Search

The Cascade arc runs from Southern British Columbia through Northern California and developed by the subduction of the Juan de Fuca plate beneath North America. Among the volcanoes of the Northern Cascades, Mount St Helens (MSH) is the most active.Inter-connected melt within the underlying magmatic systems of volcanic regions allow identification of the system via Magnetotelluric (MT) imaging due to the sensitivity of the method to electrical conductivity. Time variance of the properties of the melt due to changes in eruptive activity may result in significant conductivity changes that can be identified with continuous MT monitoring. Galvanic distortions resulting from the small near-surface conductivity variations may can be difficult to separate from the expected subtle observational differences. However, impedance phase relations in MT are free of these galvanic distortions. Thus, by exploiting changes in the phase response (the MT Phase tensor), it is possible to identify the temporal conductivity changes associated with changes occurring within the magmatic system.Prior MT results in the Washington cascades show a significant conductive anomaly underneath the MSH attributed to be partial melt supplying the volcano. With the aim of revealing conductivity changes beneath MSH, both a campaign style reoccupation of measurements completed during the 2004-2008 dome building eruption, and installation of four ‘continuous’ monitoring MT stations has been completed. Among the new MT dataset, a preliminary comparative phase tensor analysis of 56 repeated measurements identifies conductivity changes at frequencies that correlate the depth of the partial melt within the magmatic system., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

Unravelling the internal structure of volcanoes is an essential step to understand the architecture of the magmatic system and advance hazard assessment. Electromagnetic (EM) measurements are particularly suited for this task due to the high contrast in resistivity between melts and solid rock. Due to this, EM surveys of active volcanoes have gained significant attention in recent years. However, many modelling and inversion algorithms only provide a rough approximation of the significant topography of the volcanoes. Currently it is not clear to which level the topography has to be incorporated to provide reliable and robust results. We use two modern finite-element modelling algorithms for magnetotelluric data to assess the impact of different approximations. Based on the geometry of a recent survey at Mt. St Helens we will compare the results of the two algorithms with different topography approximations. We will discuss different strategies for incorporating topography and how to ensure valid calculations of magnetotelluric forward responses., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

Joint inversion of different geophysical data has been receiving significant attention in recent years due to its potential to improve the quality of subsurface models and provide additional dimensions of analysis over single-data approaches. For example, the relationship between different physical parameters can be used to provide classifications of lithology. However, the technical barrier of entry for new joint inversion methods is very high, as they require efficient forward engines for multiple methods, data and grid management algorithms and useful coupling methods between the different parameters. Jif3D is an open-source C++ joint inversion framework that provides all necessary ingredients for effective inversions in three dimensions. It comprises parallel solvers for gravity, magnetics, DC resistivity, seismic traveltimes, surface waves and magnetotellurics. In addition, a variety of coupling approaches such as Cross-Gradients and Variation of Information are implemented. It can be used as a ready-to-use inversion program or as an experimental platform to develop new approaches thanks to its modular design. Jif3D has been used in a variety of contexts from mineral exploration to large-scale tectonic imaging with different combinations of data. I will show selected case studies using jif3D and discuss possible applications of joint inversion. In particular I will demonstrate the value of joint inversion using seismic, magnetotelluric, gravity and magnetic data from the western United States where currently active tectonics interact with inherited structures to form a complex subsurface., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

We present results of three-dimensional joint inversion of seismic, magnetotelluric, and gravity data over a marine salt dome. Such structures are difficult to image with a single method, and our results demonstrate how combining different techniques can yield improved results. More importantly, we examine the reliability of velocity-conductivity relationships derived from structure-coupled joint inversion approaches. Comparison with a seismic reflection section shows that our models match the upper limit of the salt. Furthermore, velocity and resistivity logs from a borehole drilled into the salt dome's flank match, within error, those recovered by the inversion. The good match suggests that the difference in length scale does not have a significant effect in this case. This provides a strong incentive to incorporate borehole data into the joint inversion in the future and substantiates approaches that use the relationships derived from joint inversion models for lithological classification.

