Modeling P waves in seismic noise correlations: Advancing fault monitoring using train traffic sources

Abstract

The theory of Green's function retrieval essentially requires homogeneously distributed noise sources. Even though these conditions are not fulfilled in nature, low-frequency (<1 Hz) surface waves generated by ocean-crust interactions have been used successfully to image the crust with unprecedented spatial resolution. In contrast to low-frequency surface waves, high-frequency (>1 Hz) body waves have a sharper, more localized sensitivity to velocity contrasts and temporal changes at depth. In general, their retrieval using seismic interferometry is challenging, and recent studies focus on powerful, localized noise sources. They have proven to be a promising alternative but break the assumptions of Green's function retrieval. In this study, we present an approach to model correlations between P waves for these scenarios and analyze their sensitivity to 3D Earth structure. We perform a series of numerical experiments to advance our understanding of these signals and prepare for an application to fault monitoring. In the considered cases, the character of the signals strongly diverges from Green's function retrieval, and the sensitivity to structure has significant contributions in the source direction. An accurate description of the underlying physics allows us to reproduce observations made in the context of monitoring the San Jacinto Fault in California using train-generated seismic waves. This approach provides new perspectives for detecting and localizing temporal velocity changes previously unnoticed by commonly exploited surface-wave reconstructions.

Seeking Repeating Anthropogenic Seismic Sources: Implications for Seismic Velocity Monitoring at Fault Zones

“…In addition to the proposed data‐driven techniques, numerical simulations of CFs (Sager et al., 2021 ) are beneficial to understanding the correlation wavefields. Anthropogenic sources often have specific spatio‐temporal distributions, which can be adequately simulated.…”

“…In simulations, separating correlation wavefields excited by fast‐moving P‐waves from slow‐moving S‐ or surface waves is feasible. This provides guidance in explaining the observed wave packets and contributes to describing the sensitivity kernels for structural perturbations (Sager et al., 2021 ). Numerical simulations can also facilitate the study of CFs affected by source interactions, advancing a more comprehensive understanding of correlation wavefields.…”

Detecting the Unseen: Understanding the Mechanisms and Working Principles of Earthquake Sensors

“…Sensors play a crucial role in seismic monitoring and are used in various applications such as mobile gravity monitoring [30], electromagnetic wave signal detection [31], and cross-fault deformation measurement [32]. The first seismic network was established in California, USA in 1929 using Wood-Anderson seismometers [33] Modern seismic networks typically consist of broadband and strong-motion seismometers.…”

“…Artificial seismic noise is usually dominated by high-frequency body waves providing a high spatial resolution. In addition, the location of artificial noise sources is often fixed (e. g., industrial operations) or moves along a fixed trajectory (e. g., trains and cars), which is easy to track and simulate the movement of noise sources [30].…”