Integrating a Hierarchical Process and Architectural Marginal Marine Classification with a Computer Database and Expert Systems—Toward Improved Subsurface Predictions

Vakarelov, Boyan; Ainsworth, Bruce

Marginal marine classification schemes have historically not directly dealt with the different scales of geobodies that build such systems. Classifications tend to be simplistic and presented as 2-D and pseudo 3-D diagrams, based on relationships between the depositional system categories and shoreline processes (wave, tide, fluvial), and/or grain size or mode of coastal migration. While relatively straight forward and easy to apply, such a classification approach is not always effective at predicting architecture in the subsurface. The depositional system categories used are too broad and are not well related to different scales of observation. They also often refer to scales that can be much greater than the scale of an individual reservoir. Geospatial databases based on such categories also tend to display significant spread of data points.

An alternative classification approach allows for much better integration with computer database environments and sets the framework for building marginal marine expert systems by permitting an element of prediction. The process and architectural marginal marine classification uses hierarchies of architectural units that are linked through Parent-Child relationships in a Tree data structure. Each hierarchy level applies to a different scale of observation, with units covering the full spectrum of reservoir heterogeneities (entire flow units, inter-reservoir sand bodies, and intra-reservoir barriers and baffles).

The definition of Parent-Child relationships between architectural unit categories offers great advantages over traditional classification approaches. Since there is always a finite number of parent-child relationships between individual architectural categories, a unit identified on one level can be related to all possible parents to such a unit on another level. The set of potential parent categories in this case can be thought of as uncertainty. The children of predicted parent categories will have a Sibling relationship with the initially identified unit. Predicting the types of siblings that can be associated with a given architectural unit is important as these can co-exist in the same stratigraphic interval and, yet, may not be directly sampled by available data points (e.g., cores or wireline logs). The process and architectural marginal marine classification framework has been successfully integrated with a geospatial database and expert system software package that is currently under development.

AAPG Search and Discovery Article #90163©2013AAPG 2013 Annual Convention and Exhibition, Pittsburgh, Pennsylvania, May 19-22, 2013