Magnetotellurics (MT) uses a frequency-dependent impedance tensor estimated from the spectra of associated time-varying horizontal electric and magnetic fields measured at the Earth's surface to image the sub-surface of the Earth. Most current methods use Fourier transform based procedures to estimate power spectral densities and, therefore, assume that the signals are stationary over the record length. Stationarity in geomagnetic data, however, is not always ensured given the variety of source mechanism causing the geomagnetic variations at different time and spatial scales. Additional complication and bias may arise from the presence of noise in the recorded electric and magnetic file data. Sophisticated MT data processing account for a potential bias through windowing of the time series as well as robust estimates of the impedance. We explore a new heuristic method for dealing with the non-stationarity of MT time series based on empirical mode decomposition. It is a dynamic time series analysis method, in which complicated data sets can be decomposed into a finite and small number of simple intrinsic mode functions. In this paper, we use the empirical mode decomposition method to decompose MT data into intrinsic mode functions and calculate the instantaneous frequencies and spectra to determine the impedance tensor. We investigate the reliability of the impedance estimates on synthetic data by comparing the results to those obtained by analytical methods. Finally, we apply our processing scheme to data measured from the Costa Rica subduction zone, and compare the results from our new method to the frequently-used BIRRP processing method.

Magnetotelluric (MT) method determines a frequency dependent impedance tensor using the spectra of associated time-varying horizontal electric and magnetic fields measured at the Earth’s surface. In this abstract, we present a dynamic time series analysis method dealing the non-stationary MT data to infer the impedance tensor. Most current methods to determine the spectra use Fourier transform based procedure and, therefore, assume that the signals are stationary over the record length. We introduce a new method for dealing with non-stationarity of the MT time series based upon empirical mode decomposition (EMD) method, a dynamic time series analysis method. Using EMD complicated data sets can be decomposed into a finite and small number of "intrinsic mode functions" (IMFs), which are mono-component signals and allow the calculation of physical meaningful instantaneous frequencies. EMD has no bias due to non-stationary of geomagnetic time series, since the IMFs are based entirely on signal characteristics and not on any given set of base functions such as sines and cosines in the Fourier transform or wavelets in the Wavelet transform. We use the EMD method to decompose MT data into IMFs and calculate the instantaneous frequencies and spectra to determine the impedance tensor. The method is tested in synthetic and real marine MT data sets, the obtained estimate results are reliable compared to frequently-used BIRRP processing method. Furthermore, new method has the possibility of noise visualization and filtering, which is especially important in marine applications, where noise free time segments maybe short.

Joint inversion strategies for geophysical data have become increasingly popular since they allow to combine complementary information from different data sets in an efficient way. However, for joint inversion algorithms that use methods that are sensitive to different parameters it is important that they are not restricted to specific survey arrays and subsurface conditions. Hence, joint inversion schemes are needed that 1) adequately balance data from the different methods and 2) use links between the parameter models that are suited for a wide range of applications. Here, we combine MT, seismic tomography and gravity data in a non-linear joint inversion that accounts for these critical issues. Data from the different methods are inverted separately and are joined through constrains accounting for parameter relationships. An advantage of performing the inversions separately (and not together in one matrix) is that no relative weighting between the data sets is required. To avoid that the convergence behavior of the inversions is profoundly disturbed by the coupling, the strengths of the associated constraints are re-adjusted at each iteration. As criteria to control the adaption of the coupling strengths we used a general version of the well-known discrepancy principle. Adaption of the coupling strengths makes the joint inversion scheme also applicable to subsurface conditions, for which the assumed relationships are only a rough first order approximation. So, the coupling between the different parameter models is automatically reduced if for some structures the true rock property behaviors differ significantly from the assumed relationships (e.g. the atypical density-velocity behavior of salt). We have tested our scheme first on different synthetic 2-D models for which the assumed parameter relationships are everywhere valid. We observe that the adaption of the coupling strengths makes the convergence of the inversions very robust and that the final results are close to the true models. In a next step the scheme has been applied on models for which the assumed parameter relationships are invalid for some structures. For these structures deviations from the relationships are present in the final results; however, for the remaining structures the relative behaviors of the physical parameters are still approximately described by the assumed relationship. Finally, we applied our joint inversion scheme on seismic, MT and gravity data collected offshore the Faroe Islands, where basalt intrusions are present.

The finite difference time domain code TEMDDD was modified for the 3D forward modeling of marine CSEM data. After changes in the code, which make it possible to create model geometries typically encountered in marine CSEM experiments, parts of the code have been parallelized using massive parallelization on graphic cards. Parts of the singular value decomposition, which is the most time consuming part of the code, have been successfully ported with massive speed-ups (8-12x faster) observed as compared to the standard code. The full parallelization of the code is still work in progress.

