close

Вход

Забыли?

вход по аккаунту

?

Electrical and magnetic response of archaeological features at the early neolithic site of Movila lui Deciov western Romania.

код для вставкиСкачать
Archaeological Prospection
Archaeol. Prospect. 11, 213–226 (2004)
Published online 4 November 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/arp.234
Electricaland Magnetic Response of
Archaeological Features at the Early
Neolithic Site of Movilalui Deciov,
Western Romania
J. M. MAILLOL,1* D. L. CIOBOTARU2 AND I. MORAVETZ3
1
Department of Geology and Geophysics, University of Calgary, 2500 University Drive NW,
Calgary, AlbertaT2N1N4, Canada
2
Directia pentru Cultura, Culte si Patrimoniul Cultural National Timis,Timisoara, Romania
3
Department of Archaeology, University of Calgary, 2500 University Drive NW, Calgary,
AlbertaT2N1N4, Canada
ABSTRACT
An archaeologicalgeophysics survey was conducted on the early neolithic site of Movila lui Deciov, in
the Province of Banat, Romania. Magnetometry and electromagnetic terrain conductivity were used
for the main prospection effort, and a test of electricalresistivity imaging was conducted on a selected
profile.In addition, magnetic susceptibilitymeasurementswere obtained from excavation pit samples.
The magnetic survey was successfulin determining the extent of the site, in delimiting zones rich in
structures and artefacts, and in confirming the presence of a ditched enclosure that could be the earliest known in the region.The electromagnetic survey was limited by a lack of resolution of electrical
property contrast. Detailed joint modelling of the magnetic and electrical response of the subsurface
was used to confirm that electrical resistivity imaging can provide depth information to complement
magnetic mapping.
One of very fewreportedin Romania, this surveypavesthe way foranincreased use ofgeophysical
techniques in the cultural heritage management of this country. From a methodological viewpoint,
this work further demonstrates the potential of electrical resistivity imaging in archaeology.Copyright
2004 JohnWiley & Sons,Ltd.
Key words: magnetometry; electrical resistivity imaging; modelling; magnetic susceptibility;
Romania; neolithic
Introduction
Movila lui Deciov (Deciov’s knoll) is an early
neolithic site in western Romania (Banat Region).
To date very few archaeological geophysics surveys have been carried out and reported in
Romania, and the initial objectives of this study
were mainly those typical of archaeological pro* Correspondence to: J. M. Maillol, Department of Geology
and Geophysics, University of Calgary, 2500 University Drive
NW, Calgary, Alberta T2N 1N4, Canada.
E-mail: maillol@ucalgary.ca
Copyright # 2004 John Wiley & Sons, Ltd.
spection projects. Previous unpublished investigations and test excavations have revealed,
unambiguously, the presence of a neolithic, Starčevo-Körös-Criş occupation site, but no clear
appreciation of how widespread the site was
could be deduced. The first prospection objective
therefore was to attempt to delineate the total
extent of the site by mapping the abundance of
anomalies that could be related to archaeological
material.
An extremely enticing discovery of the test
excavations was evidence of a ditch feature,
possibly representing the trace of an enclosure
Received 23 August 2003
Accepted 3 August 2004
J. M. Maillol et al.
214
surrounding the entire site. Although ditched
enclosures are common features of neolithic
settlements (Whittle, 1994), they have never
been found around sites of the period of Movila
lui Deciov. When present they are often mapped
very successfully using geophysical techniques,
particularly magnetometry (Tabbagh et al., 1988;
Doneus and Neubauer, 1999). Another very
important objective of this survey was therefore
to attempt to confirm the presence of an enclosure completely surrounding the site and to map
its characteristics.
From a methodological point of view it was
also interesting to evaluate the potential benefits
of using multiple techniques on such a site, and
to integrate their findings so as to achieve an
understanding of the geophysical response going
beyond the mere superposition of data sets of
different origin. Magnetometry and area soil
resistivity are used routinely for the mapping
of archaeological sites. Although very efficient
for fast and accurate mapping, these methods
cannot in general provide an accurate image of
the vertical distribution of archaeological materials. In many cases, archaeologists rely on georadar for precise depth determination. For sites
such as Movila lui Deciov, however, and in fact
for the great majority of prehistoric sites located
on agricultural land and devoid of stone structures, the presence of fine-grained, clayey and
moist soil and sedimentary material often preclude the use of georadar because of the high
signal attenuation. In these circumstances, direct
current (DC) electrical resistivity imaging can
provide the best and only alternative to georadar.
This relatively recent technique is now widely
used in near-surface geological investigations.
