A commonly used interpretation technique is to flatten data on horizons. This removes structure and allows the interpreter to see geological features as they were laid down. For instance, after flattening the seismic data, an interpreter can see an entire flood plain complete with meandering channels in one image.
Previously, in order to flatten seismic data, a horizon needed to be interpreted. If the structure was changing then many horizons needed to be interpreted. Here we propose a method for automatically flattening entire 3D seismic cubes without any interpretation at all. Our method involves first calculating dips everywhere in the data using a dip estimation technique described in Claerbout (1992). The local dips are resolved into a local travel time via a least squares problem that we solve in the Fourier domain. Then the data is shifted according to the travel times to output a flattened volume.
In this paper, we review the method for calculating dips and describe, in detail, the dip integration method. Then we show the results of several test cases. The first test case is a 3D synthetic data set with planes dipping at a single dip everywhere. The second test case is also a 3D synthetic data set, but has curved horizons. Finally, we apply this method to flatten horizons warped by a salt piercement in a real 3D seismic data set from the Gulf of Mexico.
This method successfully flattens the synthetic test data sets and removes a lot of deformation from the real test data set. These results are encouraging and invite more testing with more complicated models. The ability of this method to flatten data with faults still needs further development.