Information representation and integration of multiple sets of spatial geoscience data

Earth science and exploration data have characteristic features in both temporal and spatial space which are distinct on the terrestrial spectrum. These data sets can be either continuous or discrete. Recently there have been a number of studies based on information representation theory and AI/expert system techniques for processing and interpretation of multiple sensor spatial data sets with both global and regional coverage. There are in general two approaches in spatial data fusion and interpretation of the results; one based on specific rules, including geological and geophysical principles, and a more direct data driven approach. In the paper, some of the basic mathematical relationships often utilized in a data driven spatial information fusion methods are reviewed. Uncertainties in the hypotheses and experimental errors can also be handled and processed using same formalism, if the proper information representation technique is carefully chosen. To quantify the basic information contained in each data layer, and to accurately represent the information, it is important to understand statistical meaningfulness of each data set on the terrestrial spectrum for the final integration using currently popular GISs. In this paper, the basic concepts in preprocessing of geoscience data and information representation techniques, including the fuzzy logic, evidential belief function, and Bayesian probability, are discussed