象山港海域水质时空格局的自组织特征映射神经网络识别

于2007—2008年对象山港23个站点(包括10个电厂站点)的水质样品进行连续2年的季节性采集,采用SOM(Self-Organizing Map)工具箱,结合k-nn(knearest neighbors)神经元聚类对15个水质参数进行分析,以探明象山港海域水质时空变化并识别敏感的影响区域.结果显示,象山港海域N/P(物质的量比)平均值为27.0.水体污染指数(AI)和海水营养指数(NI)分别指示整个象山港水质处于严重污染和富营养化状态,但水质加权指数(WDX)显示,加权水质标准未超过3类水质,说明传统的AI和NI指数不能反映象山港的实际水质状况.经SOM分析发现,象山港海域各取样站点按季节和空间格局可分为8个聚类组.从季节上看,pH和油类含量在春季最低;夏季水温、COD、NO2--N最高,而DO最低.NO3--N、DIN、DIP在秋冬季节高于春夏季节,但透明度相反.Chl-a含量以夏季最高,冬季最低.GLM(General Linear Model)方差分析显示,不同季节的安全性指数(SFT)和N/P无显著差异(p>0.05),而NI、AI和WDX差异极显著(p<0.01).空间分析显示,象山港水体可分为港底区和口中部区,其中,港底区盐度、pH显著低于口中部区(p<0.01),而NO2--N、NH4+-N、DIN、DIP、Chl-a则显著高于口中部区(p<0.05).除WDX无显著差异外,港底区的N/P显著低于口中部区(p<0.01),而NI、AI、SFT相反(p<0.05).建议港区底部宜采用养殖大型海藻方式以减轻富营养化,此外,冬季黄墩港的水体中粪大肠菌群严重超标,生食该季节贝类产品时需要检测.

Identification of spatiotemporal patterns of sea water quality in Xiangshan Bay by using self-organizing maps

In 2007 and 2008, seasonal samples were collected from 23 sites (including 10 power plant sites) in Xiangshan Bay. To explore spatiotemporal variations of sea water quality and to identify the response of sensitive sites to the variations, 15 parameters of water quality were subjected to Self-Organizing Map (SOM) toolbox in which neurons were clustered by k-nn (k nearest neighbors) clustering algorithm. The results showed that the average N/P ratio in Xiangshan Bay was 27.0 (molar ratio). Both water pollution index (AI) and seawater nutrition index (NI) indicated that severe pollution and eutrophication occurred in the bay. However, water quality weighted index (WDX) demonstrated that the weighted quality of seawater in Xiangshan Bay was not worse than the class Ⅲ national standard. Clearly, traditional indices of both AI and NI did not reflect the actual water quality in Xiangshan Bay. Based on SOM analyses, eight clustering groups for sampling sites were found according to seasonal and spatial patterns. In seasonal pattern, the lowest pH and oil content were observed in the spring. In summer, water temperature, COD and NO-2-N were the highest, while DO was the lowest. In autumn and winter, NO-3-N, DIN and DIP were higher than those in spring and summer, while higher transparency was found in autumn and winter. In addition, the concentration of Chl-a reached a maximum in summer, and a minimum in winter. General Linear Model (GLM) analyses also revealed that safety index(SFT) and N/P values were not significantly different among seasons (p>0.05). However, NI, AI and WDX showed significant differences (p <0.01). In spatial pattern, two areas of Xiangshan Bay, the bottom and the mouth-middle area, were identified with SOM. All examined parameters except WDX changed significantly between two areas. In the bottom area, salinity and pH were significantly lower (p<0.01), and NO-2-N, NH+4-N, DIN, DIP and Chl-a were significantly higher than those in the mouth-middle area (p<0.05). Furthermore, N/P values in the bottom area were significantly lower than the mouth-middle area (p<0.01), and reverse results were found in NI, AI and SFT (p<0.05). Based on these results, we recommend that the large-scale seaweed farming in the bottom area should be adopted to reduce eutrophication. Moreover, being eaten raw, shellfish was needed to be detected due to the fact that the numbers of fecal coliform in Huangdun Bay seawater during winter far exceeded quality standard of seawater.