Identification of driving factors for chlorophyll‐a in multi‐stable shallow lakes of China employing machine learning methods

Abstract

AbstractDue to the rapid increase in urbanization and anthropogenic activity in the lake catchment, nutrients like phosphorus and nitrogen are abundant in urban sewage, leading to lake eutrophication and promoting algal blooms. Chlorophyll‐a (Chl‐a) is an important indicator of algal blooms; thus, predicting concentration is vital for lake restoration and management. Literature reveals that most of the previous studies of Chl‐a prediction required multiple variables and considerable computation. Thus, it is crucial to scientifically identify the key drivers of Chl‐a in Taihu Lake for the formulation of effective environmental management measures. In this study, a long‐term dataset (1991–2015) and Daniel trend test method were employed to analyse the trend of meteorological and water quality indicators in Taihu Lake in China. The redundancy of variables was eliminated by using the rank correlation method, combined with prior knowledge that provides a solid foundation for the analysis of the driving factors of Chl‐a. Four machine learning models, named Extreme Gradient Boosting (XGBoost), random forest (RF), support vector regression (SVR) and gradient boosting decision tree (GBDT), were employed to predict the concentration of Chl‐a in different regions of Taihu Lake. The result shows that the XGBoost model outperformed the other three models in terms of R2 (>0.90) and root mean square error (RMSE) (3.69–8.40). Finally, based on the predictions of the concentration of Chl‐a using the XGBoost model, the SHapley Additive exPlanation (SHAP) framework was utilized to analyse the key factors influencing Chl‐a concentration. Finding of this study suggests that the XGBoost model performed best and could be applied to the overall stretches of Taihu Lake and other lakes in China; model prediction will be helpful in providing a zoning decision‐making approach for the rational protection of the lake environment and the prevention of algal bloom hazards.