Source identification of surface water pollution using multivariate statistics combined with physicochemical and socioeconomic parameters

Accurate identification of potential contamination sources of river water is a basis for effective pollution control and sustainable water management. Pollution source identification based on physicochemical-parameters-only method may lead to uncertainty and subjectivity. In this study along with hydrochemistry parameters (HPs), socioeconomic parameters (SPs) were considered as an auxiliary in multivariate statistics to achieve a comprehensive assessment on pollution sources with accurate estimates of source identification and apportionment. Fifteen physicochemical parameters were combined with twelve socioeconomic parameters in multivariate statistics to quantitatively assess potential pollution sources and their contributions to river water pollution. Multivariate statistics in the study included regression analysis, principal component analysis (PCA), and absolute principal component score-multiple linear regression (APCS-MLR). Regression analysis between hydro-chemical parameters and socioeconomic parameters indicated that industrial and population growths were the main factors related to ammonium nitrogen (NH4+-N), total nitrogen (TN) contamination, while total phosphorus (TP) was more correlated with domestic discharge and poultry breeding. Based on the results of PCA, four latent factors were extracted for hydrochemistry parameters (HPs) and socioeconomics parameters (SPs), accounting for 68.59% and 82.40% of the total variance, respectively. With integrating the PCA results of the two parameter groups, pollution sources were ranked as industrial effluents > rural wastewater > municipal sewage > phytoplankton growth and agricultural cultivation. Source apportionment in APCS-MLR revealed that industrial wastewater and rural wastewater averagely contributed 35.68% and 25.08% of pollution, respectively, followed by municipal sewage (18.73%) and phytoplankton pollution (15.13%) with relatively small percentage of unrecognized source. It is concluded that socioeconomic parameters assisting hydrochemistry parameters in multivariate statistics can improve the accuracy and certainty of pollution source identification, supporting decision makers to formulate strategies on protection of river water quality.