Prediction of the removal efficiency of emerging organic contaminants in constructed wetlands based on their physicochemical properties.

This study investigates the prediction of the removal efficiency of emerging organic contaminants (EOCs) (pharmaceuticals-PhCs, personal care products-PCPs, and steroidal hormones-SHs) in constructed wetlands based on their physicochemical properties (e.g., molecular weight-MW, octanol-water partition coefficient-Log Kow, soil organic carbon sorption coefficient-Log Koc, octanol-water distribution coefficient-Log Dow, and dissociation constant-pKa). The predictive models are formed based on statistical analysis underpinned by principle component, correlation, and regression analyses of a global data set compiled from peer-reviewed publications. The results show that the physicochemical properties of EOCs emerged as good predictors of their removal efficiency. Log Koc, Log Dow, and Log Kow are the most significant predictors, and combination with MW and/or pKa often improved the reliability of the predictions. The best performing model for PhCs was composed of MW, Log Dow, and Log Koc (coefficient of determination-R: 0.601; probability value-p < 0.05; root mean square error-RMSE: training set: 11%; test set: 27%). Log Kow and Log Koc for PCPs (R: 0.644; p < 0.1; RMSE: training set: 14%; test set: 14%), and a combination of MW, Log Kow, and pKa for SHs (R: 0.941; p < 0.1; RMSE: training set: 3%; test set: 15%) formed the plausible models for predicting the removal efficiency. Similarly, reasonably good combined models could be formed in the case of PhCs and SHs or PCPs and SHs, although their individual models were comparatively better. A novel decision support tool, named as REOCW-PCP, was developed to readily estimate the removal efficiency of EOCs, and facilitate the decision-making process.