The effect of map scale on geological complexity for computer-aided exploration targeting

Abstract

Mineralization in hydrothermal ore systems has proximity, association and abundance relationships with fluid pathways (e.g. faults and lithological contacts) based on the concept of deformation-induced permeability being localized along such features. To empirically calibrate the abundance relationship between these features and gold mineralization, a measure known as geological complexity has been quantified. A fractal dimension representing the degree of geological complexity is evaluated using a box-counting method on the combination of faults and lithological contacts on a series of maps over the Kurnalpi Terrane of the Archaean Yilgarn Craton. Using multi-scale geological datasets from the Kurnalpi Terrane, we show that geological complexity and its relationship to known occurrences of orogenic gold mineralization varies depending on scale and degree of interpretation of the maps analyzed. In terms of scale, stronger correlations are observed for higher resolution ‘outcrop’ geological and structural maps. Even when the percentage of outcrop was taken into consideration, the higher resolution maps still showed greater correlations between geological complexity and gold mineralization. Differing solid geological interpretations at the same scale significantly affect the degree of quantified geological complexity and its correlation with known gold mineralization. The study results also illustrate a negative correlation between gold mineralization and gradients in the geological complexity. This outcome is contrary to results obtained by previous workers, and indicates that geological complexity, rather than complexity gradients, is useful as an input predictor map for prospectivity analysis and exploration targeting in the Kurnalpi region and other regions containing similar orogenic gold systems. Furthermore, the results indicate that interpreted solid geology maps, while critical for manual conceptual targeting, may actually be less effective than outcrop maps as a predictive layer in automated conceptual and empirical prospectivity analysis.