A knowledge-data framework and geospatial fuzzy logic-based approach to model and predict structural complexity

Abstract Prediction of structural complexity for geohazard and subsurface resource applications requires constraining and interpreting data that are often ambiguous or lack key information. Moreover, structural complexity is a subjective term, requiring context for quantification. Recognizing this, a new knowledge-data framework and a geospatial fuzzy logic method is developed to represent and predict structural complexity in the subsurface. A conceptual model for known structural complexity serves as a basis for associating geospatial representations with types of damage zones. A second conceptual model for zones of structural complexity facilitates its prediction, notably in areas with limited explicit structural data. For each conceptual model, a fuzzy logic inference model is developed to incorporate geospatial data and estimate structural complexity potential. This approach is demonstrated using several public geospatial datasets within the state of Oklahoma. Explicit fault and earthquake location data were integrated using a fuzzy model of known structural complexity to train topographic, lithologic, and geophysical proxy datasets, applied to a fuzzy model to predict structural complexity, and evaluated with Receiver Operating Characteristic analyses and error classification. The final model output, displayed in maps and cross sections, offers comparison with interpreted structural data for validation. Together, these results demonstrate the effectiveness and limitations of the new approach as a screening tool for predicting structurally complex areas.