The slightly longer version
My passion for the planet beneath our feet brought me to geophysics. The combination of classical mechanics, fascinating geological processes and high-performance computing applied to usually inaccessible structures is what keeps me fascinated to this day. Exactly this inaccessibility makes inverse problems - converting data to models of the subsurface - so vital to geophysics. Any optimization in this process will lead to a better understanding of the world around us.
Most of the time, data is taken as given in inverse problems. In my project, I take a step back and investigate how much the design of an experiment can influence the expected results. The next step is how to choose an experiment that is most likely optimal in the sense of giving the most information about the subsurface given a typical dataset.
While based on a straightforward theoretical framework, solving this problem in practice is computationally very expensive, if not infeasible, even for small-scale experiments. My work over the following years will mainly focus on finding approximations that allow designing extensive surveys of one or several sensor types optimally.
After finding these methods, I will focus my research on designing experiments to answer specific questions such as: “What’s the size of a certain subsurface body?” or “Is the source of micro-seismicity moving upwards?”
All of the methods I plan on using extensively incorporate statistics and machine learning concepts. I am very excited to learn more about these to complement my geophysics background.
Finally, I want to thank the SPIN-project for giving me the unique opportunity to work on this project as part of a Marie Skłodowska-Curie Actions EU grant.
Curriculum Vitae
I keep a full length version of my CV updated and publicly available:
Download my CV in PDF
Education
MSc in Geophysics
2021
|
Department of Earth and Environmental Sciences, Technical University Munich & Ludwig-Maximilian University
Thesis: Synthetic Spectra for Synthetic Earth Models: Normal Modes as a Tool to Assess Mantle Convection Models
Advisor: Bernhard Schuberth
Download Slides
Abstract
Mantle convection models (MCM) are essential for understanding the dynamic processes governing the flow of viscous mantle material. The direct influence of this movement on plate tectonics and the global heat budget is of high importance for many related fields of geophysics. Normal modes provide a complementary dataset to earlier assessments of MCMs based on body waves. The direct misfit between spectra of real data and predictions from mantle convection models was calculated. It can be shown that for overtones that roughly probe the whole mantle, MCMs show lower misfit than the spherical symmetric reference model PREM. Fundamental modes with their sensitivity towards shallow structures do not show the same reduction of the misfit. The L2 norm used as a misfit function has been tested in the context of normal mode comparisons. While the L1 showed slightly better performance, both the L1 and L2 norm are well suited for comparing spectra. Other methods based on optimal transport could not show improvements in this regard. Under certain assumptions, the variance of a large number of stacked multiplets can be used to constrain the even degree covariance of lateral heterogeneity. This has been applied to different MCMs as well as the tomographic model S20RTS. It can be shown that the method is well suited for these comparisons and has much potential for future studies. Both measurements display a clear distinction between MCMs based on different assumptions. Models with the same past plate motion and viscosity structure follow the same trend for fundamental modes and models based on identical composition for overtone modes. Normal modes can therefore be used to differentiate between these effects. Overall isochemical (thermal) models show lower misfit and variance difference than models with a dense component in the lower mantle (thermochemical). As suggested by previous studies, thermochemical models seem to overestimate the amount of lateral heterogeneity. The results also suggest that the magnitude of heterogeneity in S20RTS may be underestimated in the lower mantle.
BSc in Earth Sciences
2019
|
Department of Earth and Environmental Sciences, Technical University Munich & Ludwig-Maximilian University
Thesis: Normal Mode Seismology in Geodynamic Models
Advisor: Bernhard Schuberth
Abstract
The behavior of normal modes in a geodynamic model is examined using the MantleConvection-Model by [Schuberth et al. 2009b], building upon the results of [Noe 2018]. The spectras were computed using the Iterative Direct Solution Method (IDSM) by [AlAttar et al. 2012], making use of its full coupling capabilities. Self-coupling or groupcoupling approximations are not exact enough, making it necessary to couple all modes used ([Aki and P. Richards 2002], [Deuss and J. H. Woodhouse 2001]). At first, the effects of scaling perturbations globally was examined. The relative deviations from PREM [Dziewonski and Anderson 1981] given by the MCM were scaled either as a whole or independently. Modes, where clear splitting can be seen, were analyzed in detail. In the second part of the thesis, the global distribution of kinetic energy through free oscillations is examined. Especially the changes caused by perturbations from spherical symmetry were examined. The most important result here was clear pattern of energy along the great circles connecting the source and the antipode. This pattern is weaker for heterogeneous models but still dominant. The same pattern also complicates analyzing the similarity of spectras globally. The similarity was quantified using the relative spectral misfit defined by [Akbarashrafi et al. 2017]. Using stacking or other techniques, this criteria may still be useful in comparing mantle/earth models.