← Back to Gallery

Session: Building, Using and Validating 3D Geophysical Models [Poster]

Type: Poster

Date: 4/25/2019

Time: 06:00 PM

Room: Grand Ballroom

Ambient Noise Empirical Green’s Function Full Waveform Tomography for the Northern Mississippi Embayment

Ambient noise data recorded by 244 broadband seismic stations within the Northern Mississippi Embayment (NME, 86^oW-94^oW, 33^oN-38^oN) over the time period of 1990 to 2018 are used to extract empirical Green’s functions (EGFs) by seismic noise interferometry. The EGFs between pairs of seismographs are estimated from the long-time cross-correlation of ambient noise. With the traditional time-domain normalization to remove transient noise and earthquakes, the signals in the cross-correlation will inevitably loose amplitude information and small earthquakes could get through the procedure and hide in the data. However, the amplitude of signals in the cross-correlations are useful for direct source location, attenuation analysis, differentiating sources and small earthquakes in the data could cause confused high amplitude signals around 0 delay time in the cross-correlations. To avoid these flaws of the time-domain normalization method, we apply a new hybrid seismic de-noising method using wavelet block thresholding to remove earthquakes and other signals. Rayleigh wave group velocity dispersion curves are then estimated from the EGFs using frequency-time analysis and used to invert for a 1D velocity model at each station. Starting with this initial model, the EGFs are next utilized as waveform data for 3D full waveform tomography to acquire the final 3D model for NME. A CUDA-enable collocated grid finite difference code is used to do the waveform modeling and inversion. The final 3D model illuminates several geological and tectonic features such as Reelfoot Rift and is useful for future seismic hazard analysis and seismic source determinations.

Presenting Author: Yang Yang

Additional Authors