Model-Based Three-Dimensional Correlation of Horizons in Seismic Data across Normal Faults
Final Report Abstract
Within the duration of the project we were able to show, how to introduce 3-d knowledge into the process of interpreting horizon continuation at faults. The resulting software includes major steps of horizon interpretation starting from an auto-tracker followed by a fault surface generator and ending with a horizon matcher. Introduction of 3-d knowledge into horizon matching has been proven to provide more accurate and more complete matches than if using 2-d slice information only. This has been validated by comparing different regions of the FOCADOS data set containing data of different quality. Introduction of a perception-based multi-resolution hierarchy in the matching process decreased computational cost for solving the matching task and also increased the robustness of the method with respect to noise and artefacts. Application of the system to various subsets of the FOCADOS data showed that - while not being perfect - horizon continuation across faults did produce similar results than the interactive reference interpretation by an expert geologist. Not all horizons may be matched but mismatches happened only when the data quality deteriorated or the fault throw did not follow the geological model (mostly because of interactions from nearby faults). However, the method has been investigated for shallow depths only and for regions that do not contain multiple faults. This was necessary in order to generate a good understanding of the roles of different parts of the underlying model. Future research will require automatic detection and classification of fault systems in order to integrate influences from fault interaction into the interpretation module. The extension to deep subsurface regions should be postponed until such fault system model exists. Modeling and counteracting effects from data distortion due to signal loss and deterioration is much more difficult and - in our opinion - requires that most other influences are much better understood.
Publications
-
F. Admasu, K. D. Tönnies. Exploiting 3d Spatial Continuity for Robust Automatic Horizon Matching across Faults. In IEEE Third International Symposium on 3D Processing, Visualization and Transmission, 111-114, Chapel Hill, USA, June 2006.
-
F. Admasu, K. D. Tönnies. Multi-scale bayesian based horizon matchings across faults in 3d seismic data. Lecture Notes in Computer Science, Vol. 4174 (DAGM-Symposium 2006), pp 384-393.
-
F. Admasu, K.D. Tönnies. An Approach towards automated fault interpretations in seismic data. Simulation und Visualisierung 2005, pp. 207-219, Magdeburg, March 2005.
-
F. Admasu, K.D. Tönnies. Anisotropie 3d seismic features for robust horizons correlation across faults. IEEE International Conference on Image Processing ICIP2005, volume 2, pp. 225-228, 11.-14. September 2005.
