Diagnosing method conditioned bias in mineral resource estimation using a mutual information and entropy uncertainty...

Area of Science:

• Geostatistics
• Mineral Resource Estimation
• Data Science

Background:

• Conventional mineral resource interpolation focuses on accuracy and compliance, overlooking method-dependent stability.
• Existing validation methods do not fully capture the impact of chosen interpolation techniques on estimation outputs.

Purpose of the Study:

• Introduce a novel indicator, Mutual Information and Entropy Based Interpolation Uncertainty Indicator (MUI), to diagnose method-conditioned instability in mineral resource estimation.
• Evaluate the coherence and stability of interpolation outputs under a defined realization protocol.

Main Methods:

• Developed the MUI indicator using normalized entropy and mutual information from controlled realization ensembles.
• Constructed comparable multi-realization ensembles for Inverse Distance Weighting (IDW) and Ordinary Kriging (OK) for Cu estimation.
• Applied the methods to a heterogeneous porphyry-skarn deposit with skarn and hornfels domains.

Main Results:

• Ordinary Kriging (OK) yielded a less coherent and stable realization ensemble compared to Inverse Distance Weighting (IDW), particularly in the skarn domain.
• Both IDW and OK demonstrated comparable predictive accuracy when benchmarked against infill drilling data.
• Realization stability and interpolation accuracy were identified as distinct aspects of estimation behavior.

Conclusions:

• The MUI indicator serves as a practical diagnostic tool for potential method-induced bias, complementing standard accuracy-based frameworks.
• MUI captures method-dependent differences not explicitly expressed by conventional validation.
• Combined with benchmarking and resampling, MUI offers a safety-oriented assessment of interpolation accuracy, robustness, and explainability.