Research Addresses Seismic Full-Waveform Inversion Using Multiples

Yingcai Zheng, associate professor in the University of Houston’s Department of Earth & Atmospheric Sciences, received word that a paper he co-authored was selected by the Society of Exploration Geophysicists as the Best Poster Paper Presented at the 2019 SEG Annual Meeting. The award will be presented at the Editor’s Reception during the 2020 SEG Annual Meeting in October.

The paper, Controlled-order multiple waveform inversion, was co-authored Zheng and lead authors Yike Liu and Bin He from the Institute of Geology and Geophysics, Chinese Academy of Sciences. The paper discusses a new algorithm to improve subsurface seismic imaging and reduce artifacts using full-waveform inversion of multiples.

The Best Poster Paper award was established in 1987 by SEG, an international academic organization present in 114 countries with more than 11,400 members.

Multiples, Deemed Harmful in Seismic Industry, Now Useful Signals

In marine seismic acquisition, a seismic wave can go up and down along a zig-zag path in the geological layers. Depending on the number of its interactions (0, 1,2, …) with the sea surface (or the free surface), a seismic wave can be classified as the 0^th (or primary), 1^st, 2^nd , … -order free surface multiples.

However, most modern seismic imaging algorithms treat all the paths as if they were just the 0^th order. For example, the recorded seismic data has the 1^st order multiple, but it would be nevertheless mapped/imaged incorrectly to create an image at the wrong location. This oversimplification creates imaging artifacts that can mislead geologists to yield an erroneous interpretation in understanding the petroleum system.

Paper’s Contribution

Recognizing that the 1^st order multiple (blue path + green path) was generated from the 0^th order (blue), the 2^nd order (blue + green + red) is from the 1^st order (blue + green), etc. The authors first separated the seismic data into datasets of different orders of multiples. They then use the 0^th order as the “source” to illuminate the subsurface to create the 1^st order data. A full-waveform inversion is applied to find an optimal subsurface model such that the interaction of the 0^th order multiple and the model creates the 1^st order data. Likewise, one can do the same for the pairs: the 1^st and 2^nd order data, 2^nd and 3^rd, etc. Multiples cover a wide range of reflection angles during wave propagation and offer the advantage of imaging the shadow zones that cannot be reached by primary reflections.

This image shows the FWI inverted Marmousi model using ONLY multiples (1^st, 2^nd, … order) from a linear-gradient starting model.

The paper has been formally accepted as a journal paper:
Liu, Y.K., Bin He, Y.Zheng (2020). Controlled-order multiple waveform inversion, Geophysics, 85 (3), R243-R250, dx.doi.org/10.1190/geo2019-0658.1

List of all winners of the Best Paper Poster award since its inception in 1987:
seg.org/About-SEG/Governance/Honors-and-Awards