Oceanic high-frequency global seismic wave propagation with realistic bathymetry

Abstract

We present a new approach to simulate high-frequency seismic wave propagation in and under the oceans. Based upon AxiSEM3D (Leng et al. 2019), this method sup- ports a fluid ocean layer, with associated water-depth phases and seafloor topography (bathymetry). The computational efficiency and flexibility of this formulation means that high-frequency calculations may be carried out with relatively light computational loads. A validation of the fluid ocean implementation is shown, as is an evaluation of the oft-used ocean loading formulation, which we find breaks down at longer periods than was previously believed. An initial consideration of the effects of seafloor bathymetry on seismic wave propagation is also given, wherein we find that the surface waveforms are significantly modified in both amplitude and duration. When compared to observed data from isolated island stations in the Pacific, synthetics which include a global ocean and seafloor topography appear to more closely match the observed waveform features than synthetics generated from a model with topography on the solid surface alone. We envisage that such a method will be of use in understanding the new and exciting ocean-bottom and floating seismometer datasets now being regularly collected.

SASSIER22: Full‐Waveform Tomography of the Eastern Indonesian Region That Includes Topography, Bathymetry, and the Fluid Ocean

“…Thus, the effects for other regions and different scales remain unclear, but our results are in good agreement with Fernando et al. (2020), who consider epicentral distances of up to 90° across the Pacific Ocean. While we only decrease the minimum period band from 20 to 15 s compared to SASSY21 , we expect that the effects of surface elevation and the fluid ocean become even more significant at shorter periods and when running additional iterations.…”

“…We observe the strongest effect on the radial and vertical components, which suggests that the ocean's effects are confined to the source–receiver plane, as previously suggested by Fernando et al. (2020).…”

“…We mesh irregularly patched oceans (rather than a global ocean layer of constant thickness as in Fernando et al., 2020), which allows us to investigate waveforms along different source‐receiver paths. Figure 10 presents several waveform matches for observed data with synthetics obtained from both ocean types.…”

“…Thus, its success is strongly dependent on being able to produce realistic synthetics from potentially complex models of the Earth's interior. In particular, topography, bathymetry, and the ocean can have a significant effect on ground motion; for example, surface topography can lead to seismic wave scattering (e.g., Lee et al., 2008) and the fluid ocean elongates the Rayleigh surface wave train (e.g., Fernando et al., 2020; Todoriki et al., 2016). However, these effects are not routinely accounted for in surface wave and waveform tomography, largely due to the need for sophisticated meshing techniques, plus the added computational burden associated with simulating waves in the low‐velocity fluid ocean.…”

Solving the wave equation with physics-informed deep learning

“…In these approaches the wave equation is discretised and iterative time-stepping schemes are used to solve it. These methods have benefited from decades of development, and highly sophisticated algorithms exist which are able to model a large range of physics; for example in geophysics, * bmoseley@robots.ox.ac.uk the effects of solid-fluid interfaces, intrinsic attenuation and anisotropy can be readily incorporated into the solution (Colombi et al 2012, Fernando et al 2020, Tesoniero et al 2020.…”