Search for: All records

Typically, inversion algorithms assume that a forward model, which relates a source to its resulting measurements, is known and fixed. Using collected indirect measurements and the forward model, the goal becomes to recover the source. When the forward model is unknown, or imperfect, artifacts due to model mismatch occur in the recovery of the source. In this paper, we study the problem of blind inversion: solving an inverse problem with unknown or imperfect knowledge of the forward model parameters. We propose DeepGEM, a variational Expectation-Maximization (EM) framework that can be used to solve for the unknown parameters of the forward model in an unsupervised manner. DeepGEM makes use of a normalizing flow generative network to efficiently capture complex posterior distributions, which leads to more accurate evaluation of the source's posterior distribution used in EM. We showcase the effectiveness of our DeepGEM approach by achieving strong performance on the challenging problem of blind seismic tomography, where we significantly outperform the standard method used in seismology. We also demonstrate the generality of DeepGEM by applying it to a simple case of blind deconvolution. 

Seafloor geophysical instrumentation is challenging to deploy and maintain but critical for studying submarine earthquakes and Earth’s interior. Emerging fiber-optic sensing technologies that can leverage submarine telecommunication cables present an opportunity to fill the data gap. We successfully sensed seismic and water waves over a 10,000-kilometer-long submarine cable connecting Los Angeles, California, and Valparaiso, Chile, by monitoring the polarization of regular optical telecommunication channels. We detected multiple moderate-to-large earthquakes along the cable in the 10-millihertz to 5-hertz band. We also recorded pressure signals from ocean swells in the primary microseism band, implying the potential for tsunami sensing. Our method, because it does not require specialized equipment, laser sources, or dedicated fibers, is highly scalable for converting global submarine cables into continuous real-time earthquake and tsunami observatories. 

Buoyancy anomalies within Earth’s mantle create large convective currents that are thought to control the evolution of the lithosphere. While tectonic plate motions provide evidence for this relation, the mechanism by which mantle processes influence near-surface tectonics remains elusive. Here, we present an azimuthal anisotropy model for the Pacific Northwest crust that strongly correlates with high-velocity structures in the underlying mantle but shows no association with the regional mantle flow field. We suggest that the crustal anisotropy is decoupled from horizontal basal tractions and, instead, created by upper mantle vertical loading, which generates pressure gradients that drive channelized flow in the mid-lower crust. We then demonstrate the interplay between mantle heterogeneities and lithosphere dynamics by predicting the viscous crustal flow that is driven by local buoyancy sources within the upper mantle. Our findings reveal how mantle vertical load distribution can actively control crustal deformation on a scale of several hundred kilometers. 

Abstract Body‐to‐surface wave scattering, originated from strong lateral heterogeneity, has been observed and modeled for decades. Compared to body waves, scattered surface waves propagate along the Earth's surface with less energy loss and, thus, can be observed over a wider distance range. In this study, we utilize surface waves converted from teleseismicSHorSdiffwave incidence to map strong lateral heterogeneities across the entire contiguous United States. We apply array‐based phase coherence analysis to broadband waveforms recorded by the USArray Transportable Array and other permanent/temporary networks to detect coherent signals that are associated with body‐to‐surface wave scattering. We then locate the source of the scattering by back‐propagating the beamformed energy using both straight‐ray and curved‐ray approximations. Our results show that the distribution of scatterers correlates well with known geological features across the contiguous United States. Topographic/bathymetric relief along the continental slope off the Pacific Border is the major source of scattering in the western United States. On the other hand, sedimentary basins, especially their margins, are the dominant scatterers in the central United States. Moho offsets, such as the one around the periphery of the Colorado Plateau, are also a strong contributor to scattering, but isolating their effect from that of other near‐surface structures without any additional constraints can be complicated. Finally, we demonstrate the possibility of using scattered surface waves to constrain subsurface velocity structures, as complementary to conventional earthquake‐ or ambient noise‐based surface wave tomography.