1. Near surface imaging and monitoring with fiber-optic sensors

Subsurface imaging and monitoring at high spatial and temporal resolution is important for assessing the sustainability and potential seismic hazards of geological structures. However, it is usually costly and logistically prohibited, especially in urban settings and harsh environments. Fortunately, in recent years, solutions have been proposed in terms of both seismic sources and receivers. For seismic sources, the well-established ambient noise interferometry uses ubiquitous long-period natural vibrations or high-frequency anthropogenic noise to turn receivers into virtual sources. For receivers, more recently, the emerging technique of distributed acoustic sensing (DAS) repurposes telecommunicate fibers into meter-spacing strain sensors, usually covering urban and offshore areas. With the advance in these cost-effective seismic sources and receivers, we are able to achieve subsurface imaging and monitoring at unprecedent resolution.

1.1 Shallow subsurface imaging explains earthquake ground shaking variability

Earthquake ground motion depends strongly on near-surface structure, which is challenging to image in urban areas at high resolution. We converted an underground dark fiber across the city of Ridgecrest, CA, into a DAS array and recorded aftershocks with substantial lateral variability in shaking intensity over only 8-km in distance. We use traffic-generated seismic noise to image near-surface structure along the fiber path, which can readily explain the lateral variations of earthquake shaking at sub-kilometer scales. The results highlight the great potential of DAS for high-resolution seismic hazard mapping in urban areas. Relevant publications: GRL paper EOS Research Highlight

1.2 Scattered waves detect and characterize hidden faults

Fault zone complexities contain important information about factors controlling earthquake dynamic rupture. It is challenging to identify small hidden faults in the absence of surface evidence or cataloged seismicity. We show how we identify and characterize the fault zones at shallow depth using scattering phases captured from traffic-generated seismic noise along a DAS cable in Ridgecrest City, CA. The results demonstrate the potential of DAS in fine-scale fault imaging without needing earthquakes. Relevant publications: JGR paper 1, JGR paper 2

1.3 Seismic velocity changes capture the vadose zone soil moisture dynamics

Vadose zone soil moisture dynamics in semi-arid regions are critical to ecosystems and water management, but their direct observation at large spatiotemporal scale remains challenging. Utilizing pre-existing fiber-optic cables as seismic sensors, we demonstrate a fiber seismic sensing principle to robustly capture the vadose zone soil moisture dynamics. Our observations in Ridgecrest, California, confirm a long-held speculation that the amount of precipitation is not a reliable predictor of vadose zone water availability in semi-arid climates. With hydrological modeling, we elucidate the highly nonlinear relationship between precipitation and shallow subsurface water storage, and the impact of long-term drought on vadose zone water availability. The findings highlight the potential of fiber seismic sensing for vadose zone hydrology in face of climate change, especially as droughts escalate. Relevant publications: NatComm paper Caltech News

1.4 Multimode wave propagation resolves polar firn structure

Understanding the density of firn (compacted snow layers) in Antarctica’s ice sheets is vital for accurate sea-level rise predictions. Seismic surveys rely on establishing empirical relations to translate seismic velocities into firn densities. Our study repurposes an existing 8-km fiber-optic cable near the Amundsen-Scott South Pole Station into a low-maintenance, continuous seismic array. Active hammer-source recordings distinguished 9 Rayleigh and 7 pseudoacoustic modes. Simultaneous inversion of these dispersion modes provides robust constraints on both P- and S- wave velocities with depth. Combined with South Pole Ice Core data, our study formulates an east Antarctica-specific empirical relation that will improve the precision of firn density estimation for the South Pole and East Antarctic region. Relevant publications: GRL paper

2. Full seismic waveform modeling and inversion with machine learning

Seismic wave propagation forms the basis for most aspects of seismological research, yet solving the wave equation is a major computational burden that inhibits the progress of research. This is exacerbated by the fact that new simulations must be performed whenever the velocity structure or source location is perturbed. Here, we explore a prototype framework for learning general solutions using a recently developed machine learning paradigm called neural operator. A trained neural operator can compute a solution in negligible time for any velocity structure or source location. We develop a scheme to train neural operators on an ensemble of simulations performed with random velocity models and source locations. As neural operators are grid free, it is possible to evaluate solutions on higher resolution velocity models than trained on, providing additional computational efficiency.

We illustrate the method with the 2D elastic wave equation and demonstrate the method’s applicability to seismic tomography, using reverse-mode automatic differentiation to compute gradients of the wavefield with respect to the velocity structure. The developed procedure is nearly two orders of magnitude faster than using conventional numerical methods for full waveform modeling and inversion. Relevant publications: TSR paper, IEEE TGRS paper

3. Fault and basin structure investigation with dense nodal arrays

3.1 Earthquake detection and location reveals hidden faults

Relevant publications: SRL paper

3.2 Converted phases provide independent constraint on LA basin depth

Converted seismic phases generated by deep events beneath or near the Los Angeles Basin (LAB) are used to determine the depth of the LAB by mapping the S-P conversion points using 3-D ray tracing. The results show that the Central Trough of the LAB is even deeper than the previously assumed value of 9 km (Wright, 1991). Relevant publications: GRL paper

4. Unraveling lithospheric structure and deformation with joint inversion

4.1 Receiver function analysis reveals strong crustal anisotropy in North China Craton

Relevant publications: Sci China paper

4.2 Joint inverstion suggests southeastern Tibet is undergoing three simultaneous tectonic modes

Relevant publications: GJI paper, Sci China paper