摘要:
以湘西北水系沉积物测量常量元素含量数据为研究对象,针对湘西北铅锌矿床的层控特征,以研究区含矿层地球化学常量元素含量的统计规律为基础,利用对数比变换( alr、clr、ilr变换)对原始数据进行预处理,并在此基础上进行偏最小二乘( PLS)降维分析,提取了地球化学组合异常,并与常用的主成分分析、因子分析的结果进行对比。分析结果表明,等距对数比变换( ilr变换)相对于其他变换方法有明显的优势,变换后的数据对称特征明显,更趋于正态分布。相对于其他两种降维方法,偏最小二乘法提取的元素组合与研究区铅锌矿的地球化学基本特征更为接近,组合异常与已知矿床分布和构造特征耦合良好,与研究区地质特征吻合,比主成分分析与因子分析得到的结果更易于地质解释。
Abstract:
As lead?zinc deposits in northwestern Hunan are controlled by strata, the authors chose statistical regularities of constant ele?ments in the ore?bearing layer as the basis of integrated geochemical anomaly extraction and used a stream sediment geochemical data?set. Firstly, three main forms of log?ratio transformation were used to explore the effects of the constant elements data closure problem. On such a basis, the PLS method was employed to study the integrated geochemical anomalies. The results show that data transformed by ilr method have obvious symmetrical statistical features and are much closer to the normal distribution. In contrast with PCA and FA, the integrated geochemical anomalies extracted by PLS method are generally more consistent with such geochemical features of lead?zinc mineralization in the study area as the known ore deposits and the fault distribution, and the results can be explained easier in geolog?ical field.
下载: