Fates of algal and natural organic matters during moderate pre-oxidation of potassium permanganate composite to enhance coagulation of algae-laden water.

Pre-oxidation has been increasingly applied to enhance coagulation during algae-laden water treatment, yet the transformation behaviors of algal organic matter (AOM) and natural organic matter (NOM) during moderate pre-oxidation of potassium permanganate composite (PPC) to enhance coagulation remain insufficiently understood. In this study, excitation-emission matrix fluorescence spectroscopy coupled with parallel factor analysis was employed to track source-resolved organic matter transformation, while multiple machine learning (ML) models were developed to predict fate of AOM- and NOM-associated fluorescent fractions. The results confirmed that PPC pre-oxidation dosed within a moderate oxidation regime without inducing algal cell rupture. PPC altered the structure and molecular weight distribution of organic matter, resulting in increased UV254 while exerting limited influence on DOC. Compared to coagulation alone, PPC-enhanced coagulation improved the removal of protein-like fluorescence substances (C2), while the enhancement of humic-like fluorescence substances (C1) representative of terrestrially derived NOM remained limited. Among the evaluated ML models, random forest achieved the best balance between predictive accuracy and robustness, particularly for C2 prediction (R2 = 0.965). The shapely additive explanations analysis further demonstrated that influent organic matter characteristics exerted greater influences on settled water quality than reagent dosages. This study provides an interpretable framework for the intelligent optimization of algae-laden water treatment using pre-oxidation of PPC followed by coagulation.