EM分类法在区域地球化学数据挖掘中的应用——以湖南省洞口地区1∶20万水系沉积物Pb异常识别为例

利用新技术、新方法从海量的区域地球化学数据中挖掘找矿信息,实现找矿突破,具有重要的理论与实际意义。以湖南省洞口地区1∶20万水系沉积物Pb异常识别为例,尝试利用EM(expectation maximization)分类法消除岩性背景的影响,提高Pb异常识别的准确率。根据研究区岩性背景特征,选择Na₂O、K₂O、Ca O、MgO、La、Nb、Th、Zr、Sr、Li、Mn、Ti和V作为分类指标,将研究区水系沉积物样品分成了5类。利用分类数据和原始数据,分别圈定了Pb地球化学异常。对比分析结果表明,分类前后圈定的Pb异常存在显著差异,主要表现为分类方法突出了低背景值区的一些低弱异常,同时削弱了高背景值区的一些虚假或无意义的异常。另外,新方法圈定的Pb异常与已知矿床(点)位置的对应关系更好。EM分类法可显著提高水系沉积物区域地球化学异常识别的精度。

Application of the EM clustering method in regional geochemical data mining: a case study of lead anomaly determination in stream sediments on a 1:200 000 scale in Dongkou, Hunan Province

It is of great importance to use new techniques and methods to mine prospecting information from a mass of regional geochemical data, achieving a breakthrough in exploration. In this study, the determination of Pb anomalies in stream sediments on a 1:200 000 scale from Dongkou, Hunan Province, was taken as an example, and an attempt was made to eliminate the lithology influence by using the EM (expectation maximization) clustering method, so as to improve the identification accuracy. According to the lithology background of the study area, Na2O, K2O, CaO, MgO, La, Nb, Th, Zr, Sr, Li, Mn, Ti, and V were selected as classification indicators, and the stream sediment samples collected from the study area were divided into five clusters. For comparison, Pb anomalies were delineated respectively using the classified data and the original data. The result shows significant differences in Pb anomalies delineated before and after classification, mainly manifested as the clustering method uncovered or enhanced some weak anomalies in the low-background area and removed or weakened some false or meaningless anomalies in the high-background area. In addition, Pb anomalies delineated by the new method correspond better with the known Pb deposits (spots). We conclude that the EM clustering method can significantly improve the identification accuracy of regional geochemical anomalies in stream sediments.