Although a few examples of archaeological
application have been reported (Noel and Xue,
1991; Sambuelli et al., 1999; Neighbour et al.,
2001), it still appears somewhat underutilized
in this field. A test of this technique and how it
could be utilized in a central European neolithic
context constituted an important methodological
objective of this work.
Archaeological setting
Movila lui Deciov is a multicomponent StarčevoKörös-Criş site in the southeastern periphery of
the Great Hungarian Plain, within the Timiş
Province of Romania, just north of the Dudeştii
Vechi village (Figure 1). This area has a limited
relief and lies generally below 100 m a.s.l.
(Chapman, 1981). It was prone to periodic flooding prior to flood-protection works undertaken
in the twentieth century. Sedimentation from the
periodic floods produced nutrient-rich soils,
making the area attractive for early agriculture.
The site is in a modern agricultural field and
rises 3 m over an area of 200 m in diameter.
Gornea Aranka, the southern tributary of the
River Aranka, which has its source 40 km
Figure1. Map of Romania and surrounding countries with site location.
Copyright # 2004 John Wiley & Sons, Ltd.
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
northeast of Dudeştii Vechi, is just north of the
site. The River Aranka is part of the Tisza drainage system, which flows into the River Danube.
Present-day vegetation is composed almost
entirely of agricultural crops and weeds; postglacial vegetation cover consisted of mixed-oak
forests similar to those noted in Hungary
(Zolyomi, 1953).
Movila lui Deciov is an important stratified site
first excavated at the turn of the twentieth century. Excavations by the local collector Nagy
Gyula Kislegi in 1906 and 1907 resulted in the
recovery of many complete artefacts currently
located at Muzeul Banatului, in Timişoara,
Romania. Recent test excavations in 2000 and
2001 identified two Starčevo occupations:
(i) a lower occupation between 140 cm and 160 cm
below the surface consisting of a cultural floor
of artefacts, charcoal, ash and fish scales;
(ii) an upper occupation level between 95 cm
and 110 cm below the surface represented
by a house floor feature and artefacts.
The discovery of the two Starčevo occupations
makes Movila lui Deciov an important site for a
temporal analysis of culture change. Preliminary
examination of pottery materials suggests an
215
occupation between the Starčevo IIB and
Starčevo IVA periods, following the chronological systematics of Lazarovici (1979). This period
of occupation encompasses the timeframe of
5950–5100 BC, corresponding to Manson’s (1995)
calibrated dates.
Architectural features uncovered so far at
Movila lui Deciov consist exclusively of surface
dwellings, a peculiarity that stands apart from
the typical Starčevo-Körös-Criş pit-house dwellings associated with the deeper occupation
levels. The site is also important in the presence
of a ditched enclosure. Future excavation will
reveal the cultural association with this ditch,
which at this time remains speculative. If this
ditch is associated with the Starčevo-Criş occupation as the preliminary archaeological data
indicates, it would be the earliest in the region.
Data acquisition and results
All measurements were collected with reference
to the same grid. The grid (Figure 2) consisted of
individual survey lines orientated W–E and
separated by 1 m. The spatial sampling interval
along each line varied according to the method
Figure 2. Oblique aerial photograph of the site with survey grid overlain. Solid lines indicate the area of the magnetic
map of Figure 3. Distances are in metres.
Copyright # 2004 John Wiley & Sons, Ltd.
Archaeol. Prospect. 11, 213–226 (2004)
216
used. For most of the survey area, square subgrids 25 m 25 m were used and surveyed independently. In some cases, rectangular subgrids
were used to achieve a more efficient coverage of
the area. Three geophysical methods were used
for this study: magnetometry, ground electromagnetic conductivity and DC electrical resistivity imaging. As a first step in the exploitation of
the data, the results from all three methods are
presented separately. The following section provides integration of the results. For a detailed
discussion of the practical and data exploitation
aspects of multimethod archaeological surveys
see Bevan and Roosevelt (2003).
Magnetic survey
Magnetometry relies on measurements of the
strength and direction of Earth’s magnetic field.
When materials with magnetic properties different from the geological background are present
in the subsurface, they create magnetic field
anomalies superimposed on Earth’s main magnetic field. The mapping of these anomalies
reveals the presence and some characteristics of
the materials possessing these abnormal magnetic properties. The magnitude of the magnetic
anomalies is related directly to the magnetic
susceptibility (MS) and remanent magnetization
of the material. The instrument used for this
survey was a Scintrex Envimag proton precession magnetometer with a single sensor measuring the total strength of Earth’s magnetic field. In
order to achieve maximum sensitivity, the sensor
was held close to the ground, at an average
elevation of 10–15 cm. This distance was the
minimum height required to ensure that at no
time the sensor would hit protuberances of the
ground surface, which would result in noisy
measurements. An automatic acquisition mode
was used based on a 1 s cycling time. Keeping a
very slow constant walking pace, this corresponded to an average measurement interval of
0.20 m. It is very difficult to directly assess the
repeatability of measurements for total field
magnetometers because of the independent
effect of diurnal variation. However, seven series
of 10 repeat measurements were taken at the
main base station on different days, and the
maximum deviation among the readings was
Copyright # 2004 John Wiley & Sons, Ltd.
