A hybrid of neural net and branch and bound search techniques for seismic horizon tracking is described in this paper and its results on real as well as synthetic data are shown. Seismic horizon tracking is tracking of reflections from a subsurface bed boundary in processed seismic data. Accurately tracking horizons is important for hydrocarbon exploration. Automatic tracking of horizons in processed seismic data is difficult because of noise, nonlhwxity of the horizon shape, and discontinuities in the horizon. The hybrid technique works better than existing commercial techniques and is able to track horizons in the presence of significant discontirtuities. The key steps of the technique are: (1) rate peaks in seismic data using a neural ne~ (2) link peaks into segments using a heuristic scheme, and (3) identify segments of the horizon of interest using the branch and bound technique. ‘fltis approach allows us to integrate seismic attribute information, spatial information, and geological constraints for tracking seismic horizons.
[1]
Nils J. Nilsson,et al.
Principles of Artificial Intelligence
,
1980,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[2]
M. Taner,et al.
Complex seismic trace analysis
,
1979
.
[3]
James L. McClelland,et al.
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations
,
1986
.
[4]
Don Wagner,et al.
8. First Break Picking Using a Neural Network
,
1991
.
[5]
Franklin Kemp,et al.
A Neural Net Branch And Bound Seismic Horizon Tracker
,
1992
.
[6]
Richard O. Duda,et al.
Pattern classification and scene analysis
,
1974,
A Wiley-Interscience publication.
[7]
M. D. McCormack.
Seismic Trace Editing And First Break Picking Using Neural Networks
,
1990
.