3D mineral prospectivity modeling using multi-scale 3D convolution neural network and spatial attention approaches

Nowadays, mineral exploration has increasingly focused on the targeting of deep-seated orebodies. Mineral prospectivity modeling is one of the important approaches facilitating exploration targeting and mitigating risks associated with mineral exploration, particularly under cover. Recent advances in 3D mineral prospectivity modeling, enable the effective extraction of predictive information from three-dimensional geological models, enhancing the accurate identification of the deep-seated orebodies. Notably, these advancements have synergized with deep learning approaches to improve the efficiency of mineral exploration based on their nonlinear and multi-layer sensing attributes, effectively identifying and extracting key relationships between the 3D predictive maps and mineralization. Currently, the main deep learning method used for 3D mineral prospectivity modeling is convolutional neural network (CNN) models. However, the related research has not considered the multiscale features of geological structures, it can be made further improvements on this regard. This paper introduces a multi-scale 3D convolutional neural network model (3D CNN) incorporating a spatial attention mechanism and an Inception module (MSAM-CNN) for 3D mineral prospectivity modeling. By integrating Inception modules and spatial attention mechanisms, the network's capability to identify multi-scale geological features and concentrate on key predictive areas is significantly enhanced. This leads to further improvement in the accuracy and generalization capability of 3D mineral prospectivity modeling. To evaluate the effectiveness of this model, we carry out a case study on 3D mineral prospectivity modeling in the Baixiangshan iron deposit within the Ningwu Basin of the Middle-Lower Yangtze River Metallogenic Belt, China. The results show that the multi-scale 3D convolutional neural network model exhibits remarkable robustness and generalization capabilities. It can effectively delineate targets within the deep and peripheral areas of the deposit, providing Indications for future exploration. The addition, performance indicators, ROC curve, and Capture-Efficiency curve consistently demonstrate that the MSAM-CNN model outperforms the Inception-enhanced CNN (M-CNN), CNN, Random Forest (RF), and Support Vector Machine (SVM) models. All of these indicate that MSAM-CNN can extract the 3D spatial features within 3D predictive maps better during 3D mineral prospectivity modeling, positioning it as a promising tool for yielding more reliable targets for deep-seated mineralization in future exploration.