J. M. Maillol et al.
never found to exceed 0.25 nT. As some of that
variation is due to diurnal variation, we estimate
that the resolution of the instrument must have
been better than 0.25 nT.
A loop technique was used for diurnal corrections and a base station was reoccupied after
walking two survey lines; a time interval of
4 min on average and never exceeding 6 min.
To minimize the walking distances and the
time intervals between successive base-station
readings, seven different base stations were
used for the entire grid. These seven stations
were tied together with 4.7 min separating the
first and last reading. An area of 12 600 m2 was
covered by the magnetic survey.
The total field measurements were processed
to obtain a complete map of the surveyed area.
The first step was to apply a time-based correction to correct for temporal variations of Earth’s
magnetic field. This was done in the field for
individual subgrids using their corresponding
base-station readings. An overall correction
using the ties between individual base stations
was later applied to reduce all measurements to a
common datum. No other correction was needed
owing to the limited extent of the survey area
and to the very flat topography.
Despiking was used to remove most very
large, short wavelength anomalies attributable
to small, recent iron-rich objects such as nails or
fence wire lying at or very close to the surface.
These objects were remarkably few in this particular agricultural field. Detrending and subgrid
boundary adjustments were then performed
resulting in a seamless map of the surveyed
area, devoid of large-scale, long spatial wavelength characteristics. Finally a light high-cut
filter was applied to enhance the continuity of
features.
The final total field map is shown in Figure 3.
Numerous anomalies are seen with a maximum
range of variation of 65 nT. The distribution of
anomalies defines very clearly the limits of the
site, which does not appear to extend beyond
the survey area. The most significant feature of
the map is possibly the presence of an approximately oval and apparently mostly positive
anomaly, with a long axis measuring 80 m and
a short axis measuring 62 m. This feature seems
to completely surround the majority of smaller
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
217
Figure 3. (a) Total field magnetic anomaly map. (b) Magnetic map with overlay showing: (1) outline of the anomaly interpreted as a
ditched enclosure; (2) test excavations of 2001 season; (3) location of test electrical resistivity imaging profile; (4) present-day
farm road.The straight NW^SE anomalies particularly visible in the southwest part of the map are caused by deep plough furrows.
Copyright # 2004 John Wiley & Sons, Ltd.
Archaeol. Prospect. 11, 213–226 (2004)
218
scale anomalies, and it intersects the excavation
trench where evidence of a ditch was found
(Figure 3b). It is therefore natural to interpret it
as a ditched enclosure. The magnetic signature of
such a large and heterogeneous feature would be
a composite of the induced magnetization of
topsoil infill, and the numerous dipole anomalies
due to the induced and remanent magnetization
of small individual objects or debris. At this
latitude, the overall effect would be a mostly
positive anomaly. This could constitute the
most important archaeological contribution of
this study, because such an enclosure would be
the earliest discovered in this part of the world.
A number of high-amplitude anomalies are
also very interesting; e.g. at grid coordinates
90 m and 50 m (Figure 3). The test excavations
have revealed the presence of house features with
burned floors. Magnetic susceptibility measurements of house material give very large values
consistent with large magnetic anomalies (see
next section). It is therefore likely that the largest
anomalies, both in amplitude and extension, are
created by house features, which will constitute
targets of choice for future excavations.
All the other smaller scale anomalies almost
certainly reveal the presence of archaeological
features and artefacts, such as individual ceramics objects, ovens, pits, or house fragments.
Very clear patterns of anomalies are also found
outside the oval enclosure, in particular in the
southwest corner of the area surveyed. Without
further archaeological investigations it is impossible to know how they relate to the material
inside the enclosure.
Magnetic susceptibility measurements
The magnetic susceptibility of soil and archaeological samples was measured in the laboratory
using a Bartington MS2 meter and MS2B sensor.
A total of 29 samples collected from the 2000 and
2001 test excavations were used. These samples
were collected initially for sedimentological and
phytolith analysis and not specifically for measurements of their magnetic properties. In addition, two hand samples of house material (daub)
were recovered during the 2002 geophysical
survey, and subsequently subsampled for magnetic analysis.
Copyright # 2004 John Wiley & Sons, Ltd.
