River algal bloom is a global ecological and environmental problem, which is characterized by complex causes, large impact range and long duration. Numerous studies have shown that river algal blooms are mainly influenced by climatic, nutrient, hydrological and hydrodynamic conditions, and the interference of human activities. In recent years, there has been a serious problem of algal blooms in the lower reaches of the Han River (HR), posing a huge threat to the ecosystem and environment of the HR Basin. Particularly, with the operation of the South to North Water Diversion Project (SNWDP) since 2014, the mechanism of algal blooms in the HR becomes more complicated, making the prevention and control of algal blooms more difficult. Here we combine the Convergent Cross Mapping (CCM) and machine learning to identify the causes and predict the occurrence of algal blooms in the HR. We also examined whether the occurrence of algal blooms would be greatly impacted by the operation of SNWDP.
The CCM results revealed that the water temperature in the HR and the water level in the YR had causal effects on algae density. Both random forest and gradient boosting machine methods accurately predicted the occurrence of algal blooms by using the antecedent 10-day period data, indicating that the models had the ability to predict algal blooms in advance. In addition, the variable importance analysis showed that the water levels in the HR and YR dominated the occurrence of algal blooms before and after the operation of the project, while the water temperature in the HR replaced the water level variation in the HR as one of the dominant factors after the operation of the project. This study provides scientific guidance for evaluating the impact of SNWDP on the algal blooms in the HR, and it is of great significance for the prediction and control of algal blooms in the HR and the ecological restoration of the HR Basin.