Integration of multi-source data in mineral exploration

This paper describes several multivariate statistical analysis applications of geochemical, geophysical, and spectral variables in mineral exploration. Mahalanobis' distance is described in some detail and based on four multisource variables this measure is applied to produce a map that gives an expression of the statistical proximity of each point in the map to a mineralized area. The four multisource variables chosen from a much larger set of variables have all been subject to extensive data processing: the geochemical variable is the noise MAF (minimum/maximum autocorrelation factor) of eleven kriging interpolated stream sediment variables; the geophysical variables are kriged aeromagnetic data iteratively moving average corrected to minimize the flight line striping and kriged Bouguer gravity anomaly data corrected for a quadratic trend; and the spectral variable is the density of automatically generated linear features based on Landsat TM data. The results indicate among other things a not previously recognized subsurface continuation of an already mapped lineament.