Fisher型逐步判别法在东海水团判别中的初步应用

The application of Fisher's Stepwise Discriminant Analysis to the water masses in the East China Sea

1998年2月份对东海区进行了大面水质调查。样品分析方法依据《中华人民共和国标准》GB17378.4-1998,所得数据应用美国StatSoft公司开发的STATISTICAS软件进行分析。聚类分析结果,东海表层主要水团配置为黄东海混合水团(HE)、东海表层水团(E)、沿岸冲淡水(C)、黑潮变性水团(EK);底层主要水团配置为东海表层水团(E)、黑潮变性水团(EK)。M为水团间过渡带。应用Fisher型逐步判别法建立判别方程式时,根据计算的统计量F值对表底层的指标值进行筛选。表层水团入选的指标值有温度、盐度、溶解氧、总磷、pH值和硝酸氮六个指标;底层水团入选的指标值有温度、盐度和叶绿素α三个指标。经显著性检验,建立的判别方程式符合聚类分析的要求,选定的指标可以有效判别水样归属。判别式回代验证结果表明,本文建立的判别式的正确率为100%。本次应用过程中没有出现空间站位划分混乱现象,说明冬季该海域水团分布均匀,没有受到上升流的影响。

In February 1998, the survey of water quality was conducted in the East China Sea. The samples were analyzed by the state standard of GB17378.4-1998.The indexes of temperature and salinity of investigating samples were clustered by the cluster analysis with the software STATISTCAS. The results showed that the surface water masses in the East China Sea included HE, E, C, EK and the bottom water masses included E, EK. M refers to the mixing zone between water masses. When building the formulations of Fisher's Stepwise Discriminant Analysis, the incoming indexes of the surface water masses include T, S, DO, PO_4-P, pH and NO_3-N. and the indexes of the bottom water masses include T, DO, and chlorophyll α. Under the test of prominence, the results were in accordance with the one that derives from the cluster analysis. The indexes in Fisher's Stepwise Discriminant Analysis can be used to efficiently divide the investigating samples into the formal clusters. In addition, the samples of this time have been correctly classified, which means the water masses are free from the impact of ascending currents.