Automated 3-D Horizon Tracking and Seismic Classification Using Artificial Neural Networks

Seismic surveys are routinely carried out in three dimensions, resulting in large volumes of high-resolution seismic data and a corresponding increase in the workload of an interpreter. An automatic tracker is described in this chapter, based on artificial neural networks (ANNs), which enables horizons to be tracked in three dimensions with less input from an interpreter compared to most commercial automatic trackers. More time can therefore be spent by the interpreter investigating geologically complex areas. A hybrid ANN is employed which combines both unsupervised (self-organising feature map) and supervised (multilayer perceptron) network paradigms. The tracker is demonstrated on a real three-dimensional (3-D) seismic data set, and is shown to be a viable technique for use as a standard tool and for enhancing efficiency in 3-D seismic interpretation. 1.5-D and 2-D methods have also been demonstrated successfully, which account for the seismic character above, below, behind and ahead of the current tracking position.