论文部分内容阅读
多变量统计方法中之主成份分析法被用于确定美国路易斯安那之火炬松及湿地松树干削度。用了四组资料(三组火炬松、一组湿地松)。资料包括单木在地面基部,0.02、0.04、0.06、0.08处及树高每隔0.1处的直径。这样每组(ⅰ)包括ni株数,每株有14个直径测定值。主成份分析用于每组资料在各个情况下单一的特征超过总方差99％。与主要特征值相关的特征向量元素图,代表了该组资料中的树木干削度的平均值。同组资料在胸径或树高级之间反映在干形上未发现有差别,但具有冠比大于0.51的树木在其树干0.3的上部有较大的削度。不同范围内生长的树木之间在干形上可见明显的差别,生长在同一范围内的两种树种间也有明显差异。同不同方法将第一特征向量内插以获得干削度模型。我们用回归技术,把第一特征向量当作因变量,相应的位置高及它们的幂次当作自变量。这样得到的方程式用于确定材积。我们相信第一主成份特征向量曲线至少是描述干削度规律的近似值。
Principal component analysis in the multivariate statistical method was used to determine the flattering of Pinus taeda and Pinus elliottii in Louisiana, USA. With four sets of information (three sets of loblolly pine, a group of Pinus). Data include the single wood at the ground base, 0.02,0.04,0.06,0.08 and the height of the tree at every 0.1 diameter. Thus each group (i) includes the number of ni strains, each having 14 diameter measurements. Principal component analysis is used for each set of data in each case for a single feature that exceeds 99% of the total variance. The eigenvector element maps associated with the main eigenvalues represent the average of the tree dryness in this set of data. No differences were found in stem diameter between DBH or tree height for the same group of trees, but trees with crown ratio greater than 0.51 had a greater taper on the top of trunk 0.3. Significant differences were observed in dry form between trees grown in different areas, with significant differences between the two species grown within the same range. The first eigenvector is interpolated in the same way to obtain a dryness model. We use regression techniques to treat the first eigenvector as the dependent variable, the corresponding high position, and their power as independent variables. The equation thus obtained is used to determine the volume. We believe that the first principal component eigenvector curve is at least an approximation that describes the law of dryness.