← Back to Sessions

Multi-Source Wavefield Reconstruction Combining Interferometry and Compressive Sensing

Session: Advances in Seismic Interferometry: Theory, Computation and Applications

Type: Oral

Date: 4/22/2021

Presentation Time: 05:45 PM Pacific

Description: 

Seismic interferometry makes it possible to determine the waves that propagate between seismometers, any one of which can act as a virtual source for the other receivers. These wavefields can then be used for tomography and imaging. However, high resolution processing and interpretation in seismic interferometry typically require seismic profiles acquired from dense and regular seismic arrays. In practice, dense and regular seismic arrays cannot always be used due to restricted receiver installation, receiver malfunction, and low signal-to-noise ratio (SNR) records in some receivers. Thus, regularization and reconstruction methods such as compressive sensing are required for incomplete and irregular seismic profiles.

We propose a joint technique called multi-source reconstruction. The method cross-correlates signals of all available virtual sources before reconstructing the missing interferometric signals using compressive sensing. We show a numerical example using a line array with different numbers of available receivers. Using SNR as a comparison, our joint method statistically improves the reconstruction of interferometric signals by approximately 5 dB (without additive noise) and 15 dB (with additive noise), compared to the reconstruction when the compressive sensing is performed before interferometry.

We also compare our technique to a conventional method, called single-source reconstruction, where cross-correlation interferometry of only one virtual source is performed before the compressive sensing reconstruction. Our reconstruction uses Fourier and Curvelet as the sparse transforms. The Fourier multi-source reconstruction considerably improves the reconstruction quality at traces close to the virtual sources, compared to the Fourier and Curvelet single-source reconstruction. The total SNR of the Fourier multi-source reconstruction also improves by 20 dB and 15 dB at 20% and 40% of available receivers along the line array, respectively.

Presenting Author: Patipan Saengduean

Student Presenter: Yes

Authors