New 3-D Combined Inversion Scheme Using Response Functions Free From Galvanic Distortion

The combined inversion using distortion-free response functions is an effective approach to robustly estimate the 3-D electrical resistivity structure against the distortions caused by near-surface resistivity anomalies. However, previous combined inversion analyses have presented a significant dependency of the inversion results on initial and prior models. Therefore, in this study, we evaluated the effectiveness of the following two new types of 3-D combined inversion using distortion-free response functions: one uses the phase tensor and the vertical and inter-station horizontal magnetic transfer functions, while the other uses the Network-MT response functions, in addition to the former. Because long dipoles are used, the Network-MT response function is negligibly affected by galvanic distortion. To access the combined inversion approach, we developed a novel 3-D inversion scheme combining the response functions of the usual magnetotelluric measurements and the Network-MT response function. The synthetic inversion analysis demonstrated that both of the proposed combined inversions can recover the characteristic resistivity distributions of the target model without a significant dependence on the initial models, at least in the shallow part. These results demonstrate that the combined inversions using only distortion-free response functions have the potential to estimate subsurface resistivity more robustly than what was previously thought. Furthermore, we confirmed that the combined inversion using the Network-MT response function can make the resultant resistivity structure closer to the actual one and enhance the stability of the inversion. This result suggests that the combined use of the Network-MT response function is the preferred approach.

Inverting magnetotelluric data with distortion correction—stability, uniqueness and trade-off with model structure

SUMMARY Galvanic distortion of magnetotelluric (MT) data is a common effect that can impede the reliable imaging of subsurface structures. Recently, we presented an inversion approach that includes a mathematical description of the effect of galvanic distortion as inversion parameters and demonstrated its efficiency with real data. We now systematically investigate the stability of this inversion approach with respect to different inversion strategies, starting models and model parametrizations. We utilize a data set of 310 MT sites that were acquired for geothermal exploration. In addition to impedance tensor estimates over a broad frequency range, the data set also comprises transient electromagnetic measurements to determine near surface conductivity and estimates of distortion at each site. We therefore can compare our inversion approach to these distortion estimates and the resulting inversion models. Our experiments show that inversion with distortion correction produces stable results for various inversion strategies and for different starting models. Compared to inversions without distortion correction, we can reproduce the observed data better and reduce subsurface artefacts. In contrast, shifting the impedance curves at high frequencies to match the transient electromagnetic measurements reduces the misfit of the starting model, but does not have a strong impact on the final results. Thus our results suggest that including a description of distortion in the inversion is more efficient and should become a standard approach for MT inversion.

Amplitude-phase decomposition of the magnetotelluric impedance tensor

The introduction of the phase tensor marked a breakthrough in the understanding and analysis of electric galvanic distortion effects. It has been used for (distortion-free) dimensionality analysis, distortion analysis, mapping, and subsurface model inversion. However, the phase tensor can only represent half of the information contained in a complete impedance data set. Nevertheless, to avoid uncertainty due to galvanic distortion effects, practitioners often choose to discard half of the measured data and concentrate interpretation efforts on the phase tensor part. Our work assesses the information loss due to pure phase tensor interpretation of a complete impedance data set. To achieve this, a new MT impedance tensor decomposition into the known phase tensor and a newly defined amplitude tensor is motivated and established. In addition, the existence and uniqueness of the amplitude tensor is proven. Synthetic data are used to illustrate the amplitude tensor information content compared with the phase tensor. Although the phase tensor only describes the inductive effects within the subsurface, the amplitude tensor holds information about inductive and galvanic effects that can help to identify conductivity or thickness of (conductive) anomalies more accurately than the phase tensor. Furthermore, the amplitude and phase tensors sense anomalies at different periods, and thus the combination of both provides a means to evaluate and differentiate anomaly top depths in the event of data unavailability at extended period ranges, e.g., due to severe noise.

Removal of galvanic distortion effects in 3D magnetotelluric data by an equivalent source technique

The galvanic distortion induced by the electric charge buildup across near-surface inhomogeneities can severely affect the interpretation of magnetotelluric (MT) data for deeper structures. In addition to the methods already available, we have developed an alternative approach for processing MT impedance data with such distortions using an equivalent source technique. One prerequisite for the method is that all data are acquired on the surface, which is nearly always the case in land-based MT surveys. The method works with the electric field scaled from the impedance data and constructs an equivalent electrical polarization layer that attempts to reproduce the signal in the data while misfitting the galvanic distortion. Because of the uncorrelated characteristics of galvanic distortions across multiple stations at the same frequency, they can be distinguished and removed by constructing an equivalent source layer of electrical polarization using a regularized inverse formulation. The tradeoff between the signal and distortion is achieved through the use of generalized cross-validation method during the equivalent source construction, whereas the choice of equivalent source parameters also affects the separation. Numerical tests indicate that good results are obtained when the depth of the equivalent source layer is slightly greater than 10 times the nominal data spacing, and the lateral extent is twice that of the data area. The simultaneous processing with multiple frequencies yields more stable apparent resistivity curves than the separate single-frequency processing. The method has performed well in removing the galvanic distortions in the synthetic- and field-data examples.

Magnetotelluric distortions directly observed with lightning data

Galvanic distortions complicate magnetotelluric (MT) soundings. In this research, we use lightning network data to identify specific sferics in MT measurements and analyse these events on the basis of the lightning source location. Without source information, identification and removal of galvanic distortion is a fundamentally ill-posed problem, unless data are statistically decomposed into determinable and indeterminable parts. We use realistic assumptions of the earth-ionosphere waveguide propagation velocity to accurately predict the time of arrival, azimuth and amplitude for every significant sferic in our time-series data. For each sferic with large amplitude, we calculate the rotation of the electric field from the measured to the predicted arrival azimuth. This rotation of the electric field is a primary parameter of distortion. Our results demonstrate that a rudimentary model for near-surface galvanic distortion consistently fits observed electric field rotations. When local features rotate regional electric fields, then counter-rotating data to predicted arrival azimuths should correct the directional dependence of static shift. Although we used amplitude thresholds to simplify statistical processing, future developments should incorporate both signal-to-noise improvements and multisite decompositions. Lower amplitude signal may also be useful after the appropriate signal processing for noise reduction. We anticipate our approach will be useful for further work on MT distortion.