High-resolution Bayesian spatial autocorrelation (SPAC) quasi-3-D Vs model of Utah FORGE site with a dense geophone array

Abstract

Summary We expand the application of spatial autocorrelation (SPAC) from typical 1D Vs profiles to quasi-3D imaging via Bayesian Monte-Carlo inversion (BMCI) using a dense nodal array (49 nodes) located at the Utah Frontier Observatory for Research in Geothermal Energy (FORGE) site. Combinations of 4 and 9 geophones in subarrays provide for 36 and 25 1D Vs profiles, respectively. Profiles with error bars are determined by calculating coherency functions that fit observations in a frequency range of 0.2–5 Hz. Thus, a high-resolution quasi-3D Vs model from the surface to 2.0 km depth is derived and shows that surface-parallel sedimentary strata deepen to the west, consistent with a 3D seismic reflection survey. Moreover, the resulting Vs profile is consistent with a Vs profile derived from distributed acoustic sensing (DAS) data located in a borehole at the FORGE site. The quasi-3D velocity model shows that the base of the basin dips ∼22o to the west and topography on the basement interface coincident with the Mag Lee Wash suggests that the bedrock interface is an unconformity.

Subsurface Imaging by a Post-Stimulation Walkaway Vertical Seismic Profile Using Distributed Acoustic Sensing at the Utah FORGE Enhanced Geothermal System Site

“…Sequentially, the 1D P-wave velocity model from ZVSP was expanded to 3D by correlating with a 3D composite velocity model from [17] and the basement interface (Figure 15). The 3D composite velocity model was contributed from two velocity models, a base model from the quasi-3D VS model of [18] and VP and VS profiles from [7]. A quasi-3D Swave velocity model was constructed [18] by applying spatial autocorrelation (SPAC) over 61 1D VS profiles over an approximately 3.5 km by 3.5 km (11,482.94 ft by 11,482.94 ft) area via Bayesian Monte Carlo inversion (BMCI).…”

“…The 3D composite velocity model was contributed from two velocity models, a base model from the quasi-3D VS model of [18] and VP and VS profiles from [7]. A quasi-3D Swave velocity model was constructed [18] by applying spatial autocorrelation (SPAC) over 61 1D VS profiles over an approximately 3.5 km by 3.5 km (11,482.94 ft by 11,482.94 ft) area via Bayesian Monte Carlo inversion (BMCI). Their quasi-3D SPAC VS model provides reliable velocity images with a resolvable depth down to 2.0 km (6561.68 ft).…”

Depth of Sudden Velocity Changes Derived From Multi‐Mode Rayleigh Waves

“…Assuming our approach is sensitive to this deeper velocity increase would mean that the deeper velocity increase is sharper than the sediment‐to‐granitoid boundary at shallower depths. Using estimated velocities from Zhang and Pankow (2021) and typical densities for sandstone and granite, we estimate the shear impedance contrast of the sediment‐to‐granitoid interface to be $$IC=\frac{2630\,\mathrm{k}\mathrm{g}/{\mathrm{m}}^{\mathrm{3}}\,3000\,\mathrm{m}/\mathrm{s}}{2000\,\mathrm{k}\mathrm{g}/{\mathrm{m}}^{\mathrm{3}}2000\,\mathrm{m}/\mathrm{s}}=1.97$$ and would, thus, be too small for our approach. Thus, we conclude that DARE may confirm a deeper increase in velocities.…”

Microseismic Monitoring and Analysis Using Cutting-Edge Technology: A Key Enabler for Reservoir Characterization

“…A single 1-D anisotropic velocity model with three layers was used in the forward model. The Swave velocity was taken from the FORGE velocity estimates by Zhang and Pankaw [61] and Wamriew et al [51], while the P-wave velocities and the densities were estimated using the Castagna [62] and Gardner [63] equations, respectively. Relatively high Thomsen anisotropic parameters [64] of ϵ = 0.51, γ = 0.36, and σ = 0.25, were chosen since previous studies have revealed that a neural network, trained on high anisotropic parameters, would generalize well when presented with waveforms from lower anisotropic models [48].…”