Thesis Defense – David Anthony (MS)

Possible Biases in the Amplitude of the Deconvolution-Based Ambient-Field Green’s Functions Revealed by Simulations in the Community Velocity Model for Southern California

David Anthony

David Anthony
MS Candidate
Advisor: Dr. Shuo Ma

Monday, December 19th, 2016
CSL 422 – 10:45 am

Abstract
We quantify possible biases in the amplitude of deconvolution-based ambient-field Green’s functions by simulating numerical Green’s functions in the Community Velocity Model CVM-S4.26 for southern California. In the microseismic band (5 – 10 s) remarkable waveform similarity is obtained among station-to-station Green’s functions obtained by deconvolution, correlation, and a finite-element (FE) approach using the CVM. However, large biases exist in the amplitude of the ambient-field Green’s functions, showing a strong azimuthal dependence. The deconvolution approach tends to overestimate the FE amplitude along the direction of noise propagation (approximately perpendicular to the coast), but generally underestimate the amplitude in other azimuths. Correlation generates similar amplitude to the deconvolution in a wide azimuthal range especially where stations are outside the basins. The large biases in the amplitude could be due to the inhomogeneous distribution of noise sources or in theory the deconvolution does not give rise to Green’s functions because the boundary condition at the virtual source is not an impulsive force. Carefully correcting these biases may allow the use of ambient-field Green’s functions in the virtual earthquake approach.

 
 
 

About the author

More posts by