Resume/ Abstract
The mining industry is adopting more the use of in-situ real time analytical techniques such as automated core scanning technology using multiple sensors and analytical techniques, directly on the drill cores. This cutting-edge system, generally equipped with high resolution camera, hyperspectral sensors spanning multiple wavelengths at sub-micron spatial resolution and X-ray fluorescence (XRF) sensors, enable rapid, high-resolution in-situ mineralogical and geochemical analysis at the core shack. By delivering dense, high-quality datasets, this technology can provide an early prediction of geometallurgical and geoenvironmental behaviour of the deposit. The integration of these rich and multi-level datasets with machine learning (ML) methodologies can further strengthen this technological impact by enabling the creation of detailed geometallurgical and geoenvironmental datasets. This, in turn, supports the development of more accurate and spatially refined sampling, leading to better 3D geometallurgical and geoenvironmental block models. Here we explore the potential of combining core scanning data with ML techniques to improve environmental characterization. The project focus is twofold: (1) optimizing sampling strategies through ML-driven insights from multi-layer core scanner data to facilitate more efficient sampling for geoenvironmental studies, and (2) predicting key parameters related to acid generation and acid neutralization for samples or areas lacking direct environmental laboratory measurements. These advancements can offer a pathway to enhance the efficiency and accuracy of environmental assessments, particularly in scenarios where traditional laboratory testing is either unavailable or time intensive. This study highlights how combining automated core scanning and ML analytics can transform environmental assessments to be more data-rich, efficient, and proactive.