Salt C3 Dataset
There are several data sets available at the SEG/EAGE 3D Modeling Data Access Site and the Salt Model Phase-C Classic Datasets (C3)was obtained through Golden Geophysical. The data set comes in two azimuths: narrow azimuth (NA) and wide azimuth (WA). Below is some basic survey information.
In the following two gif images are the placement of the shots over the salt structure in the SEG/EAGE Salt Model. On the left is the C3-WA, and in the center is the C3-NA. The C3-NA and C3-WA datasets are very similar except for the streamer spacing. The C3-NA dataset has approximately twice as many shots as the C3-WA dataset. The red structure is the salt within the velocity model and the blue points present the shot locations. The ten horizontal and vertical lines are image planes which are shown below. They include x = 5060 m, 6760 m, 7400 m, 8440 m, 10140m, and y = 1680 m, 3380 m, 5060 m, 6760 m, 8440 m.
Shot Locations Shot Locations Migration Volumes
On the right are various volumes of the migration. The red box indicates the extent of the velocity model. For our migration, we used the same velocity model used for the forward modeling. This test is interested in the performance migration algorithm, and not the velocity model building aspects. The green box indicates the volume of the final stacked image. Although the image volume extents to x < 4000m and y > 8000m, very few shots are imaging this region and beneath the salt structure from this direction.
The blue box is a standard migration volume imaged by one shot, and the shot location is at the center of the box. This migration volume was selected to match the forward modeling, but may not have a large enough aperture to image beneath the salt structure. The forward modeling used a 321x321x210 grid point cube to produce the synthetic data. Around the outside edge of this domain, a 15-20 grid point sponge region was applied to eliminate reflections from the boundaries. This reduced the useful imaging domain to approximately 271x271x210, which is what we used to produce the following images. This domain size along with the 5 sec trace length brings up the question - Have the layers beneath the salt structure been illuminated and was enough trace data recorded so that imaging can occur?
These images match the planes indicates in the pictures above. They include x = 5060 m, 6760 m, 7400 m, 8440 m, 10140m, and y = 1680 m, 3380 m, 5060 m, 6760 m, 8440 m. These images are for the C3-WA dataset. On the top half of each gif is the migrated image, and on the bottom half is the corresponding velocity model. The images for the C3-NA dataset are very similar. So similar that one may question if the C3-NA really provides any additional information. See the cut aways below and make your comparison. As can be seen below, portions of the flat reference layer have been illuminated, but many subsalt features have not been imaged. These subsalt features are dipping which requires larger apertures to image them.
Below are the cutaways for the C3-WA dataset (left), C3-NA dataset (center), and the velocity model (right). The intensities of the C3-WA and C3-NA images have been adjusted to allow for easy comparison.
The cutaway surfaces are x = 3520 m, 7400 m; y = 1880 m, 7700 m; and z = 260 m, 3500 m. At z = 3500 m, the flat reference layer can be seen. The black regions indicate that the flat reference layer has been illuminated. However, the gray portions at z = 3500 m indicate regions without illumination and mostly occur near the northwesterly (toward x = 0 and y = 0) plunging stock.
Compare one of the above cutaways from the C3 dataset with the cutaway from our defined acquisition geometry. These images are very similar. There are two major differences in the generation of these images: the acquisition geometry and the computational costs. The C3 datasets are classical streamer acquisition geometries which contained thousands of shots (~2400 shots for the C3-WA dataset and ~4800 shots for the C3-NA dataset). Our defined acquisition geometry closely matches ocean bottom cable (OBC) acquistion geometries, and contains only 45 shots. However the computational costs associated with each geometry are not as similar as the images. The costs on a per shot basis (we are using shot-record, prestack, f-x migration) are nearly equal for both the C3 and our defined datasets, since they use the same migration volume. Therefore the C3-WA image costs about 50 times more than our defined dataset image, and the C3-NA image about 100 times more based on the sheer number of shots (migrations).
Movie showing a cutting plane passing through the SEG/EAGE Salt model
MPEG Video (3 MB)
Movie showing stacking of image plane x=6760 m of the SEG/EAGE Salt
MPEG Video (2 MB)
Curtis C. Ober
Last modified: October 9, 1998
Back to top of page || Back to Seismic Imaging Home Page || Sandia Home Page
Questions and Comments || Acknowledgment and Disclaimer