Navigation – Plan du site

AccueilNuméros45-11. Case studies and archaeologica...Using Seismic Full-Waveform Inver...

1. Case studies and archaeological feedback

Using Seismic Full-Waveform Inversion (FWI) for Structural Investigation of the Danevirke

Manuel Zolchow, Daniel Köhn, Dennis Wilken et Wolfgang Rabbel
p. 151-154

Résumé

– Applicability of seismic FWI to archaeological targets with strong topography.

– Reconstruction of building phases and state of preservation using seismic FWI.

Haut de page

Texte intégral

Introduction

1Seismic full-waveform inversion (FWI) is a recently established method within the field of archaeogeophysics, but with the number of examined field data samples still limited, there are many aspects regarding practical application that remain unclear. This study, focused on shear-wave seismic data from the Danevirke (Schleswig Holstein, Germany), aims to analyze the applicability of the method to archaeological targets with strong topography and complex heterogeneous subsurface.

2The Danevirke is a 30-km long fortification system (Fig. 1a) of linear trenches and walls. It was built by the Danes during the Nordic Iron Age as simple wall embankments and enlarged repeatedly (Fig. 1b) during the Viking era. In medieval times, walls and foundations made of local field stones (mostly glacial relocated and modified plutonic rocks like granite) and fired brick, held together by clay or mortar, were integrated into the wall structure, but extensively removed in modern times for reuse as building material. The long building history results in a complex stratigraphy and widely unclear state of preservation. Geophysical prospection methods are essential in view of the size of the wall system permitting selective archaeological excavation at best. Once the electrical resistivity tomography (ERT) results were shown to be strongly affected by soil moisture (Ismaeil, 2008), seismic methods were chosen for further investigation of the building history and state of preservation of the Danevirke.

Figure 1. a) Map of the Danevirke system with blue star indicating the location of the investigation, b) Building phases of the Danevirke (I: Nordic Iron Age, II–III: Viking Age, IV/VI: Medieval Period).

Figure 1. a) Map of the Danevirke system with blue star indicating the location of the investigation, b) Building phases of the Danevirke (I: Nordic Iron Age, II–III: Viking Age, IV/VI: Medieval Period).

Methodology

3The seismic data were recorded with a GEOMETRICS seismic system in SH-configuration along a cross profile of 60 m length with 10-Hz horizontal-component geophones at 1-m distance. Shear waves were excited using a sledge hammer source at intervals of 2 m.

4The data was analysed using two-dimensional, elastic FWI in time-domain, based on inversion of dispersive Love- and refracted SH-waves. The FWI-code DENISE was used for the purpose (Köhn et al., 2012). The data fit was improved by inverting for shear-wave velocity vs and density ρ simultaneously, while the seismic quality factor Qs was assumed to be constant. The required initial model of shear-wave velocity (Fig. 3a) was determined using ray-based first-arrival traveltime tomography (FATT) on the basis of refracted SH-waves using TOMO2D software (Korenaga et al., 2000). Proceeding from this initial seismic velocity model, a density model was calculated via an empirical relation (Ulugergerli & Uyanik, 2007) valid for unconsolidated sands. The constant value of the quality factor Qs = 30 was optimized by best data fit (Fig. 2a) after multiple forward modelling runs with different quality factors. Since the existing code is based on rectangular grids, we used an oversampled spatial discretization of 0.05 m together with the improved vacuum formulation (Pan et al., 2018) to implement accurate free-surface boundary conditions in the presence of strong topography. In order to obtain stable and interpretable results despite the strong non-linearity of the full-waveform inversion, a multi-stage inversion strategy was specifically designed for this dataset. This combines a sequential low-pass filter strategy including alternating offset windowing with a high frequency bandpass filter strategy (Köhn et al., 2019) in order to ensure a similar waveform-adaptation of the Love- and SH-waves. A sensitivity analysis as well as checkerboard tests (Fig. 3b) clearly prove, that the method can theoretically resolve all relevant structures within the rampart construction.

Figure 2. Forward modelled seismograms of the initial model (a) and the FWI-result (b) in comparison to field data.

Figure 2. Forward modelled seismograms of the initial model (a) and the FWI-result (b) in comparison to field data.

Figure 3. a) Smoothed FATT result. b) Checkerboard-Test with 1 m box size and 10% velocity variations relative to the FATT model. c) FWI result d) FWI result in comparison to archaeological records [Andersen 1998]. Identified structures: top layer (T), stone mural (S), mural foundations (F), early wall phases (W1-3), trench structures (G1/G2), infilling (V), possible deeper foundations (A/B).

Figure 3. a) Smoothed FATT result. b) Checkerboard-Test with 1 m box size and 10% velocity variations relative to the FATT model. c) FWI result d) FWI result in comparison to archaeological records [Andersen 1998]. Identified structures: top layer (T), stone mural (S), mural foundations (F), early wall phases (W1-3), trench structures (G1/G2), infilling (V), possible deeper foundations (A/B).

Results and Discussion

5After inversion, the data fit (Fig. 2b) is improved significantly in comparison to the forward modelled synthetic seismogram based on the initial model. This is especially valid for the Love waveform. The result of the FWI (Fig. 3c, d) is structurally comparable with archaeological records (Andersen, 1998). The comparison shows that FWI can reproduce most of the structures that are essential to an understanding of the building history of the Danevirke, including older rampart phases and ditches, as well as various wall sections and foundations made of field stones and bricks. Masonry structures, which could not be sufficiently resolved in the ERT results, probably due to absorption of soil moisture by the compound material, were observable in the FWI outcome as distinct positive anomalies of shear wave velocity.

6A quantitative analysis of the inverted seismic model indicates that shear wave velocity differs significantly for structures made of local field stones (vs = 150–300 m/s) or bricks (vs > 300 m/s). Therefore, masonry of both kinds can be clearly identified without extensive archaeological excavation. Overall, the results show the potential of FWI for the high-resolution characterization of near surface archaeological targets and the advantages of a multi-method approach integrating seismic measurements for interpreting results obtained with other geophysical methods.

Haut de page

Bibliographie

Andersen, H.H., 1998. Danevirke og Kovirke. Arkæologiske undersøgelser 1861-1993. Aarhus University Press, Aarhus.

Ismaeil, A., 2008. Untersuchung des Dannewerk-Hauptwalls mit Hilfe von Geoelektrik, Georadar und Geomagnetik. Diplomarbeit, University of Kiel, Germany.

Köhn, D., De Nil, D., Kurzmann, A., Przebindowska, A., Bohlen., T., 2012. On the influence of model parametrization in elastic full waveform tomography. Geophysical Journal International, 191(1): 325-345.

Köhn, D., Wilken, D., De Nil, D., Wunderlich, T., Rabbel, W., Werther, L., Schmidt, J., Zielhofer, C., Linzen, S., 2019. Comparison of time-domain SH waveform inversion strategies based on sequential low and bandpass filtered data for improved resolution in near-surface prospecting. Journal of Applied Geophysics, 160: 69-83.

Korenaga, J., Holbrook, S., Kent, G., Kelemen, P., Detrick, R., Larsen, H.-C., Hopper, J., DahlJensen, T., 2000. Crustal structure of the Southeast Greenland margin from joint refraction and reflection seismic tomography. Journal of Geophysical Research, 105: 21591-21614.

Pan, Y., Gao, L., Bohlen, T., 2018. Time-domain full-waveform inversion of Rayleigh and Love waves in presence of free-surface topography. Journal of Applied Geophysics, 152: 77-85.

Ulugergerli, E., Uyanik, O., 2007. Statistical correlations between seismic wave velocities and SPT blow counts and the relative density of soils. Journal of Testing and Evaluation, 35(2): 187-191.

Haut de page

Table des illustrations

Titre Figure 1. a) Map of the Danevirke system with blue star indicating the location of the investigation, b) Building phases of the Danevirke (I: Nordic Iron Age, II–III: Viking Age, IV/VI: Medieval Period).
URL http://journals.openedition.org/archeosciences/docannexe/image/9213/img-1.jpg
Fichier image/jpeg, 228k
Titre Figure 2. Forward modelled seismograms of the initial model (a) and the FWI-result (b) in comparison to field data.
URL http://journals.openedition.org/archeosciences/docannexe/image/9213/img-2.jpg
Fichier image/jpeg, 1,2M
Titre Figure 3. a) Smoothed FATT result. b) Checkerboard-Test with 1 m box size and 10% velocity variations relative to the FATT model. c) FWI result d) FWI result in comparison to archaeological records [Andersen 1998]. Identified structures: top layer (T), stone mural (S), mural foundations (F), early wall phases (W1-3), trench structures (G1/G2), infilling (V), possible deeper foundations (A/B).
URL http://journals.openedition.org/archeosciences/docannexe/image/9213/img-3.jpg
Fichier image/jpeg, 853k
Haut de page

Pour citer cet article

Référence papier

Manuel Zolchow, Daniel Köhn, Dennis Wilken et Wolfgang Rabbel, « Using Seismic Full-Waveform Inversion (FWI) for Structural Investigation of the Danevirke »ArcheoSciences, 45-1 | 2021, 151-154.

Référence électronique

Manuel Zolchow, Daniel Köhn, Dennis Wilken et Wolfgang Rabbel, « Using Seismic Full-Waveform Inversion (FWI) for Structural Investigation of the Danevirke »ArcheoSciences [En ligne], 45-1 | 2021, mis en ligne le 16 août 2021, consulté le 10 novembre 2025. URL : http://journals.openedition.org/archeosciences/9213 ; DOI : https://doi.org/10.4000/archeosciences.9213

Haut de page

Auteurs

Manuel Zolchow

Corresponding author, Institute of Geosciences, Kiel University, Otto-Hahn-Platz 1, 24098 Kiel

Daniel Köhn

Institute of Geosciences, Kiel University, Otto-Hahn-Platz 1, 24098 Kiel

Dennis Wilken

Institute of Geosciences, Kiel University, Otto-Hahn-Platz 1, 24098 Kiel

Articles du même auteur

Wolfgang Rabbel

Institute of Geosciences, Kiel University, Otto-Hahn-Platz 1, 24098 Kiel

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search