Concentration-distance from centroids (C-DC) multifractal modeling: A novel approach to characterizing geochemical patterns based on sample distance from mineralization

Abstract Various fractal models have been implemented to separate populations and characterize spatial distributions in geochemical data derived from regional mapping programs. This study compares the conventional number-size with a proposed concentration-distance from centroids (C-DC) fractal model to detect geochemically anomalous populations. These models have been applied to centered log-ratio transformed data of VMS-style mineralization related element concentrations in till from a low-density national survey conducted across Sweden. The C-DC model has been applied to the distance between till samples and centroids of geological sub-provinces containing a number of VMS-style deposits, and is derived from the concentration-distance (C-D) approach originally developed using a radial-density (R-D) model. The efficiencies of the models in detecting a multivariate geochemical response to known mineralization are compared using a variant of the overall accuracy matrix. The C-DC model provides accuracy of classification similar to the N-S model. Therefore, Monte Carlo simulation was applied to quantify uncertainties in setting of population thresholds using the two fractal models. It demonstrated greater efficiency of the C-DC model.