在叶片中5种主要养分元素测定和分析的基础上,首次应用判别分析法建立了综合指标诊断判别系统,对不同产量类型的锥栗人工林营养进行了判别诊断分析。当将产量水平划分为2类时,采用1维判别分析进行回判,不同的分类方法可得到不同的正判率,高产类及(中产+低产)类的划分所得到的正判率均为100% ;而(高产+中产)类及低产类的划分所得到的正判率分别为88.33%和86.67%。将产量划分为高、中、低3类的情况下,用1维判别进行回判的结果其正判率分别是100%、86.70%和83.33%;用2维判别进行回判,则正判率分别为100%、100%和83.33%。因此,类别的划分及判别函数的维数是影响判别诊断分析效果的重要因子。
The paper established a diagnosis distinguishing system by synthetic indices using distinguishing analysis method firstly, and used this method to diagnose the nutrient situation of Castanea henryi plantation in different yield types based on the measuring and analysis of five main nutrient elements in leaves. There were different correct distinguishing rates in different type dividing method by using distinguishing analysis of one dimension, which were 100% both for high yield type and middle yield + low-yield type while being 88.33% and 86.67% for high yield + middle-yield type and low-yield type respectively when dividing into two types. When dividing into three type (high, middle and low yield), their correct distinguishing rates were 100%, 86.70% and 83.33% respectively by using distinguishing analysis of one dimension but were 100%, 100% and 83.33% respectively by using distinguishing analysis of two dimensions. So, type dividing and dimensions of distinguishing function are two important factors affecting the results of distinguishing diagnosis analysis.
全 文 ：