Mapping Potential Targets for Gold Mineralization Based on Geochemical Data Score Balances

Описание: Presented at The International Association Mathematical Geosciences Meeting in Throndheim, Norway, 7-10 August 2023
Geochemical exploration in deeply weathered tropical terrains is often hampered by a strong modification of the primary geochemical signals, on the surface, making it difficult to locate potential targets for gold mineralization relying on geochemical assays for Au in soil samples. Moreover, geochemical results for gold in soil samples maybe misleading due to the difficulty in obtaining representative samples coupled with analytical problems and unstructured variability such as the nugget effect.

This problem is particularly relevant in the Amapari region, Amapá state, Brazil where humid tropical conditions contributed to develop regolith profiles that may be several tens of meters thick topped by lateritic crusts or highly leached colluvial soils. Therefore, it is important to examine carefully the metal distribution near the potential gold target in order to extract geochemical vectors to Au mineralization. In this study we utilize an innovative methodology based on score balances and data mining.

The aim is to select potential gold targets within a gridded area with a favourable metavolcano-sedimentary geological setting. Although no data mining package is used, the essential steps of the SEMMA (Sample, Explore, Modify, Model, Access) paradigm are followed in this work. The data is composed of a vast soil geochemical survey conducted in the region. A training step based on mineralized and non mineralized samples is essential to select the best variables to be adopted in a balance formula whose score will be calculated over a validation area. The mathematical background is given by the compositional analysis theory. Most calculations were done with functions of the package “Compositions” in and “Selbal”. The training samples were used to generate a numerical model to be used for discriminating between mineralized and non mineralized multivariate geochemical patterns.

The proposed method was successfully applied to highlight a mineralized trend in the Amapari region, which was not previously detected considering only Au grade or any other potential pathfinder metal alone. Given similar geological conditions of very thick overburden coupled with the lack of a distinct signal by the assayed gold grades, the method may be of use in other regions and even for other valuable metals.