Automatic event picking in prestack migrated gathers using a probabilisticneural network

We describe algorithms for automating the process of picking seismic events in prestack migrated common depth image gathers. The approach uses supervi sed learning and statistical classification algorithms along with advanced signal/image processing algorithms. No model assumption is made, such as hy perbolic moveout. We train a probabilistic neural network for voxel classif ication using event times, subsurface points, and offsets (ground truth inf ormation) picked manually by expert interpreters. The key to success is usi ng effective features that capture the important behavior of the measured s ignals. We test a variety of features calculated in a local neighborhood ab out the voxel under analysis. Selection algorithms ensure that we use only the features that maximize class separability. This event-picking algorithm has the potential to reduce significantly the cycle time and cost of 3-D p restack depth migration while making the velocity model inversion more robu st.