Interannual variation and machine learning simulation of organophosphate esters in Taihu Lake

Abstract

Organophosphate esters (OPEs) are widespread in water bodies and have attracted public attention due to their hazards. This study investigated the presence of OPEs in surface water of Taihu Lake from 2012 and 2021–2022. The OPEs concentration was compared ten years ago and ten years later. Water and meteorological parameters were ranked using the random forest (RF) model, and OPEs concentration in lakes was simulated using selected parameters as inputs. The concentration of Σ7OPEs was higher ten years ago compared to ten years later. There was no significant seasonal difference in Σ7OPEs from 2021–2022, while the concentration of Σ7OPEs in 2012 was lower in summer than in other seasons. The spatial distribution of the two interannual Σ7OPEs exhibited a decreasing trend from the northwest region. The results of RF importance ranking and redundancy analysis showed that NH3-N, TN, TP, water temperature and relative humidity were the most influential factors affecting OPEs concentrations. RF models performed better for TnBP, as indicated by training R and test R values are excellent and relatively low errors. Our results demonstrated that machine learning models were useful in facilitating efficient monitoring and assessment of OPEs contamination in lakes.