基于多重主成分分析的地球化学异常提取

Geochemical Anomaly Extraction Based on Multiple Principal Component Analysis

地球化学异常识别是勘察地球化学工作中的中重要环节。充分利用地球化学元素的频率分布和空间分布规律,对识别多变量地球化学异常的方法有重要意义。本文以山东省胶东半岛西北部地区为例,提出一种基于多重主成分分析的地球化学异常提取新方法。从高维数据的角度出发,通过多重主成分分析提取地球化学数据空间维度和元素维度的主要信息。捕捉地球化学数据的元素相关性和空间结构,重建出地球化学数据,进而通过原始值与重建值之间的距离计算地球化学异常得分。已知金矿点分布在异常得分分数高的区域;使用AUC指标评价异常得分与已知金矿点之间的空间关联度(AUC = 0.856)。结果显示,本文提取的地球化学异常分数与已知金矿床之间存在密切的空间相关性,基于多重主成分分析的方法能够有效提取地球异常。

Geochemical anomaly identification is an essential part of geochemical exploration. Making full use of the distribution of frequency and spatial distribution patterns of geochemical elements is important for identifying multivariate geochemical anomalies. In this paper, a new method of geochemical anomaly extraction based on multiple principal component analysis is proposed for the northwestern part of Jiaodong Peninsula in Shandong Province as an example. From the perspective of high-dimensional data, the main information of spatial dimension and elemental dimension of geochemical data is extracted by multiple principal component analysis. The elemental correlation and spatial structure of the geochemical data are captured, and the geochemical data are reconstructed. Subsequently, the geochemical anomaly score is calculated by the distance between the original value and the reconstruction value. The known gold deposits are distributed in the areas with high anomaly scores;the spatial correlation between the anomaly score and the known gold deposits is evaluated using the AUC index (AUC = 0.856). The results show that there is a close spatial correlation between the extracted geochemical anomaly scores and the known gold deposits, and the multiple principal component analysis method can effectively extract the geochemical anomalies.