Metallogenic Pattern And Mineral Prospectivity Modeling Of The Dashui Gold Concentration District

The Dashui gold concentration district is located at the edge of the Bailongjiang structural belt at the intersection of the western Qinling orogenic belt and the Songpan-Ganzi orogenic belt.In recent years,a large number of gold deposits such as Dashui superlarge gold deposit,Xinqu large gold deposit,and Zhongqu medium-sized gold deposit have been successively discovered within the Dashui gold concentration district.The prospecting results in this area show that the geological conditions of the Dashui gold concentration district are superior,and the main metallogenic geological characteristics of the gold deposits in the area are similar,showing the district has great prospecting potential.This paper studies the technical route that combines geology,geochemistry.geophysics,and deep learning numerical modeling techniques.Taking the mineralization geological characteristics of the study area as a guide,this article summarizes the ore-forming rules of the main gold deposits in the Dashui gold concentration district and extracts ore forming prediction factors,combined with regional geochemical element combination anomaly characteristics and regional magnetic anomaly characteristics,the ore forming potential were predicted and the exploration targets were located in the Dashui gold concentration district.By comparing the characteristics of the major gold deposits in the Dashui gold concentration district,and analyzing the regional gold mineralization geological background,the characteristics of mineralization,the main ore-controlling factors,and the methods for extracting prospecting elements in the Dashui gold concentration district are summarized.Based on the analysis of regional geochemical field and spatially weighted principal component analysis,the correlation between regional geochemical anomalies and ore-controlling structures is analyzed.According to the results of the spatially weighted principal component analysis,the gold deposits in the area are closely related to the geochemical abnormaly combination of low-temperature ore-forming elements including Ag,As,Au,Hg,Sb,and W,indicating that the ore-forming areas are of epithermal ore-forming characteristics.In particular,the Au-Ag-Sb-W combination anomaly corresponds well with the main gold deposits in the ore concentration district,which can be used as an indicator of ore-forming hydrothermal activity in the study area.It is one of the key geochemical prospecting elements in the Dashui gold concentration district.Furthermore,the spatially weighted principal component analysis was used to extract the spatially weighted allocation weights between different principal components to further refine the characteristics of geochemical prospecting elements in the area.By studying the regional geophysical field and combining multi-scale wavelet analysis,the concealed structure and rock mass information that are closely related to the mineralization in the region were identified.It was found that the magnetic anomalous gradient zone around the rock mass is more closely related to mineralization than rock mass itself.The rock mass and its surrounding structural fracture zone provide heat source and place for gold mineralization and enrichment The anomalous distribution and its geological characteristics indicate that the buffer zones of intrusive rock and tectonic anomalous can be used as prospecting elements in the Dashui ore concentration district.Based on the above knowledge,the mineralization prediction factors and extraction mechanism of the Dashui gold concentration district are preliminarily summarized.Aiming at the problem of the application of mineralization prospectivity modeling in the context of geological big data era,a new model framework of mineralization prospectivity modeling based on deep learning is proposed.A deep learning model framework named GeoNet was adapted to geological big data.Aiming at the problem of possible imbalanced sampling in the application of deep learning technology to mineralization prospectivity modeling tasks,this paper proposes a cost-sensitive learning mechanism for orebody misclassification in terms of algorithm design and an OGA-SMOTE grade adaptive oversampling technology in terms of data processing.Aiming at the problem of overfitting in the deep learning model,this paper introduces the data augmentation mechanism of the mineralization prediction model.Aiming at the problem of "scale trap" of the mineralization prediction model,this paper proposes a Multi-Scaler attention mechanism for the extraction of multiscale mineralization information ROC curve analysis technology and P-A diagram were used to evaluate the prediction ability and prediction efficiency of the mineralization prospectivity model,respectively.Finally,this paper uses the GeoNet model based on the convolutional neural network algorithm to evaluate the prospective potential of the Dashui gold concentration district,delineating six prospecting prospective target areas,and verifies them by drilling in relevant units.Which provides direction for mineral exploration and drilling deployment in the Dashui gold concentration district.