Mineral prospectivity mapping for Gejiu tin district using random forest and convolutional neural network methods
Rattachement africain : cn, Rwanda. Niveau de preuve : code pays fourni par la source.
Le résumé fourni par la source
• HPF fusion technique was used to integrate multi-sensor remote sensing data. • The S-A model was used to separate the geochemical anomaly from the background. • CNN model was demonstrated superior performance in mapping tin prospectivity. The Gejiu district, the world’s most economically significant primary tin producer, served as our test area for developing an advanced mineral exploration targeting system integrating multi-source geospatial data with machine learning. In this paper, we constructed an exploration target model for tin-polymetallic deposits in Gejiu district using ASTER and Sentinel-2A images, stream geochemical datasets, and both random forest (RF) and convolutional neural network (CNN) modeling approaches. The workflow comprised: (1) generation of fused remote sensing datasets with optimized spectral-spatial resolution; (2) extraction of hydrothermal alteration features (iron oxide, Al-OH, Mg-OH) and structural elements; (3) spectrum-area fractal analysis of geochemical anomalies; and (4) integration of these predictive layers through ensemble machine learning. RF predictive modeling (AUC = 0.83) effectively delineated tin-polymetallic prospective zones through its robust handling of categorical geospatial data, with SHAP (SHapley Additive exPlanations) analysis identifying iron-stained alteration intensity and Sn-W-Bi multi-element geochemical anomalies as the most statistically significant predictors. CNN architecture (AUC = 0.96) demonstrated superior performance in exploring complex spatial mineralization patterns, particularly in recognizing non-linear relationships between complex ore-controlling factors and mineral distribution. This study confirms the tin-polymetallic potential of the Gejiu district and establishes a reproducible, data-driven exploration framework that identifies strategic targets for future mineral exploration.
Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.
Le contrôle bibliographique ouvert
DOI retrouvé dans Crossref DOI retrouvé ; titre concordant.
- Titre Crossref
- Mineral prospectivity mapping for Gejiu tin district using random forest and convolutional neural network methods
- Date Crossref
- 01/06/2026
- Éditeur
- Elsevier BV
- Type
- journal-article
Ce recoupement confirme des métadonnées liées au DOI. Il ne confirme ni la méthode ni les conclusions de l’étude, et il ne compte pas comme une seconde source scientifique indépendante.
Où se fait cette recherche
-
China University of Geosciences pays non établi dans la noticeUniversité ou école supérieure
-
Ministry of Natural Resources Rwanda (code pays fourni par la source)Organisme public
-
Nanjing Surveying and Mapping Research Institute (China) pays non établi dans la noticeEntreprise
-
Hubei University of Science and Technology pays non établi dans la noticeUniversité ou école supérieure
-
School of Earth Resources pays non établi dans la noticeUniversité ou école supérieure
-
Yunnan Institute of Geology and Mineral Surveying and Mapping Co. pays non établi dans la noticeStructure de recherche
-
Yunnan Key Laboratory of Intelligent Monitoring and Spatiotemporal Big Data Governance of Natural Resources pays non établi dans la noticeStructure de recherche
-
School of Resources and Environmental Science and Engineering pays non établi dans la noticeUniversité ou école supérieure
China University of Geosciences, Ministry of Natural Resources (Rwanda) et Nanjing Surveying and Mapping Research Institute (China), avec 5 autres affiliations. Pays d’affiliation : Rwanda.
Une affiliation ne permet pas de déduire la nationalité d’un auteur.