J. M. Maillol et al.
Figure 4. Distribution of magnetic susceptibility measurements of 31 excavation samples. The number of samples in
each category is indicated above the corresponding column.
The results of these measurements are summarized in Figure 4 in the form of a histogram
of values of low frequency volume specific
magnetic susceptibility (k). The majority of soil
samples have k in the range 0.2 103 to
1 103 u.SI with some values reaching up to
3 103 u.SI. One of the excavation samples
containing reddish material suggestive of burning or archaeological debris has a large k of
11 103 u.SI. The two daub samples have elevated but very different susceptibilities. Here
also the larger value corresponds to a mostly
red sample with clear evidence of heating,
whereas the weaker sample is mostly grey and
probably essentially unmodified. The bulk of the
soil samples provide an estimate of the background magnetic susceptibility between the surface and a depth of 2 m. The mean value for the
27 weakest samples (Figure 4) is 1.03 103 u.SI.
The daub samples and the more strongly magnetic excavation samples provide estimates of
the k values for actual archaeological material.
Although only four such samples are available,
the results indicate that in this case k can vary
between 3.7 103 and 14.6 103 u.SI. Because
of the mode of sampling and the difficulty to
extrapolate imperfect laboratory measurements
to in situ susceptibility values, these numbers
represent only estimates. However, the differences between extreme values and the background value provide some useful constraints
on the possible k contrast responsible for the
magnetic anomalies detected and mapped by
the surveys. These constraints will be used in a
later section to reach a better understanding of
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
the geophysical response of the materials present
at the site.
Electromagnetic survey
A conductivity map was produced with a lowinduction number electromagnetic unit, Geonics
EM38, which directly measures the apparent
electrical conductivity of the ground. The operator followed exactly the same survey lines as for
the magnetic survey. Measurements were carried out in ‘step mode’ with a 0.5 m interval, the
operator stopping at each station. The EM38 was
used in ‘vertical dipole mode’ and held parallel
to the survey lines. In this configuration, the
instrument sensitivity is highest for a depth of
investigation equal to the coil separation, 1 m in
this case. As the magnetic survey preceded the
conductivity survey, only the portion of the full
survey grid shown to contain significant magnetic anomalies was covered with the EM38, a
total of 8000 m2.
The EM38 conductivity measurements were
compiled and processed in a similar fashion to
the magnetic measurements, except for the
temporal corrections not needed for an active
219
method. Each 25 25 m square was processed
individually using the following steps. A 0.25 0.25 m interpolation of the original 1 m 0.5 m
data was performed. A high-pass filter was
applied by subtracting the result of a 7.5 7.5 m
moving average from the original data. The
resulting effect of applying such a filter is to
effectively remove any large-scale conductivity
variations, possibly geological in origin. Any
differences between subgrids due to slightly
different calibrations applied on different days
or to an overall change in moisture conditions are
also eliminated or at least greatly reduced. The
subgrids were then all combined together to
produce a conductivity map. To give more meaning to inherently relative values, the average of
all measurements for the entire survey area was
added to each of the filtered values; after this
step, values ranged from 17 to 36 mS m1.
The complete conductivity map (Figure 5) is
disappointing because it shows very few significant features. The most apparent characteristics
are the road and the clear effect of plough furrows seen as very straight NW–SE stripes of
alternatively high and low conductivity values.
The only clearly significant feature of the
Figure 5. Terrain conductivity map (EM38): (a) field data; (b) filtered data. The circle indicates the only really significant feature,
which also corresponds to a very strong magnetic anomaly seen on Figure 3.The NW^SE streaks correspond to plough lines.
Copyright # 2004 John Wiley & Sons, Ltd.
Archaeol. Prospect. 11, 213–226 (2004)
220
conductivity map is located in the southwest of
the surveyed area. It coincides exactly with the
very strong magnetic anomaly interpreted earlier
as a house feature. We would expect such compact, low porosity material to have a low conductivity (high resistivity), and this is confirmed
by the resistivity section, which shows generally
high resistivity zones colocated with magnetic
anomalies (see next section). However, the conductivity measurements indicate the opposite
behaviour. The absence of additional features
on the conductivity map is somewhat disappointing because the archaeological material is
expected to lie at a depth of about 1.5 m or less,
which is exactly in the range of the EM38. These
apparent contradictions will be examined in the
discussion section.
Resistivity imaging
A test of two-dimensional resistivity imaging
was conducted along a profile selected on the
basis of the preliminary results of the magnetic
survey in order to intersect prominent features.
