An optimized decision model for area selection in massive sulphide exploration

CIM Bulletin, Vol. 72, No. 804, 1979

G. FAVINI Consultant, Quebec, Que., ROBERT ASSAD Professor, University Laval, Quebec, Que.

A Bayesian decision model of minimum risk, based on the probability and costs of mine discoveries and non-economic projects, is presented for massive sulphide exploration in the Canadian Shield. The sample used for the model is composed of data relating to forty mines, sixty non-economic sulphide occurrences, and sixty occurrences of graphitic and pyritic material. These economic and barren samples consist of drilltested occurrences which were, in varying degree, characterized by conductivity anomalies.Given the large size, the reliability and the homogeneity of the data base available for massive sulphide exploration, it is shown that certain statistical models can and should be utilized as a tool for decision-making by explorationists. Given this data base, information theory tells us that only those signals that are independent of one another will transmit information with minimum error. An ensemble of six significant multilevel signals, markedly independent of one another and consisting of numeric and geometric form descriptions derived from Bouguer anomaly, aeromagnetic and physiographic maps, were identified and evaluated by case studies. The use of these signals as input in a simple statistical procedure, or in a carefully planned and experienced visual appreciation scheme, can be of significant aid in decisions concerning the choice of favourable areas or sectors to explore in detail. The method of discriminant analysis is theoretically and practically suited for the statistical modelling involved. By way of a simulation on our statistical model, the efficiency of the model and of the underlying criteria is determined for historic cases and, by inference, for its validity in future exploration. An approximate optimized solution is derived from the model which suggests that over 90% of the known ore could be found by screening, according to a model such as that herein described, and drilling only 30% of the targets (mostly barren) normally identified and tested in current and past exploration.
Keywords: Exploration techniques, Mineral exploration, Sulphide deposits, Massive sulphides, Models, Decisional models, Bayesian models, Precambrian Shield, Discriminant analysis, Geostatistics.