地学前缘 ›› 2019, Vol. 26 ›› Issue (4): 67-75.DOI: 10.13745/j.esf.sf.2018.6.25

• 矿床大数据研究 • 上一篇    下一篇

勘查地球化学数据挖掘与弱异常识别

左仁广   

  1. 中国地质大学(武汉) 地质过程与矿产资源国家重点实验室, 湖北 武汉 430074
  • 收稿日期:2018-01-22 修回日期:2018-04-12 出版日期:2019-07-25 发布日期:2019-07-25
  • 作者简介:左仁广(1981—),男,博士,教授,博士生导师,主要从事数学地质与矿产勘查方面的研究。
  • 基金资助:
    国家自然科学基金项目(41772344,41522206);湖北省自然科学基金项目(2017CFA053);地质过程与矿产资源国家重点实验室自主研究课题(MSFGPMR03-3)

Exploration geochemical data mining and weak geochemical anomalies identification

ZUO Renguang   

  1. State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (Wuhan), Wuhan 430074, China
     
  • Received:2018-01-22 Revised:2018-04-12 Online:2019-07-25 Published:2019-07-25
  • Supported by:
     

摘要: 我国积累的大量高质量、多元素、多尺度的地球化学数据,为矿产勘查与环境评价提供了有效的数据支撑。如何对这些数据进行二次开发和再利用,提取有价值的地球化学异常信息并带动找矿突破,是缓解当前矿产资源短缺的重要途径之一。在覆盖区和深部的找矿实践中,由于矿体埋深和覆盖层的影响,往往在表生介质中形成弱小的地球化学异常,识别和评价弱小地球化学异常是当前勘查地球化学数据处理的重要方向之一。本文围绕地球化学异常信息的提取和评价,主要从以下几个方面讨论了相关的国内外研究进展和发展趋势:勘查地球化学数据处理与异常识别方法和模型,勘查地球化学数据闭合效应的影响及其解决方案,基于大数据和机器学习的勘查地球化学数据处理以及弱小地球化学异常的识别和评价。研究发现,在地质环境的约束下,基于大数据思维和机器学习相结合的方法,注重地球化学空间分布模式与已发现矿床的相关关系,同时使用所有地球化学变量能有效刻画具有非线性特征的地球化学空间分布模式,可识别出传统方法无法识别的异常,为开展地球化学空间模式识别与异常提取提供了新的途径。

 

关键词: 勘查地球化学, 弱小地球化学异常, 分形, 机器学习

Abstract: We have collected in China many high-quality, multi-element and multi-scale geochemical exploration datasets, which provide an effective data support for mineral exploration and environmental studies. It is of great interest to further explore these datasets to identify geochemical anomalies associated with mineralization, in supporting breakthroughs in the next round of mineral exploration to alleviate current mineral shortage distress. However, deep buried or covered mineral deposits pose significant challenge to mineral exploration, what we are currently facing is how to identify and evaluate weak geochemical anomalies in the field of exploration geochemical data processing. In this paper, we reviewed the state-of-the-art methods and models for exploration geochemical data processing and geochemical anomaly identification. We also discussed the data closure problem and its solutions, as well as application of big data and machine learning methods to geochemical data processing. It is shown that the hybrid method, which combines big data thinking and machine learning methods under the constraints of geological settings, is a powerful tool to explore geochemical patterns and identify geochemical anomalies. The hybrid method takes into account the correlations between geochemical patterns and locations of mineral deposit, involving all geochemical variables, and can reveal non-linear characteristics of geochemical patterns.

Key words: geochemical prospecting, weak geochemical anomalies, fractal, machine learning

中图分类号: