Publications

This paper introduces the problem of mining frequent integrated-geophysical data patterns and spatial association rules that are prevalent in spatio-temporal data. Due to the heterogeneous nature, large volume in size and file storage requirements, data patterns and trends hidden in these voluminous integrated data, non-trivial data mining and interpretation solutions are required. Data in different formats and domains have different dimensions, properties and attribute strengths. Conventional data mining and interpretation methods alone cannot discern data patterns and unlock the latent geological knowledge. Use of a single geophysical method of exploration and field development in any project may yield ambiguous results. Even after integration, geophysical anomalies and their trends are not well understood. Gravity, magnetic, seismic, electrical and electromagnetic datasets are typical geophysical domains, including their sub-surface geophysical data domain, when integrated for processing and interpretation of different geophysical anomalies in a single prospect domain can be effective for data mining and interpretation. Authors propose a novel approach to extract trends, correlations and patterns from different geophysical data domains, especially when they are interpreted with petroleum databases that are narrated by period and geographic dimensions using geo-ontologies representing prior knowledge. The aim of this study is to show how large amount of knowledge represented * School of Information Systems, Curtin Business School, Perth, WA, Australia ** MEDCO Energy *** Schlumberger in geo-ontologies, can be used to avoid the extraction of data patterns that are previously known to be “ambiguous”. Authors take advantage of the fact that voluminous natural geographic, spatial and periodic associations are inherently hidden in different geophysical data domains, including their associated data schemas and geo-ontologies. In spite of different geophysical properties of different domains, integrated domain ontologies generate correlatable data patterns and trends, when these data patterns are mined and interpreted, significant geological knowledge is discovered, with important economic leverage in terms of adding prospect volumes and drillable assets.