Improving tomographic models of the Pyrenees region (NE Iberia): Preliminary results from Full-Waveform Inversion

Detailed imaging of subsurface crustal structures requires inverse modeling from massive amounts of waveform data, including regional and/or local seismic events of lower magnitudes and high frequencies, aspects not conventionally used in Full-Waveform Inversion (FWI). The use of higher frequency signals notably boosts spatial model resolution but increases the computational cost of inversion, affordable only with HPC infrastructures. Preliminary tomographic models for the Eastern Pyrenees region have been constrained through a visco-elastic FWI using seismic data. Three-component recordings were provided by over 120 stations from a high-density seismic deployment. To optimize coverage, in addition to the far-field waveforms from M>4.5 events, data from 23 smaller magnitude earthquakes were included from a temporary broad-band seismic network deployed between September 2019 and November 2020. Moment tensor solutions of these additional seismic data were determined, and the seismograms were bandpass filtered. Through this computationally intensive workflow, multiple iterations resulted in a better fit with the observed seismograms, and point to improved spatial resolution of shallow structures. This contribution, aims to disclose and discuss the difficulties and challenges faced when using relatively high frequency seismic recordings in visco-elastic FWI applications, as well as addressing the benefits of the resulting models. Improved tomographic models could be the basis for more accurate waveform modeling and we hope that they will contribute to improve ground-motion simulations on a regional scale. This research is funded and contributes to the objectives of the Horizon Europe Project DT-GEO: A Digital Twin for GEOphysical extremes (ID 101058129).
The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)