GRSG 34th Conference 2023

Title: Marigold on the DL Platform: Utilising Bare Earth Composites and Cloud Compute for Regional Scale Mineral Exploration – A Case Study over the Maricunga Belt, Chile

Author: Sam Casement


ASTER and Sentinel-2 satellites have been the principal open-source, regional-scale multispectral optical imagery for mineral exploration for some time. With ASTER being launched in 1999, and Sentinel-2 in 2015 (and still actively collecting), there is a large archive of data to draw from globally. Despite this, satellite data has remained relatively inaccessible to the exploration geologist, with specialist software and knowledge required to pre-process, process and extract information from this data. At Descartes Labs (DL), we have generated our Bare Earth Composites, selecting the best pixels on a global scale for mineral exploration. Using machine learning algorithms, we have detected the pixels with the lowest cloud, shadow, vegetation, snow, and water, from the entire ASTER and Sentinel-2 archives, corrected them for look angle differences and optimised them for outcrop detection. This data has been pre-processed to surface reflectance so that it is analysis ready on a global scale for geologists to use in their mineral exploration workflows straight away.

Analysing global datasets presents a challenge in terms of data management and processing speed. To tackle this, at DL we have leveraged an online cloud computing approach to manage high data volumes and allows for scalability, significantly accelerating the speed of workflow generation over regional scale areas with a mineral exploration focus. To further improve this speed and ease of access, DL have created Marigold – a no-code, analytics interface designed specifically for mineral exploration. Multispectral datasets can be assessed quickly and thoroughly, through a number of processing tools including: band ratios, mineral indices, transforms, classifications and spectral mapping, allowing for a larger proportion of time spent analysing an area, or multiple or larger areas at the same time. With mining companies looking to save cost and improve productivity, Marigold provides solutions to this by reducing idea-to-prospect exploration costs, empowering larger groups to access remote sensing data and allow for mineral targets to be identified and evaluated faster.

The growing demand for copper and other minerals associated with the energy transition means that underexplored areas need to be considered for targeting, and established mineral districts need to be reassessed. Marigold presents tools for both of these opportunities and we will showcase how the tool can rapidly assess areas of interest for mineral targeting. We present a case study, using our Bare Earth Composites, over the Maricunga Belt, Chile, for targeting porphyry and epithermal-style deposits. Found along the western crest of the Andes, the Maricunga belt hosts a series of porphyry and high sulphidation epithermal deposits within Oligocene to Miocene volcanics. Alteration zones associated with mineralisation are large and widespread across the belt, making the region an ideal candidate for remote sensing. In this example, we present how spectral analysis from known deposits within the region can provide insight into targeting alteration to inform a field campaign.