Enhanced assessment of chlorophyll-a and total nitrogen dynamics using unmanned aerial vehicle-based model and hyperspectral imagery in coastal wetland water

The Liao River Delta is at risk of eutrophication and water pollution due to oil extraction and aquaculture. Rapid estimation of chlorophyll-a (Chla) and total nitrogen (TN) is essential for effective water quality management. Hyperspectral remote sensing facilitates dynamic monitoring and spatiotemporal mapping of water quality, offering significant advantages over traditional methods. This study collected 102 water samples and airborne hyperspectral data (350-1000 nm) from the Liao River. Gaofen-5B (GF-5B) satellite images from 2022 to 2024, representing various months, were also utilized to map Chla and TN distributions. Key findings include: 1) Based on spectral absorption features from airborne hyperspectral data, new spectral indices for Chla (TBI Chla [R(682 nm) - 1 − R(691 nm) - 1 ]×R(763 nm) ) and TN (TBI TN [R(721 nm) -1 −R(674 nm) -1 ]×R(969 nm) ) were developed; 2) The direct model provided the most accurate prediction for Chla (R 2 = 0.88, RMSE = 0.0037 mg/L, MAE = 0.0029 mg/L), while an auxiliary model incorporating hue angle enhanced TN prediction (R 2 = 0.86, RMSE = 0.1727 mg/L, MAE = 0.1331 mg/L); 3) the best-performing unmanned aerial vehicle (UAV)-based models, optimized for atmospheric correction, were effectively applied to multi-temporal GF-5B images, enabling large-scale mapping of Chla and TN distribution across the watershed; 4) Intensive human activities and oil extraction were found to elevate urban Chla levels, whereas high TN concentrations in autumn were linked to the aquaculture fattening season. The integration of UAV and satellite hyperspectral data significantly improved the spatial representation of Chla and TN in the Liao River, capitalizing on the complementary strengths of high-resolution UAV imagery and broad-scale satellite coverage. • A pioneering UAV-derived spectral index for Chla and TN retrieval was developed. • An auxiliary inversion framework based on hue-angle for TN prediction was integrated. • The cross-platform transferability from UAV to satellite for dynamic water quality monitoring was demonstrated. • The mechanistic link between human activities and spatiotemporal trophic state changes was determined.