By Norman Mark, norm_mark@hotmail.com
Using modern seismic interpretation procedures seismic data is loaded into software with 3D viewers in which the interpreters use built-in auto-picking software to transform seismic segments/horizons into surfaces. This is a time-consuming process which often takes months of mouse drawing and clicking.
Last year I published an article in this journal showing automation of the picking of all of the continuous segments longer than a few pixels within a 2D seismic data profile. This video presented here enables the viewer (the person, not the software) to see that the extension of my method to 3D works. It shows the algorithm in action.
The video opens with a text file of adjacent seismic crossline sections with 5 segments each being loaded into the left-side 3D viewer. The text file contains x-y coordinates of 5 continuous horizons in depth order for each of five seismic crosslines. Crosslines are separate and adjacent x-y planes with crossline number increasing along the z axis.
The crosslines are loaded one at a time into the viewer on the left side. The segments in the crossline closest to the origin are red and the segments in the crossline farthest from the origin are purple. Again, segments left in this format on-screen require months of manual and semi-automatic effort to be transformed into surfaces using currently available commercial software.
Caption: this video is a proof-of-concept and the method needs to be applied to a multi-gigabyte 3D seismic data volume to prove its validity.
In the viewer on the right the crossline segments are shown after having been transformed into surface segments. They appear in depth order and are red-orange for the shallowest and blue-purple for the deepest horizons. The horizon spheres have some transparency applied. Timeouts have been written into the code to help visualize the sequence of calculations which is extremely fast without the timeouts.
The key to the transformation from segments on vertical slices to segments within surfaces is a geometric transformation, my proprietary code. I wrote the viewers using JavaScript, ThreeJS and of course, HTML with some CSS.
