. However, the reliability of 3 model simulations depends partly on the completeness of input information, particularly variables associated with biological processes in aquatic systems. This study investigates the feasibility of integrating Chlorophyll-a (Chl-a) information derived from Sentinel-2 imagery into a water-quality model for the Thach Han River system, Vietnam. Chl-a was estimated using spectral reflectance from the B2 and B3 bands of Sentinel-2 and validated against field-measured Chl-a data before being incorporated as additional information into the water-quality simulation. Two scenarios were evaluated using an inherited water-quality modeling framework: Scenario 1 (S1) representing the baseline model configuration and Scenario 2 (S2) incorporating Sentinel-2-derived Chl-a. Model performance was evaluated against observed total nitrogen (TN) and total phosphorus (TP) using RMSE, MAE, ME, and R². The results demonstrate a substantial improvement in TN simulation following Chl-a integration, with RMSE decreasing from 0.1263 to 0.0537, MAE from 0.1191 to 0.0487, and R² increasing from 0.665 to 0.977. For TP, RMSE and MAE slightly increased from 0.00727 to 0.00773 and from 0.00443 to 0.00591, respectively, although R² increased from 0.032 to 0.207. These results indicate that integrating Sentinel-2-derived Chl-a can provide additional biological information and substantially improve TN simulation, whereas its effect on TP remains inconclusive and requires further validation. The simulation results were applied on a pilot basis in a WebGIS-based water-quality monitoring and earlywarning system.