成分数据统计分析的定和问题解决方案

Solutions of Composition Translation in Composition Data Analysis

利用成分数据统计分析方法处理数据时,数据的定和问题的处理办法直接影响到统计结果的自然属性和实际意义。本文结合成分因子在估计沉积物端元中的应用,探讨定和问题的两种解决方案,即:"剩余变量法"和"平均含量限制法"。将这两种方法及传统的端元估计结果与实际端元进行对比,对比结果表明本文方法在端元估计方面比传统方法具有更好的效果。

Two solutions of composition translation in composition data analysis are introduced in this paper: one is residual variable composition translation,and the other is average sum composition translation. By using these methods and the traditional method on Okinawa Trough oceanic sediment datasets,the Changjiang-origin end-members are estimated from those datasets respectively.Their application indicates that the methods presented here are more effective than traditional ones in endmember estimation.