KMeans-Riemannian model for classification mineral resources in a copper deposit in Peru

This study applies the KMeans clustering model with Riemannian geometric distance to classify mineral resources in a copper deposit in Peru. Covariance matrices of Ordinary Kriging estimates, kriging variance, and average sample distances are used to represent multivariate spatial structures for classification based on intrinsic geometry. The new automated method obtains similar results to the Qualified Person (QP), offering a reproducible and consistent framework aligned with geological variability and expert interpretation. The Riemannian approach improves spatial coherence and segmentation, making it suitable for deposits with complex geometries. This methodology supports objective, automated resource classification while preserving geological integrity.