Skip to navigation – Site map

HomeNuméros45-12. Methods and innovationsHighlighting the Potential of 3D ...

2. Methods and innovations

Highlighting the Potential of 3D ERT by Comparing its Results with GPR and the Excavation Map of a Roman Building

Mandana Parsi, Roland Linck, Jörg W. E. Fassbinder and Michael Eitel
p. 183-186

Abstract

– Comparative measurements and analysis of 3D ERT and GPR results.

– Evaluation of different ERT data processing methods.

Top of page

Full text

Historical background

1The villa rustica near Peiting (Bavaria, Germany) is situated on a plateau 1 km east of the Lech river, which is followed in the west by an ancient Roman road called the Via Claudia. The first known document concerning the existence of an archaeological site in this location is a letter from 1837. Boulders in the field indicated the presence of subsurface stone buildings, leading the Bavarian State Department of Monuments and Sites to excavate parts of the site in 1957. The excavation was later backfilled and covered with topsoil. Road construction in 1990 and the laying of a gas pipeline necessitated further excavations of a bathhouse constituting part of the villa. It was found then that the villa had been built in the 2nd century AD and was substantially rebuilt after being destroyed in the mid-3rd century AD, before being ultimately abandoned in the 4th century AD when the Romans left the province of Raetia for good (Leicher, 2018). The site, like all typical Roman villa complexes, consists of the main building with a layout more common for central Italy, a bathhouse, several outbuildings and storage units in the near environs and a courtyard wall. The bath was equipped with a Roman hypocaust heating system.

Field survey

2Jörg Fassbinder conducted the first geophysical prospection (resistivity mapping with Geoscan-RM15) in this area in 2002. In the summer of 2020, the site was revisited and further prospection methods were applied, including a large-scale survey with ground penetrating radar (GPR). Magnetic prospection was not an option because of the presence of a pipeline, an electrical power cable and two main streets next to the research area. Part of the main building was surveyed with 3D ERT to obtain more detailed results.

3A comparison of the capabilities of the GPR and ERT methods, as applied at the site, is the focus of this paper. The instruments used in this research were a GSSI SIR 4000 with a 400 MHz antenna for GPR and a Lippmann 4-point light 10 W for ERT. In addition, the 3D ERT result was verified with an excavation map to establish the accuracy and precision of the result and hence also the potential of the particular instrument for ERT measurements of ancient sites. Moreover, a portable Time Domain Reflectometry (TDR) device was used to monitor the soil temperature, moisture and conductivity during the survey in question.

Results

4For the GPR measurement, a 400 MHz antenna was used to obtain a good balance between resolution and penetration in an 80 × 80 m grid north and east of the Roman bath. The GPR result shows significant substructures made of stone; these are obviously part of the main building of the Roman villa (Fig. 1 for part of the measurements in this area). A suitable area was chosen based on these results to continue with 3D ERT.

Figure 1. GPR depth slice of an ancient Roman building, depth layer 60-80 cm. The red rectangle shows the area chosen for ERT prospection (GSSI SIR 4000 with 400 MHz antenna, sample interval 6 x 50 cm, interpolated to 25 x 25 cm).

Figure 1. GPR depth slice of an ancient Roman building, depth layer 60-80 cm. The red rectangle shows the area chosen for ERT prospection (GSSI SIR 4000 with 400 MHz antenna, sample interval 6 x 50 cm, interpolated to 25 x 25 cm).

5The ERT method calls for a DC current to be emitted in order to measure the voltage and derive from it the resistivity distribution of the subsurface (Schmidt, 2013). To proceed with 3D ERT, we decided to measure several parallel 2D profiles and collate the data. To obtain the most accurate result, we used both robust and smoothness-constrained inversion and the forward resistivity calculations were based on the finite-element method. Within two days 33 parallel, west-to-east profiles of 20 m length were measured with a dipole-dipole configuration. The spacing between electrodes on each profile was 0.5 m. Furthermore, for better 3D resolution, we chose a 0.5 m spacing between parallel profiles. Figure 2 illustrates three ERT depth layers and the corresponding GPR depth slices.

