Answers are grounded in this paper and cite their sources.
Urban lakes exhibit complex optical properties, which challenge the accurate estimation of fluorescent dissolved organic matter (fDOM), a key ecological indicator of water quality. This study targets Wuhan's urban lakes, integrating inherent optical property (IOP) features and interpretable machine learning to quantitatively analyze the spatiotemporal dynamics and multi-source drivers of fDOM. The results indicate that the eXtreme Gradient Boosting (XGBoost) model integrating IOP features achieves robust predictive performance (R 2 = 0.742, RMSE = 1.238 RFU). The SHapley Additive exPlanations (SHAP) framework demonstrates that the model's feature utilization aligns with established bio-optical principles. Spatiotemporal analysis reveals a concentration gradient decreasing from the urban core to the rural outskirts, distinguishing persistent pollution hotspots in hydrologically restricted lakes from sporadic hotspots in open waters. Covariance-Based Structural Equation Modeling (CB-SEM) quantifies the structural associations, identifying built-up area as a primary macroscopic driving factor (standardized total effect = 0.352). This factor exerts a direct negative effect on fDOM (direct effect = −0.168) while showing a significant positive indirect pathway mediated by surface chlorophyll-a concentration proxied by NDCI (indirect effect = 0.520). Furthermore, precipitation functions as a dual suppressive constraint, operating through pathways that theoretically correspond to volumetric dilution and habitat disturbance. By elucidating these macroscopic spatiotemporal patterns and structural pathways, this study provides a quantitative scientific basis for the differentiated management of urban aquatic ecosystems.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: