Geophysikalisches Institut (GPI)

Challenges in near surface 2D SH full waveform inversion field data applications

  • Tagungsort:


  • Datum:


  • Referent:

    Dr. Daniel Köhn
    Universität Kiel (CAU)

  • Zeit:


Classical seismic inversion approaches like first arrival travel-time tomography or surface wave dispersion
analysis are limited in their resolution to the size of the first Fresnel zone. By incorporating the
full information content of the recorded seismic wavefield, the Full Waveform Inversion (FWI) method
is able to resolve sub-wavelength structures with high fidelity.

Since 2014 the working group "Applied Geophysics" at the Christian-Albrechts-University Kiel applies
the FWI to seismic SH data with different degrees of complexity for near-surface characterization tasks
related to archaeological prospection and engineering geophysics. In this talk, I will discuss 4 challenges
which have to be tackled in order to successfully apply FWI to near-surface field data and draw appropriate
conclusions from the inversion results.

1. To accurately model the dispersion properties of surface waves, required for waveform fitting in the
FWI framework, intrinsic damping effects have to be incorporated into the underlying governing equations
describing wave propagation in visco-elastic media. The impact of visco-elastic medium properties will be
demonstrated for a field dataset, acquired to characterize the basement shape of a medieval canal in southern

2. Significant surface topography poses another challenge for accurate surface wave FD modelling on Cartesian
grids. Possible solutions to this problem will be discussed for a SH dataset related to the archaeological
prospection of the Danewerk fortification wall in northern Germany.

3. Each geophysical inversion approach requires a careful choice of smoothness constraints to distinguish
subsurface structures from inversion artifacts. After introducing two approaches to incorporate model
smoothness in the FWI workflow, different degrees of smoothness constraints will demonstrate the challenge
to find the right balance between the suppression of inversion artifacts and loosing model resolution.

4. A very important question is how reproducible are FWI results? This is actually a very complicated problem,
involving a wide range of aspects. Here, I want to focus on the effect of the inversion of SH-data with
different polarity on the FWI result for the Danewerk problem. Archaeological excavations allow a groundtruthing
of the inverted S-wave velocity models.

Einladungen für das Zoom - Meeting werden per E-Mail an die gpistud-Liste und gpi-personal-Liste geschickt. Zusätzlich werden die Informationen in der Seminargruppe auf ILIAS gepostet. 
Außenstehende kontaktieren bei Interesse bitte Mike Lindner.