Simulation-based probabilistic inversions of 3D magnetotelluric (MT) data are arguably the best option to deal with the non-linearity and non-uniqueness of the MT problem. However, the computational cost associated with the modeling of 3D MT data has so far precluded the community from adopting and/or pursuing full probabilistic inversions of large MT datasets. In this contribution, we present a novel and general inversion framework, driven by Markov chain Monte Carlo (MCMC) algorithms, which combines i) an efficient parallel-in-parallel structure to solve the 3D forward problem, ii) a reduced order technique to create fast and accurate surrogate models of the forward problem, and iii) adaptive strategies for both the MCMC algorithm and the surrogate model. In particular, and contrary to traditional implementations, the adaptation of the surrogate is integrated into the MCMC inversion. This circumvents the need of costly offline stages to build the surrogate and further increases the overall efficiency of the method.We demonstrate the feasibility and performance of our approach to invert for large-scale conductivity structures with two numerical examples using different parameterizations and dimensionalities. In both cases, we report staggering gains in computational efficiency compared to traditional MCMC implementations. Our method finally removes the main bottleneck of probabilistic inversions of 3D MT data and opens up new opportunities for both stand-alone MT inversions and multi-observable joint inversions for the physical state of the Earth’s interior.
In this work, we study seismoelectric conversions generated in the vadose zone, when this region is traversed by a pure SH wave. We assume that the soil is a 1-D partially saturated lossy porous medium and we use the van Genuchten's constitutive model to describe the water saturation profile. Correspondingly, we extend Pride's formulation to deal with partially saturated media. In order to evaluate the influence of different soil textures we perform a numerical analysis considering, among other relevant properties, the electrokinetic coupling, coseismic responses and interface responses (IRs). We propose new analytical transfer functions for the electric and magnetic field as a function of the water saturation, modifying those of Bordes et al. and Garambois & Dietrich, respectively. Further, we introduce two substantially different saturation-dependent functions into the electrokinetic (EK) coupling linking the poroelastic and the electromagnetic wave equations. The numerical results show that the electric field IRs markedly depend on the soil texture and the chosen EK coupling model, and are several orders of magnitude stronger than the electric field coseismic ones. We also found that the IRs of the water table for the silty and clayey soils are stronger than those for the sandy soils, assuming a non-monotonous saturation dependence of the EK coupling, which takes into account the charged air-water interface. These IRs have been interpreted as the result of the jump in the viscous electric current density at the water table. The amplitude of the IR is obtained using a plane SH wave, neglecting both the spherical spreading and the restriction of its origin to the first Fresnel zone, effects that could lower the predicted values. However, we made an estimation of the expected electric field IR amplitudes detectable in the field by means of the analytical transfer functions, accounting for spherical spreading of the SH seismic waves. This prediction yields a value of 15 µV m −1 , which is compatible with reported values.
Joint inversions of two or more geophysical data sets are becoming common practice for imaging the Earth's interior and elucidating the physical state of the planet. When the inverted data sets have complementary sensitivities to the properties of interest, joint inversions significantly reduce the ambiguity inherent in single-data set inversions, achieve more stable solutions, increase identifiability of features and enhance model resolution. Perhaps more importantly, certain properties of the Earth's interior can only be revealed by combining observations from different techniques. An example is the bulk composition of the lithospheric mantle, which requires independent constrains on the bulk density (e.g., from gravity data sets) and shear-wave velocity (e.g., from surface-wave data). Recent discussions on the benefits and limitations of joint approaches for imaging the structure of the lithosphere and upper mantle can be found in, for example, Khan et al. (2006); Afonso, Fullea, Griffin,
Abstract. The seismo-electromagnetic method (SEM) can be used for non-invasive subsurface exploration. It shows interesting results for detecting fluids such as water, oil, gas, CO 2 , or ice, and also help to better characterise the subsurface in terms of porosity, permeability, and fractures. However, the challenge of this method is the low level of the induced signals. We first describe SEM's theoretical background, and the role of some key parameters. We then detail recent studies on SEM, through theoretical and numerical developments, and through field and laboratory observations, to show that this method can bring advantages compared to classical geophysical methods.
Summary Simulation-based probabilistic inversions of 3D magnetotelluric (MT) data are arguably the best option to deal with the non-linearity and non-uniqueness of the MT problem. However, the computational cost associated with the modeling of 3D MT data has so far precluded the community from adopting and/or pursuing full probabilistic inversions of large MT datasets. In this contribution, we present a novel and general inversion framework, driven by Markov chain Monte Carlo (MCMC) algorithms, which combines i) an efficient parallel-in-parallel structure to solve the 3D forward problem, ii) a reduced order technique to create fast and accurate surrogate models of the forward problem, and iii) adaptive strategies for both the MCMC algorithm and the surrogate model. In particular, and contrary to traditional implementations, the adaptation of the surrogate is integrated into the MCMC inversion. This circumvents the need of costly offline stages to build the surrogate and further increases the overall efficiency of the method. We demonstrate the feasibility and performance of our approach to invert for large-scale conductivity structures with two numerical examples using different parameterizations and dimensionalities. In both cases, we report staggering gains in computational efficiency compared to traditional MCMC implementations. Our method finally removes the main bottleneck of probabilistic inversions of 3D MT data and opens up new opportunities for both stand-alone MT inversions and multi-observable joint inversions for the physical state of the Earth’s interior.
The behaviour of CO 2 deposition sites-and their surroundings-during and after carbon dioxide injection has been matter of study for several years, and several geophysical prospection techniques like surface and crosshole seismics, geoelectrics, controlled source electromagnetics among others, have been applied to characterize the behaviour of the gas in the reservoirs. Until now, Seismolectromagnetic wave conversions occuring in poroelastic media via electrokinetic coupling have not been tested for this purpose. In this work, by means of numerical experiments using Pride's equations-extended to deal with partial saturations-we show that the seismoelectric and seismomagnetic interface responses (IR) generated at boundaries of a layer containing carbon dioxide are sensitive to its CO 2 content. Further, modeling shear wave sources in surface to borehole seismoelectric layouts and employing two different models for the saturation dependence of the electrokinetic coefficient, we observe that the IR are sensitive to CO 2 saturations ranging between 10% and 90%, and that the CO 2 saturation at which the IR maxima are reached depends on the aforementioned models. Moreover, the IR are still sensitive to different CO 2 saturations for a sealed CO 2 reservoir covered by a clay layer. These results, which should be complemented by the analysis of the IR absolute amplitude, could lead, once confirmed on the field, to a new monitoring tool complementing existing ones.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.