Bayesian Data Fusion: a Reliable Approach for Descriptive Modeling of Ore Deposits

Abstract

Recognition of ore deposit genesis is still a controversial challenge for economic geologists. Here, this task was addressed by the virtue of Bayesian data fusion (BDF) implementing available proofs: semi-schematic examples with two (Cu and Pb + Zn) and three (Cu, Pb + Zn and Ag) evidences. The data, in current paper are just concentrations of indicated elements, were collected from Angouran's deposit in Iran at prospecting and general exploration stages. BDF was used for discrimination between three geneses of Massive Sulfide, Mississippi and SEDEX types. Better genesis recognition with clear discrimination between the geneses was achieved by BDF as compared with earlier studies. The results showed that uncertainties were reduced from 50% to less than 30% and deposit recognition was improved greatly. Furthermore, we believe that using more properties can have a beneficial effect on the overall outcome. The comparison made between 2 and 3 properties showed that the amount of probable belonging values to any type of deposit was greater in 3 properties. It was also confirmed that using the completed information from the various stages of exploration progress can be amplified and be used for genesis recognition via BDF.