New Ideas in Seismic Imaging 

Technological innovation in extracting and refining petroleum has given us a set of sophisticated tools that work under a wide range of challenging reservoir conditions. But even the best extraction technology depends upon an essential first condition:

We must know where the oil is.

The days of randomly drilling wells in hopes of making a big strike are ancient history. Oil exploration and production rely upon seismic techniques that can “see” underground and predict reservoir structures and physical characteristics. Unfortunately, limitations in focus quality meant that seismic reservoir images were often expensively wrong.

Theoretical Breakthrough

A seismic survey is like a stereo camera, with one very important difference: Software plays the role of the camera’s lens, combining and focusing the huge volumes of data collected by seismic instruments into images of the earth’s subsurface. While automated focusing is essential, it remained a long-standing technical challenge. A solution came through advanced applied mathematics.

Developed at Rice, differential semblance optimization (DSO) focuses seismic data in a way similar to split-image focusing in an ordinary camera: Various parts of the data produce independent images, and the overall earth structure is correct when these independent images align.

This theoretical breakthrough came years ahead of the necessary computational power. In 1992, The Rice Inversion Project (TRIP) embarked on a mission to accelerate DSO’s performance and document its potential for accurate subsurface imaging. TRIP provided the computational and applied mathematics advances needed to make DSO feasible. But realizing its full potential requires more than mathematics.

 

[image]

(above) Marine seismic data that depicts the structural image of sub-seafloor properties. The image on the left renders an image based solely upon regional geologic information. The depicted strata, clearly identifiable at the sides of the image, become blurred and unintelligible at the center of the image, especially at the lower depths (roughly 6 km). In the right-side image, the same data is enhanced using the differential semblance optimization technique, a Rice innovation, to produce dramatically clearer focus near the center.

 [image]

(below) The Chevron Visualization Laboratory’s Visualization Wall measures 14x8 feet and renders stereoscopic 3D images using more than 33 million pixels powered by 2,034 processor cores. Rice researchers have applied full waveform tomography techniques on a continental scale to create a detailed seismic model of the substrata in China and the neighboring region. Our research – like the energy challenge – spans the globe.