Integrate physics-driven dynamics simulation with data-driven machine learning to predict potential targets in maturely explored orefields: A case study in Tongguangshan orefield, Tongling, China
{"title":"Integrate physics-driven dynamics simulation with data-driven machine learning to predict potential targets in maturely explored orefields: A case study in Tongguangshan orefield, Tongling, China","authors":"Liangming Liu , Feifu Zhou , Wei Cao","doi":"10.1016/j.gexplo.2024.107478","DOIUrl":null,"url":null,"abstract":"<div><p>The physics-driven dynamics simulation (DS) and data-driven machine learning (ML) are two general approaches to predict complex systems whose complexity is a hardship impediment to prediction. Based on the 3D geological modeling (GD), we embedded the DS into ML to predict high potential targets and to evaluate ore-controlling and ore-indicating factors in the Tongguangshan (TGS) skarn orefield that has undergone intensive exploration and 4 Cu and Au deposits discovered. The 3D geological models show that the heterogeneous distribution of orebodies around intrusions is associated with the wall rock lithology and contact zone (CZ) characteristics of intrusions, and the resistivity can only provide some ambiguous clues for interpretation of underground geological architectures rather than a direct ore-indicator. The DS results show heterogeneous distribution of temperature, pore pressure, differential stress, volume strain and shear strain, among which the volume strain is closest associated with ore formation. Based on the prediction of Random Forest (FR) model of which the feature variables are combination of DS and 3D modeling results, the SHAP valuing results show a descending importance rank of ore-controlling factors and ore-indicators as lithology, volume strain, distance to CZ, distance to Devonian-Carboniferous interface, curvature of CZ, pressure, temperature, CZ azimuth, resistivity, differential stress, shear strain and CZ dip. The DS results are more important than the resistivity. We have run 6 RF models, consisting of different feature variables which were assigned by DS and 3D modeling, to predict ore-formation favor spaces. The prediction performances on test data sets suggest that, integrating of geological features with dynamics features can enhance performance of RF prediction, the RF model consisting of pure dynamics features can predict mineralization different from the training samples. All RF models' predictions support that there are no significant high potentials at the depth of the orefield, except one small target at its eastern south corner.</p></div>","PeriodicalId":16336,"journal":{"name":"Journal of Geochemical Exploration","volume":null,"pages":null},"PeriodicalIF":3.4000,"publicationDate":"2024-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Geochemical Exploration","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0375674224000943","RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"GEOCHEMISTRY & GEOPHYSICS","Score":null,"Total":0}
引用次数: 0
Abstract
The physics-driven dynamics simulation (DS) and data-driven machine learning (ML) are two general approaches to predict complex systems whose complexity is a hardship impediment to prediction. Based on the 3D geological modeling (GD), we embedded the DS into ML to predict high potential targets and to evaluate ore-controlling and ore-indicating factors in the Tongguangshan (TGS) skarn orefield that has undergone intensive exploration and 4 Cu and Au deposits discovered. The 3D geological models show that the heterogeneous distribution of orebodies around intrusions is associated with the wall rock lithology and contact zone (CZ) characteristics of intrusions, and the resistivity can only provide some ambiguous clues for interpretation of underground geological architectures rather than a direct ore-indicator. The DS results show heterogeneous distribution of temperature, pore pressure, differential stress, volume strain and shear strain, among which the volume strain is closest associated with ore formation. Based on the prediction of Random Forest (FR) model of which the feature variables are combination of DS and 3D modeling results, the SHAP valuing results show a descending importance rank of ore-controlling factors and ore-indicators as lithology, volume strain, distance to CZ, distance to Devonian-Carboniferous interface, curvature of CZ, pressure, temperature, CZ azimuth, resistivity, differential stress, shear strain and CZ dip. The DS results are more important than the resistivity. We have run 6 RF models, consisting of different feature variables which were assigned by DS and 3D modeling, to predict ore-formation favor spaces. The prediction performances on test data sets suggest that, integrating of geological features with dynamics features can enhance performance of RF prediction, the RF model consisting of pure dynamics features can predict mineralization different from the training samples. All RF models' predictions support that there are no significant high potentials at the depth of the orefield, except one small target at its eastern south corner.
期刊介绍:
Journal of Geochemical Exploration is mostly dedicated to publication of original studies in exploration and environmental geochemistry and related topics.
Contributions considered of prevalent interest for the journal include researches based on the application of innovative methods to:
define the genesis and the evolution of mineral deposits including transfer of elements in large-scale mineralized areas.
analyze complex systems at the boundaries between bio-geochemistry, metal transport and mineral accumulation.
evaluate effects of historical mining activities on the surface environment.
trace pollutant sources and define their fate and transport models in the near-surface and surface environments involving solid, fluid and aerial matrices.
assess and quantify natural and technogenic radioactivity in the environment.
determine geochemical anomalies and set baseline reference values using compositional data analysis, multivariate statistics and geo-spatial analysis.
assess the impacts of anthropogenic contamination on ecosystems and human health at local and regional scale to prioritize and classify risks through deterministic and stochastic approaches.
Papers dedicated to the presentation of newly developed methods in analytical geochemistry to be applied in the field or in laboratory are also within the topics of interest for the journal.