Using 3D Methods in the Management of Risk in Exploration Targeting

The surface geochemistry techniques widely employed during the 1960’s to 1980’s are effective at discovering near-surface deposits, but are largely ineffective for exploring under deep cover or for deep extensions of known mineralisation in a near-mine environment. Instead, mineral explorers must focus their attention on rapidly aggregating and synthesising disparate data to build into holistic mineralisation models that can be interrogated within the framework of 3D models. For example, the Far East Zone at Red Lake in Ontario was targeted through 3D geological modelling, which identified a previously unrecognised fault that off-set mineralisation by approximately 500m in a right-lateral sense. In addition to the traditional empirical approach to targeting, automated methods such as weights of evidence, neural networks and probabilistic techniques are also being used. However, the extensive use of Bayesian statistics (eg weights of evidence) places reliance on a technique that can introduce false positives and false negatives that reduce the chances of success. In an attempt to improve targeting success, the minerals industry is increasingly looking to other industries that have tackled this problem, in particular, the oil industry which is dealing with similar parameter space. The oil industry has seen a marked increase in exploration success by assigning risk to the variables required for an oil or gas resource to be present; e.g. probabilities of the presence of source, maturation, pathway, reservoir and trap. Crude oil exploration is however dealing with a single type of deposit represented by a simple system with relatively few variables, laminar fluid flow, and relatively simple basin architecture and structure in the shallow crust. In addition, the location and form of oil deposits is controlled by the current structural geometry in the crust making it much easier to determine if a trap is present. In minerals exploration, different models are necessary to understand each type of ore deposit, and many of these models (e.g. for IOCG deposits) are not well understood. Ore deposits commonly form in complex structural environments, from highly pressurized fluids sourced from the deep crust. Many ore deposits are also long lived and have experienced significant post-deposition deformation and overprinting obscuring an understanding of the original factors that led to ore deposit formation. The coincident set of events and variables required to form a mineral deposit are considerably more complex and challenging than those involved in oil accumulation. Geoinformatics Exploration is applying a probabilistic approach in the search for ore deposits such as porphyry copper-gold systems. In porphyry copper-gold environments the variables required to form an ore deposit are a fertile source region, melting of the source region to generate hydrous and metalliferous magmas, migration pathways for the magma, and a trap to stop the magma’s ascent at an appropriate depth to allow metal precipitation. Geology and lithochemistry of volcanic and intrusive rocks in a region can be used to determine whether prospective hydrous magmas were generated. Migration pathways are likely to be major structures that could be located using upward continued geophysical methods that can be used to “see” deep into the crust. The best evidence for a trap is often the presence of a porphyritic intrusion of appropriate age, which can commonly be located under cover using magnetic or gravity data. Knowledge of the regional geology and the palaeosurface can be used to estimate if the depth of emplacement is in the pressure range for porphyry copper formation. These data sets can be used to estimate the probability of all the necessary factors coinciding at any point in a project area. This provides an estimate for the probability of finding an ore deposit assuming the model being tested is correct. For prediction to have a spatial component that can be used in looking for buried targets, one of the main inputs will be the 3D model and the accuracy and validity of all data used as inputs must be known and recorded.