An L&R Minires resistivity unit was used. This
system is a four-electrode, manual recording
unit, so the required combinations of electrode
positions were achieved by manually moving the
four electrodes. A dipole–dipole array was used
with a fixed 1 m potential and current dipole size
and an inter-dipole separation varying from
0.5 m to 5 m. For each dipole separation a measurement interval of 0.25 m was used to cover the
entire profile. A total profile length of 24 m was
acquired yielding 383 data points. Resistance
values were measured ranging from 0.659 to
11.559 , well above the sensitivity limit of the
instrument.
The dipole–dipole resistance measurements
were converted to apparent resistivity and
inverted into a resistivity model. The inversion
was performed with the software RES2DINV
(Loke and Barker, 1996). A good fit between
observed and predicted apparent resistivities is
easily achieved, which indicates a coherent data
set with little noise. The reconstructed resistivity
model and a tentative interpretation are shown
in Figure 6. The range of model resistivity is 15 to
146 -m, or to 7 to 67 mS m1, which agrees well
with the range of apparent conductivities mea-
Copyright # 2004 John Wiley & Sons, Ltd.
J. M. Maillol et al.
sured by the EM38 (17–36 mS m1). A generally
high resistivity layer from approximately 0.3 to
1.5 m depth is readily apparent, and significant
lateral changes in resistivity are present in this
layer. Underneath, lower resistivities typical of
fine-grained sediment are found. Between 5 and
9 m (horizontal distance), this lower resistivity
layer is interrupted by a funnel-shaped higher
resistivity zone extending at least down to the
depth limit of the section (3 m).
The profile was selected to intersect major
magnetic anomalies and a brief comparison of
resistivity image and magnetic map reveals that
a very good correlation exists between high
resistivity zones and magnetic anomalies. A
detailed comparison of magnetic and resistivity
results is discussed later (see Figure 8). For
comparison purpose, it is important to take into
account that the positive peaks of magnetic
anomalies are shifted towards the south relative
to the actual location of the anomaly source in the
ground. The highest resistivity values (15.5 to
21.5 m, horizontally) coincide with large magnetic anomalies attributed to house features.
Therefore, the high resistivity top layer most
probably corresponds to the layer of archaeological material, which thus appears to extend
down to about 1.5 m, in agreement with the test
excavations. Perhaps most interesting of all, the
horizontal position of the funnel-shaped resistivity zone exactly coincides with the position of the
oval feature interpreted as an enclosure. A ditch
filled with debris of various sizes would indeed
be expected to have a higher resistivity than its
surroundings, and if this interpretation is valid,
the resistivity section could provide an estimate
of the size of this feature. Based on the contoured
resistivity model, it appears to have a maximum
width of 3–4 m, and a depth below the surface of
more than 3 m. These dimensions are quite comparable with other known neolithic ditches.
Discussion and integration of results
Comparison of electromagnetic and DC
resistivity results
The EM38 survey has resulted in a mostly featureless conductivity map, and the only significant feature is a high conductivity anomaly
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
221
Figure 6. (Top) Electrical resistivity image obtained along the test profile shown on Figure 3. (Bottom) Same image with proposed
interpretation: 1, main archaeological layer with individual features represented by highest resistivity zones; 2, relatively high
resistivity zone interrupting the horizontal sequence and possibly representing the vertical cross-section of a ditch. Note that the
resistivity scale is logarithmic.
coinciding with a strong magnetic anomaly. The
poor results of the EM38 can be explained by a
lack of resolution of conductivity contrasts. The
whole layer of archaeological material is probably characterized by a relatively low conductivity, and the horizontal contrasts between
different units of this layer may be too small to
be adequately resolved by the EM device. The
presence of a high conductivity anomaly over
archaeological material can have several explanations. First, the material could in fact have a
high conductivity. This is not likely given the
nature of the site and it is also in clear contradiction with the results of the DC resistivity
survey, which show a low conductivity subsurface anomaly at the same location. Another possibility is that the high magnetic susceptibility of
the material is affecting the readings. The effect
of susceptibility is more pronounced on the inphase component of the measurements, but with
susceptibilities as high as the ones measured here
it also likely to affect the readings of the quadrature component used to calculate the conductivity. Finally EM38 readings may be more
affected by changes in moisture caused by the
Copyright # 2004 John Wiley & Sons, Ltd.
presence of archaeological material, than by the
material itself.
A detailed comparison between the DC conductivity image and the corresponding EM38
conductivity profile helps shed some light on
the lack of results and apparent contradictions.
Figure 7 shows for the same profile: (i) the EM
apparent conductivity variations as extracted
from the main conductivity map; (ii) variations
of inverted conductivities at two different depths
extracted from the DC image. To facilitate the
comparison, the profiles are all normalized to an
average of 0 mS m1.