Figure 2. Comparison of GPR and 3D ERT depth layers (ERT: Lippmann 4-point light 10 W in dipole–dipole configuration, sample interval 50 x 50 cm). ERT (L1) is the result of the robust inversion and ERT (L2) is the result of the smoothness-constrained inversion. RES3DINV software was used for the calculation of inversions. Dashed lines illustrate soil compression due to ancient walls (either removed or destroyed) in the ERT results. The last column illustrates a combined interpretation based on geophysical data and excavation.

Figure 2. Comparison of GPR and 3D ERT depth layers (ERT: Lippmann 4-point light 10 W in dipole–dipole configuration, sample interval 50 x 50 cm). ERT (L1) is the result of the robust inversion and ERT (L2) is the result of the smoothness-constrained inversion. RES3DINV software was used for the calculation of inversions. Dashed lines illustrate soil compression due to ancient walls (either removed or destroyed) in the ERT results. The last column illustrates a combined interpretation based on geophysical data and excavation.

6Both instruments show in each depth layer walls and part of the floor of a rectangular substructure made of stone. These walls are located at depths of 30 cm and 90 cm. Two different types of data processing were applied to the ERT data. Robust inversion (L1 norm) first in view of the expected sharp boundaries for the anomaly. The absolute error for this type of inversion for this dataset with five iterations was 1.26 %. Smoothness-constrained inversion (L2 norm) was applied next to detect the smooth variation of resistivity values. The RMS error for this inversion with three iterations was 3.78 %. The latter helped to detect the changes in soil density corresponding to compressed fill in trenches left by destroyed walls, which could thus be traced.

7Moreover, as some soil characteristic changes can play an important role in both GPR and ERT prospection, we monitored the topsoil moisture, temperature and conductivity with the TDR instrument. The soil moisture percentage was in the range of 37 % with a maximum change of 0.62 %; the maximum soil temperature change was 2.4°C and the topsoil conductivity was stable with a maximum change of 0.33 dS/m.

8Figure 3 shows the final interpretation based on the initial shallow excavation in 1957, and the GPR, RM15 and ERT prospection results. Red lines here are the general interpretation lines of the substructures; the blue lines show sections partly detected by geophysical methods, destroyed either after the excavation by construction work or located outside of the survey areas; and the green lines illustrate substructures detected by RM15.

Figure 3. Final interpretation of the substructure. Red lines show the general interpretation lines; blue illustrates the part mapped only by excavation (partly detected remnants in the geophysical survey results) and green represents the substructures detected by RM15 in 2002, which lay outside of the GPR grid and hence were not covered by 3D ERT.

Figure 3. Final interpretation of the substructure. Red lines show the general interpretation lines; blue illustrates the part mapped only by excavation (partly detected remnants in the geophysical survey results) and green represents the substructures detected by RM15 in 2002, which lay outside of the GPR grid and hence were not covered by 3D ERT.

Discussion and conclusion

93D ERT provides the same excellent result as GPR prospection in this area. Therefore, 3D ERT is a trustable substitute for GPR wherever field circumstances are unsuitable for the latter method. Both instruments have their limitations. GPR is great to detect stone-made features, but its signal will be dampened dramatically in clayey soil. ERT is a reliable instrument to map the underground substructure in detail, but applying 3D ERT is highly time consuming.

10Moreover, soil without moisture makes it impossible to conduct the study efficiently because the ERT electrodes sometimes have high contact resistance. Consequently, a TDR measurement is important, because it renders it possible to monitor the moisture percentage during the ERT measurement. In this research, TDR results show stable and suitable soil conditions during the day.

11The results of both instruments illustrate a rectangular stone-made substructure with some inner rooms, located at a depth between 30 cm and 90 cm. Some walls were destroyed or removed during the first excavation and they can be recognized just barely by the different, greater density compared to the periphery soil, a so-called “ghost feature” that could be overlooked in routine excavation practice (Schleifer, 2004).

