Integration and fusion of geological exploration data: a theoretical review of fuzzy logic approach

Abstract

A geological survey project, including the exploration of nonrenewable resources, typically includes the following steps: planning of the survey, field mapping, acquisition of data, achieving of data, processing and fusion of all relevant information, which are closely followed by an interpretation of the final survey results. In this process, important spatial data layers are topographic map, often digital elevation model (DEM), geological map, several sets of geochemical and geophysical survey maps, airborne or space-borne remote sensing data and, sometimes, old archived data. Each of these multiple layers of geological exploration data can first be preprocessed, and input into a chosen geographic information systems (GISs). This will be followed by information representation and fusion steps for the final imaging of the processed and fused information. Most commercially available GISs however do not have information fusion and focusing capabilities. Users of the slected GIS must first be able to represent the preprocessed geological survey information with respect to the target geological hypothesis. In the case of mineral exploration, the target hypothesis can be a specific mineral deposit(s) being explored.

Compared to the model driven exploration approach, the data driven geological exploration utilizes multiple sets of exploration data and there are several mathematical tools available for information representation and fusion. Some of these include the traditional probability approach, evidential belief function methods, AI/Expert systems and the fuzzy logic approach.

In this paper, the fuzzy logic approach of quantifying the exploration information with respect to the target hypothesis, and several of the fuzzy operators which are frequently used for geological resource exploration are critically reviewed. Although our understanding of the fuzzy information representation and fuzzy operators is still incomplete, many case studies of applying fuzzy logic approaches to various exploration projects have concluded that the final fuzzy membership function maps or the final fuzzy theme maps are considerably more accurate than the results obtained using any conventional intuitive approach, including the brute stack of the exploration (spatial) data using a GIS. In general, the efficiency and accuracy of the final fused information with respect to the target hypothesis increases with the increasing number of geological exploration data layers.

If a specific Earth system model is available, or a mineral deposit model in the case of mineral exploration projects, is available or is known a priori, fuzzy logic approach can easily be combined with currently popular machine-reasoning processes such as the neural network approach. In this type of quantitative spatial reasoning and fuzzy logic information fusion, uncertainty and error analysis is also important. In most cases, propagation of errors and uncertainties can be handled and estimated in the same manner as the processing and fusion of main exploration data.