It is first noticed that the EM apparent
conductivity variations, although coherent, are
of very small amplitude. The maximum variation
is only 5.15 mS m1, whereas the DC model resistivity variations reach amplitudes of 59 mS m1
for the shallow profile and 25 mS m1 for the deep
profile. This small amplitude variation is due to
the smoothness inherent in an apparent conductivity value as opposed to a ‘true’ or modelled
conductivity value. The second important observation is that the two DC conductivity profiles
seem to follow opposite behaviours with rises in
Archaeol. Prospect. 11, 213–226 (2004)
222
J. M. Maillol et al.
indicate that at least in the circumstances of this
survey, the EM38 conductivity variations are
more sensitive to the effect of archaeological
material on its sedimentary surroundings than
to the material itself. A similar observation was
made in a completely different environment
for data acquired with a larger system (EM31)
(Maillol et al., 2002).
Comparison of magnetic and electrical
resistivity results
Figure 7. Comparison of conductivity profiles obtained from
the DC electrical image (2 depths 0.36 and 1.19 m, converted
to conductivity) and of the corresponding EM38 profile (top).
Profile locations shown in Figure 3b; see text for explanation.
one profile corresponding to falls in the other
profile. This indicates that the low conductivity
anomalies present at a depth of 1 m or more are
associated and coincide horizontally with high
conductivity anomalies at a very shallow depth.
Finally, the third crucial observation is that the
EM38 profile appears to follow much more closely the shallow DC profile than the deep one. It
is possible to infer from these observations that
the EM38 readings are in fact mostly controlled
by the very near-surface (less than about 50 cm),
where higher conductivities and larger amplitude variations are present. The lower conductivities present at larger depth and associated
with the archaeological materials do not present
sufficient lateral contrasts to greatly affect the
apparent conductivity measured at the surface
by the EM38. The opposite behaviour of the
shallow and deep variations can be explained
by assuming that the top ground layer contains
more moisture where it overlies relatively low
porosity archaeological material than when only
soil or sediments are present. It is plausible that
the presence of archaeological features results in
a poorer drainage and an increase in moisture
near the surface. These observations seem to
Copyright # 2004 John Wiley & Sons, Ltd.
In order to better understand the dual magnetic
and electrical response of the site materials, it
appeared interesting to carry out a detailed comparison between the results of the DC imaging
profile and those of the magnetic survey. A brief
examination of the resistivity image and the
corresponding magnetic profile has indicated
that a very close correlation existed between the
locations of high resistivity anomalies and the
locations of magnetic anomalies. Although it is
generally safe to attribute an anthropogenic origin to magnetic anomalies, resistivity variations
may in many cases have a geological origin. A
way to evaluate the similarity of the anomaly
sources is to compare the resistivity model
resulting from inversion of electrical data with
a magnetic property model obtained from magnetic field data.
Any geophysical model has two fundamental
components: a physical property component and
the spatial distribution of this property. To
further develop the integration, the magnetic
model was built from susceptibility measurements for the physical property component and
from the resistivity model for the spatial distribution of susceptibility. In order to construct an
initial model, the basic assumption was that high
resistivities in the electrical image corresponded
to high susceptibility material in the magnetic
model. In smooth models such as those typically
produced by electrical resistivity imaging it is
always hazardous to try to assign specific boundaries to subsurface features by identifying them
with specific contour lines. In this case, the preliminary model was constructed by arbitrarily
choosing the 43 -m contour line for moderately
elevated susceptibilities and the 73 -m contour
line for high susceptibilities. The presumed ditch
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
was assumed to be represented by the 22.5 -m
contour line. The apparently pointless precise
numbers are a result of the logarithmic scale
used for contouring the resistivity image. The
susceptibility values (k) were selected to be compatible with the actual sample measurements.
The initial model consisted of six bodies with
k ¼ 1 103 u.SI and two bodies with k ¼ 5 103 u.SI.
In very shallow and high-resolution surveys of
this nature it is very important to consider carefully the reference of elevation used for data
acquisition and for computing the models. It is
especially important when comparison and integration between results from different survey
methods are attempted. In the case of the magnetic survey, the sensor was held at an average of
0.15 m above the surface of the ground; allowing
for the finite size of the sensor bottle, it is estimated that the measurements correspond to an
elevation of 0.20 m above the surface of the
ground. Therefore, the depths of the magnetized
bodies in the model have to be decreased by the
same amount to obtain their true distance from
the surface.