12In the data processing for ERT, these features are traced mainly thanks to smoothness-constrained inversion, hence the obvious necessity of applying different data processing methods for any given dataset. Operating a combination of different geophysical methods is therefore essential in order to model the initial shape of substructures (especially when they are only partly preserved).

Top of page

Bibliography

Leicher, J., 2018. Die Villa Rustica von Peiting: Die Chronik der Wiederauferstehung eines antiken Landgutes.

Schleifer, N., 2004. Ghost features: a proposal for appropriate management and a forum for discussion. In Newsletter of the International Society of Archaeological Prospection, Bradford. 1, 6-8.

Schmidt, A., 2013. Earth resistance for archaeologists. AltaMira Press.

Top of page

List of illustrations

Title Figure 1. GPR depth slice of an ancient Roman building, depth layer 60-80 cm. The red rectangle shows the area chosen for ERT prospection (GSSI SIR 4000 with 400 MHz antenna, sample interval 6 x 50 cm, interpolated to 25 x 25 cm).
URL http://journals.openedition.org/archeosciences/docannexe/image/9485/img-1.jpg
File image/jpeg, 3.8M
Title Figure 2. Comparison of GPR and 3D ERT depth layers (ERT: Lippmann 4-point light 10 W in dipole–dipole configuration, sample interval 50 x 50 cm). ERT (L1) is the result of the robust inversion and ERT (L2) is the result of the smoothness-constrained inversion. RES3DINV software was used for the calculation of inversions. Dashed lines illustrate soil compression due to ancient walls (either removed or destroyed) in the ERT results. The last column illustrates a combined interpretation based on geophysical data and excavation.
URL http://journals.openedition.org/archeosciences/docannexe/image/9485/img-2.jpg
File image/jpeg, 349k
Title Figure 3. Final interpretation of the substructure. Red lines show the general interpretation lines; blue illustrates the part mapped only by excavation (partly detected remnants in the geophysical survey results) and green represents the substructures detected by RM15 in 2002, which lay outside of the GPR grid and hence were not covered by 3D ERT.
URL http://journals.openedition.org/archeosciences/docannexe/image/9485/img-3.jpg
File image/jpeg, 2.1M
Top of page

References

Bibliographical reference

Mandana Parsi, Roland Linck, Jörg W. E. Fassbinder and Michael Eitel, “Highlighting the Potential of 3D ERT by Comparing its Results with GPR and the Excavation Map of a Roman Building”ArcheoSciences, 45-1 | 2021, 183-186.

Electronic reference

Mandana Parsi, Roland Linck, Jörg W. E. Fassbinder and Michael Eitel, “Highlighting the Potential of 3D ERT by Comparing its Results with GPR and the Excavation Map of a Roman Building”ArcheoSciences [Online], 45-1 | 2021, Online since 16 August 2021, connection on 18 January 2026. URL: http://journals.openedition.org/archeosciences/9485; DOI: https://doi.org/10.4000/archeosciences.9485

Top of page

About the authors

Mandana Parsi

Geophysics, Department of Earth and Environmental Sciences, Ludwig-Maximilians-University, Theresienstrasse 41, 80333 Munich, Germany

By this author

Roland Linck

Geophysics, Department of Earth and Environmental Sciences, Ludwig-Maximilians-University, Theresienstrasse 41, 80333 Munich, Germany; Bayerisches Landesamt für Denkmalpflege (BLfD), Munich, Germany

By this author

Jörg W. E. Fassbinder

Geophysics, Department of Earth and Environmental Sciences, Ludwig-Maximilians-University, Theresienstrasse 41, 80333 Munich, Germany

By this author

Michael Eitel

Geophysics, Department of Earth and Environmental Sciences, Ludwig-Maximilians-University, Theresienstrasse 41, 80333 Munich, Germany

Top of page

Copyright

CC-BY-NC-ND-4.0

The text only may be used under licence CC BY-NC-ND 4.0. All other elements (illustrations, imported files) may be subject to specific use terms.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search