Fusion Learning-Driven High-Resolution Chromophoric Dissolved Organic Matter Estimation Using Remote Sensing Data: A Case Study in China

Chromophoric Dissolved Organic Matter (CDOM) plays a vital role in freshwater lake ecosystems, where remote sensing serves as a crucial tool for monitoring its spatiotemporal dynamics. However, accurately retrieving CDOM concentrations in small urban water bodies remains challenging due to the limited spatial and temporal resolution of current satellite data and estimation methods. To address this, we propose a highresolution CDOM estimation method for urban waters using a multi-source remote sensing fusion learning framework. A total of 110 water samples collected from water bodies in Wuhan in 2021 were used for model development and validation. This approach integrates Gaofen-2 (GF-2) satellite data to derive highresolution CDOM distribution, incorporates in-situ measurements simulated as GF-2 observations to enhance model accuracy, and employs Sentinel-2 imagery to provide spatial constraints. The results indicate that the proposed model achieves high estimation accuracy, with R2 of 0.892, RMSE of 0.101m-1, and MRE of 14.7% under leave-one-out cross validation. These findings confirm the efficacy of the fusion learning framework in accurately estimating high-resolution CDOM concentrations in complex urban aquatic environments.