In the case of the resistivity image the situation
is less straightforward. In the image reconstruction process, it is assumed that the electrical
sources (electrodes) are located at the surface of
the ground; at a depth of 0 m. This is never quite
true as electrodes have to be driven a certain
distance into the ground. In addition, although
most of the current flows from the tip of the
electrodes, a very poorly known amount also
emanates from their length. These effects make
it very difficult to assign a precise depth reference in an electrical image. A difference of 0.1 to
0.2 m is negligible when a section of several tens
of metres is reconstructed, but when structures at
a depth of about 1 m are imaged it becomes very
significant. Owing to the very good ground contact at the site, the length of electrode buried into
the ground could be kept at constant 0.15 m
value. It is therefore estimated somewhat arbitrarily that a correction of 0.10 m has to be introduced in determining the depths of the magnetic
bodies as inferred from the electrical image.
The two-dimensional magnetic response of the
model was computed with a program based on a
Talwani type algorithm. Using this method it is
Copyright # 2004 John Wiley & Sons, Ltd.
223
possible to calculate the induced and/or remanent magnetic anomaly caused by uniformly
magnetized polygonal bodies. The shape, position and magnetic susceptibility contrast of the
polygons were assigned based on the principles
explained above. It was assumed that the profile
of interest was perpendicular to the strike of most
significant features (see Figure 3b). Only induced
magnetization was considered, and its direction
was defined using declination and inclination
values of þ3.5 and þ62.5 respectively. These
values are approximations calculated from the
International Geomagnetic Reference Field
(IGRF) ninth generation model (revised 2003)
(International Association of Geomagnetism and
Aeronomy, 2003).
The model predictions were then matched
with the observed magnetic profile extracted
from the map. The first pass actually resulted
in a good match as far as the horizontal position
of the magnetized bodies and the overall trend of
the magnetic anomalies were concerned. The
major discrepancies concerned the amplitude
and width of the individual anomalies. The fit
between predicted and observed anomalous
magnetic field values was then improved by
adjusting both body shapes and susceptibility
values in a standard approach. Two additional
bodies on both ends of the section had to be
added to achieve an excellent match. Figure 8
shows the final result in the form of superimposed resistivity image and magnetic model at
the bottom and superimposed predicted and
observed magnetic anomaly profiles at the top.
Caution must be used in comparing two subsurface models obtained with two very different
techniques. The resistivity model is inherently
smooth and gradual changes of resistivity can be
assigned within the constraints imposed by the
finite sizes of the mesh elements. In contrast, the
magnetic model is constrained to discrete polygonal bodies of constant magnetic susceptibility.
In an archaeological context where one expects to
find well delimited structures with strong physical property contrast relative to their sedimentary
background, a polygonal model might in many
cases be a closer representation of the reality. The
positive aspect of this disparity is that if similarities are found between the two models, despite
their fundamentally different origins, they are
Archaeol. Prospect. 11, 213–226 (2004)
224
J. M. Maillol et al.
Figure 8. Comparison of magnetic and electricalresistivity models.Bottom figure shows the test electricalimage (greyscale) and
the outline (in white) of the magnetized bodies used to compute the best-fitting magnetic response.Magnetic susceptibility values
(k) for each body are given in the table.The top figure shows the fit between computed and observed magnetic anomaly values.
See text for discussion.
almost certainly due to the underlying true geophysical characteristics of the subsurface.
Comparison of the two models reveals that a
generally good agreement between the locations
of magnetized bodies and high resistivity zones
is observed, which confirms that their sources are
for the most part identical. A clear positive
correlation between the resistivity and susceptibility is also observed, with the highest resistivities coinciding with the highest susceptibilities.
These two results clearly demonstrate the complementarity of the magnetic and electrical methods on the study site. It appears that resistivity
imaging can be used effectively to determine the
depth distribution of the structures mapped by
magnetometry.
Some discrepancies between the two models
are also observed, and much can be learned from
them in order to better interpret resistivity
Copyright # 2004 John Wiley & Sons, Ltd.
images in general. The bodies in the magnetic
model had to be made systematically thinner and
more angular than in the initial model that
attempted to follow resistivity contour lines.
The features of the resistivity model are therefore
broader and rounder than the modelled magnetic sources; this is exactly what would be
expected because the resistivity inversion tends
to produce a smooth model. This clearly illustrates why caution has to be exerted when
assigning specific resistivity contours to subsurface features. In some locations, in particular
from 13 to 16 m the shape of the final magnetic
body is completely different from the resistivity
contour lines. In this case, the likely explanation
is that although the locations of magnetic and
electrical anomaly sources coincide, the spatial
distribution of the two properties cannot be
identical. The imperfections of both modelling
Archaeol. Prospect. 11, 213–226 (2004)
Archaeological Features at Movila lui Deciov
processes are also of course a determining factor.
Another interesting discrepancy is seen at the
northeast end of the section, around 23 m, where
the magnetic body is displaced from the resistivity anomaly by about 1 m laterally. This is most
likely the result of an edge effect in the resistivity
inversion process. The two small bodies at the
southwest end appear to correspond to a single
relatively high resistivity zone occupying a deeper location. This also could be due to an edge
effect or a lack of resolution of the resistivity
method.
