MULTIVARIATE DATA ANALYSIS OF WATER QUALITY AND SOURCE IDENTIFICATION IN AN URBANIZED RIVER

摘要

Multivariate methods were utilized to simultaneously investigate spatial variations of water quality, to seek latent factors, to identify pollution sources and optimize sampling sites in Baitapuhe River in Shenyang City of northeast China. Thirty-six mixed water samples were collected from the mainstream and tributaries along a human impact gradient, i.e. rural, town and urban regions. No-multidimensional scaling anticipatorily clustered the sampling sites into three groups, which indicated that the water quality level of the river across the rural region was the highest, followed by the town and urban regions. In the rural region, four latent factors (temperature, pH, COD and DO) were detected using principal component analysis. It represented natural pollution from temperature effect and soil erosion, apart from livestock/poultry wastewater. In the town region, TN, NH4-N, TP and DO, were sought as the latent factors, revealing that pollutants mainly originated from domestic and industrial wastewater, except for the accumulation pollution from the rural region. In the urban region, the latent factors included COD, TP, EC, NH4-N and DO. It evidenced point pollution, and pollutants mainly derived from treated/untreated domestic sewage and industrial wastewater. Based on discriminant analysis, the sites M10 and M11 should be adjusted, and two or three sampling sites added in the town region.