Research on total phosphorus prediction in the Moyang River Basin based on machine learning models
Total phosphorus(TP)is an important indicator of water quality.Accurate prediction of TP variation in rivers is the basis for precise prevention and control of water pollution in river basins.On this basis,taking the Moyang River Basin as an example,TP data were collected from three sections,namely Zhonglang,Hekouzhen,and Jiangcheng,from January 2022 to June 2023 to analyze the distribution characteristics of TP,and to predict the TP at Jiangcheng section based on support vector machine model and random forest model.The results indicated that there was significant spatiotemporal heterogeneity in TP in the Moyang River Basin.The average annual TP concentrations at the Hekouzhen and Jiangcheng sections were similar(0.198 mg/L and 0.193 mg/L,respectively),higher than those at the Zhonglang section(0.149 mg/L).There was a clear covariant relationship between TP and rainfall,and the TP concentration at each section showed an upward trend during rainfall period.Compared with support vector machine model,the prediction accuracy of random forest model was higher,in which the correlation coefficient(R2)between the predicted and actual values during the testing period was 0.67,the mean relative error(MRE)was 12.70%,and the mean absolute error(MAE)was 0.036 mg/L.After incorporating the current TP as a time explanatory variable into the model,the prediction accuracy of the random forest model was further improved,with R2 increasing to 0.75,MRE decreasing to 6.44%,and MAE decreasing to 0.018 mg/L.Overall,the random forest model owned good generalization ability and could predict the variations of TP concentration in the Moyang River Basin well.
Total phosphorusrandom forestsupport vector machinepredictionMoyang River Basin
NETL
NSTL
万方数据
