Efficient monitoring of chlorophyll-a concentration in urban water bodies based on UAV multispectral images and ensemble machine and deep learning method

With the accelerating process of urbanization, water eutrophication has become an increasingly severe issue. As a key indicator of eutrophication, chlorophyll-a (Chla) concentration directly reflects the ecological health of water bodies, making efficient and accurate monitoring a critical challenge for urban ecological management and sustainable development. In this study, high-resolution unmanned aerial vehicle (UAV) multispectral imagery and derived spectral features were integrated into a newly developed EMD (ensemble machine and deep learning) method to achieve high-precision inversion of Chla concentrations. The EMD method outperformed individual models, demonstrating both high accuracy and strong stability (average R² = 0.797, average RMSE = 18.96 mg/m³). Land-use analysis was further conducted to identify major drivers and mechanisms of urban water eutrophication. Significant differences were observed among land-use types: rivers adjacent to industrial areas exhibited substantially higher mean Chla concentrations than those influenced by agricultural land or mixed green spaces and residential zones, which showed comparatively similar levels. Industrial land expansion was identified as the dominant driver of elevated Chla concentrations, while well-planned green spaces and residential layouts effectively mitigated eutrophication risks. The UAV-based remote sensing and EMD method developed in this study provides an innovative tool for precise Chla monitoring in urban water bodies. By incorporating differentiated land-use characteristics, the findings offer a scientific basis for developing location-specific strategies to control eutrophication and support sustainable urban environmental governance.