Abstract

The collection of low-amplitude seismic waves generated by persistent sources such as human activity and oceans is called seismic ambient noise. Correlations of ambient noise often contain coherent seismic arrivals, which makes them useful for imaging the subsurface. This dissertation presents three studies concerning the full-waveform and traveltime inversion of noise correlations. 

The first study challenges a common assumption used in ambient noise inversion. While the spatial distribution of noise sources is usually uneven, seismologists often assume that noise sources are uniformly distributed to simplify noise correlation inversion. In this work, I conduct numerical experiments to study how this assumption impacts subsurface models estimated from noise correlation waveforms. My results show that if the noise source distribution is deemed uniform, the misfit of the estimated models increases. This indicates that the assumption of uniform noise sources introduces source-dependent model errors. Thus, to avoid these errors, it is necessary to account for the noise source distribution. 

The second study presents a traveltime tomography of the Zabargad Fracture Zone, northern Red Sea. Using ambient noise recorded by a new seismic network, I compute noise correlations from which I measure traveltimes of Rayleigh and Love waves. Then, I invert traveltime measurements using a Bayesian inference approach to estimate a 3-D shear-wave velocity model. For the first time, my model shows the shallow crustal structure of the Zabargad Fracture Zone. The estimated model reveals a salt layer and basement with thicknesses of 2 and 4 km, respectively. Also, it shows a high-velocity gradient between 7 to 8 km, which agrees with the expected Moho depth in the region. 

The last study investigates the origin of a spurious arrival observed in the noise correlations from the second study of this dissertation. Spurious arrivals are seismic waves that are not observed in traditional observables like earthquake waveforms. They originate in different manners, making their use in seismic tomography challenging. Understanding the origin of spurious arrivals is relevant, as this will open the opportunity to use them in subsurface studies. Here, I conduct full-waveform modeling to determine the origin of spurious arrivals. My preliminary analysis indicates that these arrivals are P waves generated near the location of the stations.