Testing the advantages of ultra-dense GPR antenna arrays for the Archaeological Prospection
Roland Linck 1,2* , Andreas Stele 1, Hans-M art in Schuler
3
1
Bavarian State Depart ment of M onument s and Sites, Archaeological Prospect ion, M unich, Germany
2
Ludw ig M aximilians Universit y, Depart ment for Eart h and Environmental Sciences, Sect ion for Geophysics, M unich, Germany
3
IGM Ingenieurgesellschaft für Geophysikalische M esstechnik mbH, Überlingen, Germany
* Contact details: roland.linck@blfd.bayern.de
Introduction:
During t he last few years, t he use of Ground-Penet rat ing-Radar (GPR) mult i-channel antenna arrays in t he Archaeological Prospect ion increased dramat ically. The main advantage of t his t ype of survey is a much faster data acquisit ion combined w it h a dense profile spacing. However, most of t he
common mult i-channel arrays consist of antennas w it h a spacing of not smaller t han 8 cm. The aim of our test survey was to evaluate how an even denser spacing of 4 cm, t hat is provided by t he IDS St ream-C GPR device at a cent re frequency of 600 M Hz, can improve t he detect ion of faint
archaeological feat ures.
Test site:
As a test site, t he Great Roman Bat h of Kempten-Cambodunum was chosen.
Kempten is located in t he sout hwestern corner of Bavaria in a region called
Allgäu, ca. 80 km sout hwest of Augsburg (Fig. 1).
Fig. 1: Topographical map show ing t he
locat ion of Kempten (© Geobasis data:
Bayerische Vermessungsverw alt ung)
From an archaeological point of view, Cambodunum was t he first capital of
t he Roman province Rhaet ia. It was founded in t he first cent ury AD and soon,
most of t he cit y was const ructed as stone buildings.
One of t he most
prominent ones had
been t he Great Bat h
t hat covered an area of
4000 m² and hence
belonged to t he hugest
complexes of t his kind
nort h of t he Alps
(Gott lieb & Weber,
1989; Weber, 2000).
The
reason
for
choosing t his building
as a GPR test site, is
t hat several rooms st ill
have part ly preserved
hypocaust pillars (Fig.
2),
w hich
are
challenging to map Fig. 2: M ap (a) and photo (b) of t he excavated remains of t he Great
w it h standard GPR Roman Bat h of Cambodunum show ing t he w alls and t he locat ion of
t he hypocaust pillars. The excavat ion took place already in 1911 and
devices.
w as refilled afterw ards.
w w w.blfd.bayern.de
Survey results:
The test survey was executed on a 40 x 40 m
grid covering t he main part of t he former
Roman Bat h house. The same area was
measured simultaneously wit h t he IDS St reamC mult ichannel array and a standard single
channel IDS Duo to compare t he corresponding
dept h slices (Fig. 5a).
The St ream-C data shows a mult it ude of small
hypocaust pillars w it h a lateral lengt h of 25 cm
(Fig. 5b). In total, around 22% of t he former
excavated pillars can be mapped in t he GPR
data. The reason for t his, at a first glance, quite
low value is t hat t he hypocaust s consisted of
unstable layered bricks t hat collapsed during
excavat ion or refill. Furt hermore, t he original
excavat ion map also has some mistakes in t he
locat ion of t he pillars, as not all of t hem were
exact ly levelled and simply added in a regular
raster. Such deviat ions of t he real posit ion of
t he archaeological remains and t heir mapped
ones are also documented several t imes during
modern
excavat ions
in
Cambodunum .
Nevert heless, more t han 89% of t he potent ial
hypocaust pillars ident ified in t he GPR data
correspond to t he excavat ion map. The
remaining error is condit ioned by t he fact t hat
in some cases, t he GPR data simply shows
bigger stones of t he infill. The effect of t he
excavat ion’s refilling w it h t he digging can be
Fig. 5: a) Comparison of t he GPR dept h slices of 10 cm t hickness in 65-75 cm dept h of t he IDS
seen in t he shallower dept h slices up to 50 cm
St ream-C (left ) and t he IDS Duo (right ). (b) A sample hypocaust picked in t he 65-75 cm dept h slice
below modern surface t hat show an ext remely
and t he corresponding inline and crossline profiles.
dist urbed signat ure inside t he building.
Looking at t he single channel dataset of t he IDS Duo, reveals t hat , for sure, t he well preserved walls of t he Roman Bat h are
visible too (Fig. 5a right ). These feat ures are not t reated furt her, as t hey are already published comprehensively in Linck
(2013). However, in t his case only 3% of t he hypocaust pillars can be ident ified in t he corresponding dept h slices. Hence, such
small archaeological remains can only be mapped by ult ra-dense GPR antenna arrays and w ill be missed by standard single
antenna systems. In case of expected small-scale subsurface st ruct ures, t he applicat ion of an antenna array is t hus advisable.
w w w.geophysik.uni-muenchen.de
Survey device IDS Stream-C:
Fig. 3: Layout of t he IDS
St ream-C antenna array.
The IDS St ream-C mult ichannel array consist s of 23 antenna dipoles in VV
configurat ion t hat are 4.4 cm spaced and 9 HH dipoles w it h 10 cm spacing
(Fig. 3). The antenna frequency is 600 M Hz. Hence, t he VV dipoles have a
distance below t he Nyquist interval and t he corresponding data is fully 3D.
Wit hin one survey swat h, a coverage of 96 cm can be achieved (Fig. 4). As
t he HH data does not show more informat ion t han t he VV one due to t he
dense sample spacing, only t he latter is presented here.
Fig. 4: The IDS
St ream-C
array
mounted onto a
self-built sled in
act ion at t he t est
site in Kempten.
References:
Gott lieb, G., & Weber, G. (1989). Die Frühgeschichte Kemptens. In V. Dotterweich, K. Filser, P. Fried, G.
Gott lieb, W. Haberl, & G. Weber (Eds.), Geschichte der Stadt Kempten (pp. 1-68). Verlag Tobias
Dannheimer.
Linck, R. (2013). M et hodische Untersuchungen zur Weiterent w icklung der Boden- und
Satellitenradarprospekt ion in der Archäologie. Shaker Verlag.
Weber, G. (2000d). Bauboom in Stein - eine Stadt nach mediterranem Vorbild. In G. Weber (Ed.),
Cambodunum - Kempten. Erste Haupt stadt der römischen Provinz Raet ien? (pp. 49-80). Verlag Philipp
von Zabern.
w w w.igm-geophysik.de