The XGBoost and the SVM-based prediction models for bioretention cell decontamination effect

Surface water flooding is in a crisis for most large cities of the world. Sponge city is an innovative idea conceptualized designed to cope with urban surface water flooding and other relevant water management issues. Bioretention cell is one of the most important green drainage facilities in the construction of sponge cities, with the function of storing rainwater and reducing surface pollution. In this study, a non-vegetative bioretention cell has been constructed, and experimental data on inlet water and outlet water concentrations are obtained through a long period of operation. The XGBoost and the SVM methods are employed to simulate the water quality model based on the input and output sewage concentration of the bioretention cell. The experiment results show that the XGBoost model has a better generalization ability than the SVM model to improve the prediction accuracy with higher decontamination effect. The proposed work can provide a significant guide to the watershed planning of water reuse system.