Modeling mineralization potential of catchment basins: Insight from geochemical fingerprint

Abstract

Stream sediments are common sampling media in cost-effective regional exploration programs. Catchment basins are the basic sampling units of steam sediments, which have distinct geochemical characteristics. In this study, the mineralization potential of catchment basin using stream sediments is evaluated from the perspective of geochemical fingerprints. Firstly, the spatial distribution of catchment basins was acquired based on digital elevation model data. The local singularity analysis was employed to reduce the impact of the overburden, revealing the weak geochemical anomalies. The catchment basins with known mineral occurrences as the prior dataset were used to train the random forest model, which improved the model accuracy and interpretability. The geochemical fingerprints of different types of mineralization were constructed based on the importance of geochemical variables in the random forest modeling, which revealed the importance of geochemical elements for different types of mineralization occurrences. All catchment basins lacking mineral occurrences were categorized based on the trained random forest model, achieving the potential types of mineralization for each catchment basin. Most of catchment basins with the same types of mineralization occurrences exhibited the spatial cluster, suggesting the migration characteristics of metal element in stream sediments through river system. Geochemical fingerprinting is a valuable tool for classifying the different types of mineralization occurrences at the scale of catchment basins. The delineated potential mineral occurrence types for catchment basins provided an indicator for future mineral exploration endeavors.