Integrating Soil Geochemistry and Machine Learning for Enhanced Mineral Exploration at the Dayu Gold Deposit, South China Block

Abstract

Combining traditional geochemical methods with advanced analytical techniques is a hallmark of contemporary exploration efforts. This study explores the intricate geological dynamics of the Dayu gold deposit, located in the Dayao Uplift of the South China Block. Using a multidisciplinary approach that includes soil geochemistry, conventional geochemical methods and advanced computational techniques such as machine learning and Discriminant Projection Analysis (DPA), we aim to uncover the deposit formation information. Our results reveal a complex pattern of element anomalies, which serve as a geochemical fingerprint of the Au mineralization processes that shaped the deposit over geological time. Principal Component Analysis (PCA) and cluster analysis on soil samples highlight significant correlation among Au and its pathfinder elements. By leveraging the predictive capabilities of machine learning algorithms, particularly Convolutional Neural Networks (CNN), we improve exploration strategies, enhance the precision of target delineation and guide sampling efforts. DPA further identifies distinct discriminant functions, aiding in group differentiation and providing insights into prospective mineralization zones. This study exemplifies the integration of traditional and innovative methodologies, offering a pathway to a deeper understanding of mineralization processes and improving the effectiveness of exploration in complex geological terrains. The findings advance our knowledge of the Dayu gold deposit and demonstrate the potential of these integrated approaches in similar geological settings.