EGU2010-4184-2 Joint inversion of different datasets is emerging as an important tool to enhance resolution and decrease inversion artifacts in structurally complex areas. Performing the inversion in 3D allows us to investigate such complex structures but requires computationally efficient forward modeling and inversion methods. Furthermore we should be able to flexibly change inversion parameters, coupling approaches and forward modeling schemes in order to find a suitable approach for the given target. We present a 3D joint inversion framework for scalar and full tensor gravity, magnetotelluric and seismic data that allows us to investigate different approaches. It consists of two memory efficient gradient based optimization techniques, L-BFGS and NLCG, and optimized parallel forward solvers for the different datasets. In addition it provides the necessary flexibility in terms of model parametrization and coupling method by completely separating the inversion parameters and geometry from the parametrization of the individual method. This separation allows us to easily switch between completely different types of parameterizations and use structural coupling as well as coupling based on parameter relationships for the joint inversion. First tests on synthetic data with a fixed parameter relationship coupling show promising results and demonstrate that 3D joint inversion is becoming feasible for realistic size models.

Hongyi's Master's thesis (M.Sc. in Geophysics at LMU Munich and TUM), supervised by Prof. Alice-Agnes Gabriel and Dr. Max Moorkamp Summary: 1) Built a regional-scale unstructured tetrahedronmesh for the Botswana region(700km by 700km)withtwonon-planar fault planesbased on a local 3D resistivity model which wasinverted byMoorkamp et al. (2019). 2) Establisheda physics-based dynamic rupture base model for Mw 6.5 Botswana intraplate earthquake using SeisSol (based on Discrete Galerkin Method).Coupleslip on the fault with 3D elastic seismic wave propagation. 3) Usedlaboratory-based rapid velocity weakening friction law,the friction law isadapted from the formulation introduced byDunham et al. (2011). 4) Comparedsynthetics to recorded seismicand geodeticdata. Despite the simplesetup, the synthetics generally agree with the observed seismogram (arrival time, phasepolarity) and InSAR data(subsidence up to 5 cm,Materna et al., 2019). 5) Discussed the source ofmisfit betweensynthetic and observedseismograms. 6) Discussedfuture work,including the possibility ofusing anavailablekinematic model (Materna et al. 2019) to better constrain initial stress distributiononthe fault. &nbsp

Refereed/Peer-reviewed The tectonic history of southern Africa includes Archean craton formation, multiple episodes of subduction and rifting and some of the world's most significant magmatic events. These processes left behind a compositional trail that can be observed in xenoliths and measured by geophysical methods. The abundance of kimberlites in southern Africa makes it an ideal place to test and calibrate mantle geophysical interpretations that can then be applied to less well-constrained regions. Magnetotellurics (MT) is a particularly useful tool for understanding tectonic history because electrical conductivity is sensitive to temperature, bulk composition, accessory minerals, and rock fabric. We produced three-dimensional MT models of the southern African mantle taken from the SAMTEX MT data set, mapped the properties of ∼36,000 garnet xenocrysts from Group I kimberlites, and compared the results. We found that depleted regions of the mantle are uniformly associated with high electrical resistivities. The conductivity of fertile regions is more complex and depends on the specific tectonic and metasomatic history of the region, including the compositions of metasomatic fluids or melts and the emplacement of metasomatic minerals. The mantle beneath the ∼2.05 Ga Bushveld Complex is highly conductive, probably caused by magmas flowing along a lithospheric weakness zone and precipitating interconnected, conductive accessory minerals such as graphite and sulfides. Kimberlites tend to be emplaced near the edges of the cratons where the mantle below 100 km depth is not highly resistive. Kimberlites avoid strong mantle conductors, suggesting a systematic relationship between their emplacement and mantle composition. Plain Language Summary: The present-day composition of Earth's tectonic plates results from past geological processes. We can learn about Earth's composition from deep rock samples that are carried to the surface during volcanic eruptions and by probing its physical properties, like electrical conductivity, with geophysics. In southern Africa, there are extensive deep rock samples, which have been brought to the surface by kimberlite volcanoes that also host diamonds, and also extensive geophysical data. In this article, we compare the rock compositions with electrical conductivity to learn more about Earth's composition. Our results show that the oldest parts of the plates, which retain compositions similar to their initial composition, appear resistive. On the other hand, regions that have been intruded by deep fluids or molten rock can be resistive or conductive, depending on the types of minerals that were formed during the intrusion. The kimberlite volcanoes mostly erupted through the edges of the most resistive parts of the plates and did not erupt through the conductors. These results will help us to make more accurate interpretations about the composition of parts of the Earth where we do not have deep rock samples.