COMPARISON ON DETECTION METHODS WITH DISSOLVED ORGANIC MATTER IN RIVER WATER OF RAINY PRIMARY SEWAGE DISCHARGE FROM A COMBINED SEWAGE TREATMENT PLANT BASED ON CONFUSION MATRIX

Abstract

 We investigated changes in dissolved organic carbon (DOC) and the abundance of fluorescent dissolved organic matter (FDOM) components in a river before and after the occurrence of primary treatment discharge from a combined sewage treatment plant. We proposed DOC, the score value of the protein-like component (Comp. 2) of the FDOM, and FDOM/DOC as indicators detecting the change in river water quality due to the inclusion of untreated sewage during rainy events. Then, we evaluated the suitability of the three indicators by comparing their absolute values and their change rates with ratios of estimated primary treatment discharge to river flow. The suitability of the indicators was judged based on binary classification using machine learning, that is confusion matrix with accuracy, and F-value. As the results, it was found that the absolute values of DOC and FDOM were effective in detecting of the occurrence of primary treatment discharge at the downstream point of the receiving river because of the large accuracy rates and F values.