3D Field Data Example

We also test the proposed method on a 3D dataset. Figure 20a is the Teapot Dome seismic dataset. After inline and crossline slopes (Figure 21) are estimated, the RT volume (Figure 20b) is easily obtained from the predictive painting with multiple reference traces. With the RT volume, the 3D seislet transform can be constructed by cascading the 2D seislet transform along inline and crossline directions. To be specific, we first apply the 2D RT-seislet transform along inline direction, and then another 2D transform along crossline is applied on the coefficients from the previous step to obtain the coefficients in the 3D transform domain. The coefficients of the 2D seislet transform along inline direction and the 3D seislet transform are shown in Figure 22a and 22b. In the seislet domain, most of the coefficients concentrate in the coarse scales, which shows the excellent compression ability of the RT-seislet transform. Figure 22c is the data reconstruction result using only 5% of the most significant coefficients. The difference between Figure 20a and the reconstructed data, contains mainly noise, is shown in Figure 22d. The proposed method has good performance for reconstruction and preserving the structures of the dataset.

cuber,pick
Figure 20. (a) Teapot dome dataset. (b) RT volume estimated by the predictive painting.

dip1,dip2
Figure 21. Inline (a) and Crossline (b) Dip.

rtseis-inline,rtseis,teapot-rtseisrec5,teapot-diff
Figure 22. (a) The coefficients in 2D seislet domain along inline only. (b) The coefficients in 3D seislet domain. (c) Data reconstruction using only 5% of significant coefficients by the inverse RT-seislet transform. (d) Difference between (c) and the original seismic volume (Figure 20a).