A novel algal bloom risk assessment framework by integrating environmental factors based on explainable machine learning

In recent years, the algal blooms have intensified, posing mounting threats to aquatic ecosystems and water security. However, most previous studies merely detected algal blooms according to the characteristics of the water body at the time of algal bloom occurrence, overlooking the influence of environmental factors on algae proliferation. This study proposes a novel algal bloom risk assessment framework that integrates explainable machine learning with multivariate environmental analysis. Specifically, the Shapley Additive Explanations (SHAP) effect values were used to separately explore the relationship between chlorophyll a (Chla) and six factors, namely the total phosphorus (TP), total nitrogen (TN), TN: TP ratio (RNP), dissolved oxygen (DO), temperature, and precipitation, across riverine and lacustrine ecosystems. Results identified TP and temperature as dominant regulators, accounting for the first two in lakes and the second and third positions in rivers. The thermal effect varies between different ecosystems: Chla decreases after reaching a peak in lakes, while Chla increases linearly with temperature in rivers. In addition, DO played an important role in rivers. The Chla concentration was estimated using Random Forest and thresholds for bloom identification were adjusted (25 μg/L for lakes and 40 μg/L for rivers), reflecting hydrodynamic and optical disparities. The risk framework was applied to the Qiantang River Basin (2020−2022), and results showed low annual risk (mean Algal Bloom Risk Index <0.5) but identified spring susceptibility related to nutrient resuspension and thermal stratification. By quantifying the impact of environmental factors on algal blooms, this study improves algal bloom risk assessment in rivers and lakes, which advances proactive bloom management in mixed river-lake basins under intensifying anthropogenic and climatic pressures.