next up previous [pdf]

Next: Conclusions Up: Fomel: Predictive painting Previous: Predictive painting in 2-D

Predictive painting in 3-D

The challenge of predictive flattening in 3-D is in selecting a recursive path that the reference trace should follow to paint its neighbors. For choosing this path, I adopt a version of Dijkstra's shortest path algorithm (Dijkstra, 1959; Cormen et al., 2001). Dijkstra's algorithm finds the path between two nodes in a network of nodes, where there is a cost associated in connecting each node with its neighbors. In our case, the nodes are seismic traces in a 3-D cube. I use the semblance between neighboring traces as a cost function. Dijkstra's algorithm finds the shortest (minimum-cost) path by effectively arranging all nodes in a sequence from low to high cost and evaluating each new node using the information from previous nodes. I run the shortest-path algorithm starting from the reference trace and paint other traces in a recursive sequence using the information from previously painted traces. Using semblance as a cost function helps avoiding 3-D misties by forcing the shortest path to go around possible fault areas.

The 3-D data test is reproduced from Lomask et al. (2006). It uses a portion of a depth-migrated 3-D image with structural folding and angular unconformities (Figure 5(a)). Inline and crossline dips are measured automatically from the image using plane-wave destruction (Figures 6(a) and 6(b)). Figure 7(a) shows painting of individual strong horizons in the volume. Figure 5(b) displays automatic flattening using predictive painting of the relative geologic age. Figure 7(b) shows some of the corresponding equal-relative-age horizons displayed on top of the original image. Predictive painting is able to correctly identify the most significant three-dimensional structural features.

win wflat
win,wflat
Figure 5.
A North Sea image from Lomask et al. (2006) (a) and its predictive flattening (b).
[pdf] [pdf] [png] [png] [scons]

wdip1 wdip2
wdip1,wdip2
Figure 6.
Inline (a) and crossline (b) slopes in the North Sea Image estimated by plane-wave destruction. Blue colors indicate negative slope; red colors, positive slope.
[pdf] [pdf] [png] [png] [scons]

wpaint wcont
wpaint,wcont
Figure 7.
Predictive painting (a) and automatic picking (b) of major horizons in the North Sea Image.
[pdf] [pdf] [png] [png] [scons]


next up previous [pdf]

Next: Conclusions Up: Fomel: Predictive painting Previous: Predictive painting in 2-D

2013-03-02