Critical Zone Structure by Elastic Full Waveform Inversion of Seismic Refractions in a Sandstone Catchment, Central Pennsylvania, USA

Abstract

Seismic imaging provides key information for revealing structures within Earth's critical zone (CZ) and quantifying subsurface fluid properties. The P‐wave velocity (Vp) models estimated by seismic refraction tomography or acoustic full waveform inversion (FWI) are useful to delineate the thicknesses of weathered bedrock but ambiguously map fluid properties in CZ. Considering the complementary sensitivity of S‐wave to subsurface fluid saturation, we explore advanced elastic full waveform inversion (EFWI) to estimate both Vp and S‐wave velocity (Vs) models simultaneously. Several strategies are proposed for robust EFWI implementation of noisy single (vertical) component refraction data: (a) we window seismic data to preserve early arrivals mainly including P‐wave, converted waves, and S‐wave refractions; (b) we use a correlative misfit rather than the classic L2 misfit to alleviate the interference of unreliable data amplitudes; and (c) we perform inversion using a multiscale frequency strategy with the iterative constraint with the rule of Vp/Vs > 1. We validate that the EFWI approach reliably reconstructs Vp and Vs models using synthetic data. Then, we apply our EFWI approach to seismic refraction data acquired at the Garner Run site of the Susquehanna Shale Hills Critical Zone Observatory. The inverted Vp and Vs models indicate three distinguished layers and with significant lateral and depth heterogeneities (e.g., low Vp and Vs zones). Joint analyses of Vp, Vs, and Vp/Vs with rock‐physics knowledge reveal potential gas or water gathering zones.

Understanding Subsurface Fracture Evolution Dynamics Using Time‐Lapse Full Waveform Inversion of Continuous Active‐Source Seismic Monitoring Data

“…To avoid the challenge of matching exact amplitudes varied with shots and traces/offsets, we solve the correlative‐based misfit that measures the global phase closeness between $${d}_{m}$$ and $${d}_{obs}$$ as follows: $$$f\left({\mathbf{v}}_{p}\right)=\int \int \left(1-\frac{\int {d}_{m}\left({\mathbf{x}}_{r},t;{\mathbf{x}}_{s}\right){d}_{obs}\left({\mathbf{x}}_{r},t;{\mathbf{x}}_{s}\right)dt}{\sqrt{\int {d}_{m}^{2}\left({\mathbf{x}}_{r},t;{\mathbf{x}}_{s}\right)dt}\sqrt{\int {d}_{obs}^{2}\left({\mathbf{x}}_{r},t;{\mathbf{x}}_{s}\right)dt}}\right)d{\mathbf{x}}_{r}d{\mathbf{x}}_{\boldsymbol{s}},$$$ where t , $${\mathbf{x}}_{\boldsymbol{r}}$$, and $${\mathbf{x}}_{\boldsymbol{s}}$$ represent the time, receiver, and source coordinates, respectively. The correlative misfit can relax the requirement for offset‐ and shot‐dependent amplitude fitting and enables FWI to be robust; this approach has been applied to several field data studies (e.g., Liu et al., 2022).…”

“…To avoid the challenge of matching exact amplitudes varied with shots and traces/ offsets, we solve the correlative-based misfit that measures the global phase closeness between 𝐴𝐴 𝐴𝐴𝑚𝑚 and 𝐴𝐴 𝐴𝐴𝑜𝑜𝑜𝑜𝑜𝑜 as follows: where t, 𝐴𝐴 𝐱𝐱𝒓𝒓 , and 𝐴𝐴 𝐱𝐱𝒔𝒔 represent the time, receiver, and source coordinates, respectively. The correlative misfit can relax the requirement for offset-and shot-dependent amplitude fitting and enables FWI to be robust; this approach has been applied to several field data studies (e.g., Liu et al, 2022).…”

2D Near‐Surface Full‐Waveform Tomography Reveals Bedrock Controls on Critical Zone Architecture

“…X. Liu et al (2022) used the NC norm to apply FWT in the CZ despite working with relatively noisy data, while Borisov et al (2020) found that using the NC norm helped contend with unmodeled anelastic and 3D effects. In choosing the NC norm, the corresponding adjoint source for each shot at Earth and Space Science 10.1029/2023EA003248 each receiver location is given by f † i,j = ûi,j ∫ T ûi,j • û0 i,j dt) -û0 i,j (Choi & Alkhalifah, 2012), where i and j are the shot and receiver indices respectively.…”

“…To date, however, FWT has only been used to study the near surface on a handful of occasions (e.g., Cai & Zelt, 2022;J. Chen et al, 2017;Köhn et al, 2018;X. Liu et al, 2022;Sheng et al, 2006;W.…”

Low Vp/VsValues as an Indicator for Fractures in the Critical Zone

“…Other observations report a larger‐than‐expected range, including high V p /V s values greater than 4 (Flinchum et al., 2020; Pasquet et al., 2015a; Stümpel et al., 1984) or even as high as 20 (Salas‐Romero et al., 2021). There are reports of V p /V s values less than 1.4 in the CZ of Brazil (Trichandi et al., 2022), in dry soils (Salem, 2015), and in full waveform inversion results in the CZ of Pennsylvania (Liu et al., 2022). These observations raise an important question: what can anomalous V p /V s values tell us about physical and chemical processes that drive CZ evolution?…”