沁河流域水体污染物时空分异特征及潜在污染源识别

应用多种多元统计方法对晋城市沁河流域2005—2010年的水质数据(11个监测断面、24个水质指标)进行时空分异特征分析及潜在污染源识别.将聚类分析与判别分析相结合,在空间尺度上识别出重污染区和轻污染区,在时间尺度上识别出水质的年内差异性,揭示并探讨河流水质对降雨延迟的响应现象,进而将时间及空间聚类分析结果有机结合,开展时空联合分析与因子分析,有针对性地进行污染源识别.结果表明,监测断面可分成2类:A组(4个断面)及B组(其余7个断面),其中,丹河污染程度重于沁河,应当作为流域污染控制的重点;全年可分为2个时段:时段Ⅰ(7—11月)和时段Ⅱ(其它月份),时段Ⅰ水质略优于时段Ⅱ,流域面源污染或内源污染较严重;A组的典型污染指标共12项,分别为重金属、有机物、有毒有机物和其他污染物,主要来自于工业点源;B组的典型污染指标共11项,分别为重金属、有机物和生物指标,主要来自于工业和生活点源、生活和农业面源.

Spatio-temporal characteristics and source identification of water pollutants in River Qinhe Basin

An integrated model of various multivariate statistical techniques was used in this study for water quality assessment and potential pollution sources identification of River Qinhe Basin in Jincheng City, China. The data were collected from 2005 to 2010, including 19008 observations. The spatial classification of monitoring sites and temporal classification of months were determined using Cluster Analysis (CA) and backward Discriminant Analysis (DA). Then the spatio-temporal analysis and Factor Analysis (FA) were used for specific and accurate identification of pollution sources. The results demonstrated that there were two distinct clusters among monitoring sites, i.e. cluster A with 4 sites and cluster B with 7 sites, indicating that it is essential to conduct more powerful remediation measures for Cluster A. The 12 months were classified into two periods, i.e. period I from July to November with better water quality and period II for the rest months with worse water quality. There was a significant difference for the typical water quality indicators between Cluster A and B. Further investigation found that industrial activities contributed greatly to the heavy metals, organic pollutants, toxic organics and other pollutants in Cluster A, while urban point sources and agricultural and rural non-point sources contributed greatly to heavy metals, organic and fecal pollutants for Cluster B. The current study demonstrated the powerfulness of multivariate statistical analysis in assessing spatial and temporal characteristics and in identifying potential sources of pollution for large and complex dataset, by effectively reducing dimension of the whole data set.