作者：Huang, N., Li, P., Jiang, Y., Chen, L. & Wang, D.

The spatial heterogeneity of dissolved organic matter (DOM) is synergistically modulated by hydrodynamic and aquatic chemical factors. Although several studies have suggested the DOM transformation in the influence of water diversion projects (WDPs), their real connections remain unclear. In this work, we studied the relationship between DOM transformations in recipient lakes and WDPs for the Eastern Route of the South-to-North Water Diversion Projects (ER-SNWDPs) by integrating spectral techniques, global spatial autocorrelation analysis, and machine learning analysis. The results demonstrated that DOM characteristics in recipient lakes showed spatial heterogeneity during the non-water diversion period (NWDP), whereas WDPs transformed the characteristics of DOM into a homogeneous state. A linear fitting analysis revealed that during the water diversion period (WDP), only total fluorescence intensity was dependent on water conveyance distance, while aromaticity Index, fluorescence index, and humification index, which were originally dependent on geographic characteristics, turned to be independent because of WDPs. The 3-layer Deep Neural Network model coupled with Shapley Additive Explanations analysis identified WDPs operations, conveyance distance, and nitrogen concentration as the primary drivers of DOM homogenization. Hierarchical partitioning revealed that total nitrogen replaced geographical factors as the primary contributor (15.5%) for DOM transformation during the WDP, confirming that the nitrogen variability induced by WDPs was the key factor reducing DOM spatial heterogeneity. Our study revealed the impacts of WDPs on the spatial heterogeneity of DOM in recipient lakes, providing valuable insights for the ecological risk assessment and water quality management of ER-SNWDPs.

溶解有机质（DOM）的空间异质性受水动力与水化学因子的协同调控。尽管已有若干研究提示调水工程（WDPs）影响下的DOM转化，但二者间的真实联系仍不明确。研究综合光谱技术、全局空间自相关分析与机器学习方法，考察了南水北调东线工程（ER-SNWDPs）受水湖泊中DOM转化与调水工程之间的关系。结果表明：非调水期（NWDP），受水湖泊的DOM特征呈现空间异质性；而调水期（WDP），调水工程使DOM特征趋于均质化。线性拟合分析显示：调水期间，仅总荧光强度仍依赖于输水距离，而芳香性指数、荧光指数与腐殖化指数——原本依赖地理特征——则因调水而变为独立。三层深度神经网络模型结合SHAP（Shapley加性解释）分析识别出调水工程运行、输水距离与氮浓度是DOM均质化的主要驱动因子。层级方差分解进一步表明：调水期间，总氮取代地理因子成为DOM转化的首要贡献者（贡献率15.5%），证实调水工程诱导的氮盐变率是降低DOM空间异质性的关键因子。本研究揭示了调水工程对受水湖泊DOM空间异质性的影响，为东线工程的生态风险评估与水质管理提供了重要参考。

（来源：Journal of Hydrology 2026  DOI: 10.1016/j.jhydrol.2026.135649）