An alternative method based on region fusion to solve L0-norm constrained sparse seismic inversion

Abstract

L0-norm constrained sparse seismic inversion is an effective way to invert reflectivity series of underground rocks. A blocky impedance model with sharp formation boundaries can be estimated from sparse reflectivity series. At present, the most two common used methods to solve L0-norm constrained sparse seismic inversion problem are iterative hard thresholding (IHT) and matching pursuit (MP). IHT has low computational complexity and fast convergence rate, but low restoration precision. On the contrary, MP has high restoration precision, but high computational complexity and slow convergence rate. Hence, one needs much more computational resource to execute MP. An alternative and effective method based on region fusion (RF) is proposed to solve L0-norm constrained sparse seismic inversion problem. RF is a new descent strategy and converges fast while providing a good approximation of L0-norm with relatively high restoration precision. The numerical example demonstrates the effectiveness of RF method. The results show that RF has higher restoration precision than IHT and is significant faster than MP type methods. At last, a real seismic data example demonstrates the applicability of RF method in practice.

KEYWORDS:

• Sparse seismic inversion
• L0-norm
• region fusion
• blocky impedance

Acknowledgements

We are grateful to the reviewers for their constructive comments on this paper. This research is supported by the following funds: the Science and Technology Cooperation Project of CNPC-SWPU Innovation Alliance (No.2020CX010203), the Initiative Projects for Ph.D. in China West Normal University (No.19E063), the National Natural Science Foundation of China (no.41874146), and the National Natural Science Foundation of China (No. 41704134).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research is supported by the following funds: the Science and Technology Cooperation Project of CNPC-SWPU Innovation Alliance [grant number 2020CX010203], the Initiative Projects for Ph.D. in China West Normal University [grant number 19E063], the National Natural Science Foundation of China [grant number 41874146], and the National Natural Science Foundation of China [grant number 41704134].