|
The Skellefte region, in northern Sweden, consists mainly of Svecofennian supracrustal sequences and associated intrusive rocks. It has been regarded as one of the largest early Proterozoic VMS districts in the world. For VMS prospectivity mapping in the region, we applied weights-of-evidence (WofE) modeling (Bonham-Carter et al., 1989) to (a) determine useful spatial evidences (i.e., having positive spatial association with VMS deposits) and (b) create a predictive VMS prospectivity map. WofE involves updating of prior probability of deposit occurrence using conditional probability of deposit occurrence given presence/absence of evidence to derive posterior probability of deposit occurrence. Two conditional probabilities of deposit occurrence are estimated, W+ and W-, given presence and absence, respectively, of evidence. Predictor maps are integrated with prior probability of deposit occurrence under assumption of conditionally independence (CI) with respect to deposits.
Regional-scale features initially considered for VMS prospectivity mapping were: bedrock geology; contact between Skellefte Group (SG) and overlying Vargfors Group (VG); shear zones; geochemical anomalies; heat sources; total magnetic intensity; and airborne radiometry. We used 66 VMS occurrences (closed mines, prospects) for training and five VMS occurrences (active mines) for validation. The evidences found useful (i.e., based on studentized (C) for regional-scale VMS prospectivity mapping are: three SG members; proximity to SG-VG contact; proximity to NW-, WNW- and ENE-trending shear zones; >70th percentile Cu values and >50th percentile Zn values in till samples; proximity to heat sources; and <40th percentile K/Th ratios of airborne radiometric data.
Results of WofE calculations were used to create binary predictor maps. The three SG members were aggregated into one binary map and the weights were recalculated. Binary predictor maps of proximity to WNW- and ENE trending shear zones were combined into one binary predictor map because they violate CI with respect to deposits. Evaluation of performance of the posterior probability map (Agterberg and Bonham-Carter, 2005) indicates that, if 10% of the area is considered exploration targets, based on the upper 10 percentile of posterior probabilities, then there is at least 92% chance of finding an undiscovered VMS deposit. All the five active VMS mines, which were assumed undiscovered deposits, are delineated within <50% of the exploration targets.
|
