摘要
本文应用聚类分析的方法对卢龙县低山丘陵区112个落叶阔叶灌丛样地资料进行了分类。样地间的相异系数采用欧氏距离、聚合策略选用最近邻体法、最远邻体法、中线法、形心法和平方和增量法。结果表明,最近邻体法空间压缩性较强,中线法和形心法存在逆转的缺点,最远邻体法和平方和增量法对样地资料的利用率最高,错分类最少,是较为理想的分类方法。文中还按平方和增量法将112个灌丛样地资料分为9个群系,23个群丛组。
This paper classifies 112 releves ( 24 species ) of the deciduous broad-leaved brush-wood in the hilly district in Lulong county using the cluster analysis method. The distance formula Euclidean is adopted to calculate similarity between the releves. 'We use five methods in the cluster analysis. They are, nearest neighbour, furthest neighbour,median, centroid and incremental sum of squares m'ethod. The results show that the nearest neighbour method has weak point in space-contracting, the median and centroid methods have the defect of reverse, the furthest neighbour and incremental sum of squares are the better classification nrethods, for they have the highest utilization ratio of releves and the least erroneous classification rate. Based on the result of incremental sum of squares method, 112 releves are divided into 9 formation, 23 association group.