Spatially-Weighted Factor Analysis for Extraction of Source-Oriented Mineralization Feature in 3D Coordinates of Surface Geochemical Signal

This contribution proposes a spatially weighted factor analysis (SWFA) to recognize effectively the underlying mineralization-related feature(s) in geochemical signals. The 3D spatial properties of the sampled surficial earth materials provide the opportunity to orient the results toward the potential sources through defining the proper localization functions. Perceiving the hydrothermal alterations as mineralization-indicating vectors in geochemical systems, a weighting function integrates the distance to prospective alteration zones and the geometry of productive geo-objects into a monolith formulation to achieve the source-oriented results. A covariance matrix tuned by the system localization function reformulates the standard factor analysis (FA) model to manifest source-oriented mineralization factor(s). The established mathematical setting was adapted to the compositional nature of multi-elemental signals and implemented via a combination of programming in MATLAB platform and R packages. An experiment on a porphyry Cu deposit was subjected to the outlined procedure for performance appraisal and comparison with the FA. The results indicated that the use of a weighting function configures the permutation of eigenvalues in such a way as to reflect spatial zoning from proximal to distal signals while providing clearly interpretable eigenvectors for ore-forming elements. By amplifying the signal of interest and reducing the signal of uncalled-for geo-processes, SWFA modulates the frequency distribution and spatial continuity of the feature of interest in such a way that the continuous-value mineralization landscape is allowed to be more consistent with the subsurface metallogenic reality in the survey area. A receiver operating characteristic analysis was adopted to evaluate quantitatively the factorized signal in predicting the mineralized ground to narrow down on the target areas. The results revealed a significant spatial coincidence between the source-oriented metallogenic pattern and mineralization evidence, implying superiority over the model derived by standard FA. The suggested scheme holds potential to information as a more efficient basis for follow-up exploration.