RSC’s role in advancing mineral resource estimation
AusIMM’s upcoming Mineral Resource Estimation Conference will share leading best practices in resource estimation, covering topics such as Data Science, Machine Learning, Geological Modelling, Estimation, and Reporting. RSC brings advanced geostatistical techniques and data-driven solutions to the mining industry. In this article, RSC showcases its contributions to these areas and the future of mineral resource estimation.
RSC is the proud Platinum Sponsor of the Mineral Resource Estimation Conference 2025.
About RSC
RSC’s resource estimation services aim to offer the perfect synergy between data analytics, geology and geostatistics. RSC’s Resources and Reserves team includes globally respected and published specialists in data quality management, QA/QC, geological modelling, geostatistics, and advanced resource estimation with a broad range of geological and mineral system expertise.
At RSC, we love data and take great pride in building the best possible geological models to underpin estimation domains. Our toolkit covers the full spectrum of estimation techniques, including uniform conditioning, indicator kriging, multivariate compositional simulation, and conditional simulation.
Quality of the Informing Data
We make sure that the data informing our models are sound using state-of-the-art assessment processes. We work with our clients to identify data issues and areas for improvement, optimise SOPs, and develop comprehensive data quality management systems (DQMS). We love the challenge of creating systems that measure quality, truly move the needle at an operational level, and benefit all stakeholders.
We design systems that reduce variance in data, increase confidence in models and estimates, and ultimately better separate ore from waste. Specifying the accuracy and precision requirements of data is necessary to ensure meaningful outcomes from quality control and quality testing. Determining what is acceptable and what is not must occur before, not after, the data collection. Without this being in place beforehand, an operation runs the risk that its DQMS ends up being a bureaucratic, ineffective system that is a drain on company resources.
Our recent projects include developing a DQMS for a seabed nodule sampling exploration campaign to support a mineral resource upgrade targeting an Indicated classification, reviewing duplicate fire assay and Chrysos PhotonAssay™ analyses for an orogenic Au resource, and data quality and resource audits for several major and mid-tier mining companies.
Data-driven Domain Modelling
The way data are constrained in the estimation process is the most significant factor contributing to the quality of the eventual resource estimate yet is a key area of weakness (see, for instance, Sterk et al., 2019). A solid understanding of mineral systems and the geological processes that form and preserve a mineral deposit is the foundation of a quality model and estimate.
Geological information should be the primary input for domains and complemented by proxies for geology, e.g. geophysical, geochemical, petrophysical data, etc. Recent technological advances such as multivariate data analysis, machine learning, and downhole optical televiewers are powerful tools to support the proper delineation of domains.
Embracing new methods of interpreting and extracting value from complex datasets, e.g. through artificial intelligence, machine learning tools and workflows, can add to our deposit understanding by finding subtle correlations we cannot see using traditional approaches. The fundamentals of exploration geology (geological logging, mapping, and structural interpretation) are still as critical as ever, but we now have access to innovative approaches that can handle large volumes of data and move us further in evaluating domains and, just as importantly, the uncertainty attached to their definition.
Getting the most out of these large stores of data requires converting them to knowledge and not letting them languish in a database. Our in-house principal component analysis and machine-learnt scripts can use multi-element geochemistry to identify whether lithology and alteration logging codes are robust and whether geological domains are multi-modal or require further resolution to support high-quality domaining and estimation. The affinity for clusters to be dominated by specific logged lithologies, alteration or mineralisation types, and/or to be spatially contiguous when no spatial information is included in their calculation suggests that the clusters have meaning in terms of geological process.
We are seeing more deposits where multi-variate geochemical domains reflect mineralogy to a degree that is difficult or near on impossible to produce using geological mapping, visual logging, or conventional geochemical data approaches. Using a modern advanced toolkit allows us to look deeper and define domains at the required resolution for an efficient mineral resource estimate to be prepared.
Recent projects include remodelling of geological domains for a complex porphyry, skarn and epithermal project using geochemical proxies (e.g. Co/Al ratios for jasperoid and massive pyrite); multi-element geochemistry domain modelling for Nb-REE mineralisation in weathered-carbonatites where visually distinguishing lithology and the degree of chemical weathering in core was challenging; and a high-level review of the mineralisation controls and grade architecture for an intrusion-related Au deposit to assess the appropriateness of an existing geologically unconstrained model and multiple indicator kriging estimate.
Value Across the Mining Cycle
In addition to typical resource estimation, our resource services also include drillhole planning and spacing optimisation, resource auditing, operational performance improvement, reconciliation studies, blending and stockpiling optimisation, and risk analysis through conditional simulation. RSC is currently working on an extensive reconciliation study to support a system upgrade for a major African orogenic Au deposit.
Our team also recently contributed as domain experts to a brownfields orogenic Au targeting exercise that applied both supervised and unsupervised machine learning workflows using geological, geophysical, and geochemical data. One component of the exercise involved extracting the spatial characteristics of a shear zone by processing its wireframe and then conducting exploratory data analysis to investigate the existence of numerical relationships between Au grade and the prominent geometric features of the shear zone.
We are proud to be the platinum sponsor for the AusIMM’s Mineral Resource Estimation Conference and look forward to discussing our projects and favourite resource topics with conference attendees next month.
References
Sterk, R, (2019). Domaining in Mineral Resource Estimation, in Proceedings Mining Geology 2019, p 9 (The Australasian Institute of Mining and Metallurgy: Melbourne).