Loading...
Thumbnail Image
Publication

Wavefield reconstruction using seismic interferometry and compressive sensing

Saengduean, Patipan
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Seismic data acquisition and processing are essential steps for seismic exploration, the determination of deep earth structure, and subsurface monitoring. Common problems in seismic data acquisition are missing or unusable receivers, and a low signal-to-noise ratio of recorded signals. These problems apply to seismic interferometry, a passive-source technique to estimate inter-receiver wavefields. In this thesis, I propose novel methods that use seismic interferometry and compressive sensing to mitigate these problems. I use inter-source interferometry to estimate body waves that propagate between two earthquakes from waves that are recorded at a surface receiver array. Inter-source interferometry, which turns an earthquake into a virtual receiver, allows receivers to be virtually placed at earthquake locations (e.g. inside volcanoes and subduction zones) that are normally inaccessible for receiver installation. To accurately estimate inter- source body wavefields, I derive a criterion for how densely receivers must be spaced to retrieve the waves that propagate between earthquakes. Using waves recorded near the San Andreas Fault, I estimate the inter-earthquake wavefields. The problem of missing or unusable receivers is important for receiver arrays that do not regularly sample the wavefield at the surface. I propose a compressive-sensing-based multi-source wavefield reconstruction to alleviate the problem of missing or unusable receivers. Using the Fourier and Curvelet domains for sparse transforms, I show that a multi-source method, which reconstructs correlated wavefields at the locations of unavailable seismometers from correlograms of all available virtual sources, improves wavefield recovery compared to single-source reconstruction, which uses a correlogram from a single virtual source. I show successful applications of multi-source reconstruction for wavefield-recovery improvement over single-source reconstruction using synthetic data on linear and areal arrays and present a data example using distributed acoustic sensing data. The methodology is, in principle, applicable to active source seismic surveys. Also, I develop weighted compressive sensing to mitigate the imprint of noise on the reconstructed wavefield, and to impose a priori information about the nature of the recovered wavefields. Wavefields obtained from seismic interferometry may contain spurious arrivals. Using the Fourier basis as a sparse transform, I show that weighted compressive sensing can suppress these spurious arrivals.
Associated Publications
Rights
Copyright of the original work is retained by the author.
Embedded videos