Two fundamental aspects of magnetic and
resistivity modelling have not yet been considered. One is the role played by remanent magnetization in the magnetic model. The model
predicts the response of a magnetized object,
but this magnetization can be induced or remanent. Although it was not measured, it is quite
clear that some of the material present at the site
must possess a very significant remanent magnetization. Considering the age of the site
(less than 7000 years BP) the directions of
induced and remanent magnetizations are not
very different for in-place features. For the fairly
large, most likely in-place features considered in
this modelling exercise, the consequence of
neglecting the effect of remanent magnetization
is therefore only to overestimate the susceptibility, which does not invalidate the predictions of
the model.
The second important aspect of modelling that
has to be considered is the consequence of modelling three-dimensional structures with a twodimensional approximation. The strong anomaly
sources present in the northeast part of the
profiles, when observed on the magnetic map,
do not extend more than approximately 4 m on
either side of the profile. A two-dimensional
approximation is therefore not really adequate,
although the shallowness of the sources (less
than 1 m) should limit the magnitude of errors.
Owing to their different physical basis, magnetic
and electrical models are affected differently,
with the stronger effect expected for the electrical model. Consequently, some discrepancies
between the two models—and of course with
the as yet unknown reality—are to be expected.
However, this should not affect the main results
of the model comparison.
Copyright # 2004 John Wiley & Sons, Ltd.
225
Conclusion
From an archaeological prospection viewpoint
all the objectives have been met. The most useful
results are provided by the magnetic map, with
valuable depth information provided by the
electrical resistivity imaging profile. The extent
of the site has been well defined, which will
greatly help future investigation and conservation efforts. Numerous anomalies have been
found; some are very strong with lateral extensions of several metres and represent priority
targets for detailed excavations. Possibly the
major contribution of the survey is the discovery
of an oval anomaly surrounding the site, which
helps build a very strong case for claiming the
existence of a ditched enclosure at this early stage
of the central European neolithic.
From a methodological perspective, some
valuable information is obtained from the comparative analysis of the magnetic, electromagnetic and electrical imaging surveys. The
disappointing results of the terrain conductivity
surveys, with the additional insight provided
by the test resistivity image, illustrate the limitations of this technique in terms of resolution
of conductivity contrasts. The test of electrical
resistivity imaging is quite successful, and it
demonstrates that this technique can be used
advantageously to obtain depth information on
archaeological sites with no stone structures. The
joint modelling of electrical and magnetic data
gives a unique insight into the origin of both
kinds of anomaly, and in this case it shows that
their sources are mostly identical. This is an
important result because it shows how complementary the two techniques can be; with magnetometry providing a fast mapping tool and
electrical imaging supplying targeted depth
information at key locations.
As always in archaeological geophysics studies, the ultimate feedback can be provided by
subsequent excavations and it is hoped that the
results presented here will provide the impetus
for more extensive archaeological studies of this
particular site. In addition, a significant number
of similar sites exist in western Romania and, as
in all parts of the world, they are coming under
increasing pressure from human development.
This study will thus hopefully contribute to
Archaeol. Prospect. 11, 213–226 (2004)
226
imparting a new momentum to the use of geophysical techniques for the study, management
and conservation of these sites.
Acknowledgements
The project was funded by an International Project Grant from the University of Calgary (I.M.).
It was also made possible through the logistical
support of the Museum of Banat, Timisoara. We
thank Dominique Cossu for her contribution to
data acquisition and for drafting some of the
figures. All the participants are deeply thankful
to the citizens of Dudestii Vechi, and in particular to the Kalcov family for field assistance and
for a wonderful welcome.
References
Bevan B, Roosevelt AC. 2003. Geophysical exploration of Guajarará, a prehistoric earth mound in
Brazil. Geoarchaeology 18(3): 287–331.
Chapman J. 1981. The Vinca Culture of South-east
Europe. International Series No. 117, British Archaeological Reports: Oxford.
Doneus M, Neubauer W. 1999. Two-dimensional
combination of prospection data. Archaeological
Prospection 5: 29–56.
International Association of Geomagnetism and
Aeronomy (IAGA), Division V, Working Group
8. 2003. The 9th-generation International Geomagnetic Reference Field. Geophysical Journal International 155(3): 1051–1056.
Lazarovici G. 1979. Neolithicul Banatului. Biblioteca
Muzei Napocensis: Cluj.
Copyright # 2004 John Wiley & Sons, Ltd.
J. M. Maillol et al.
Loke MH, Barker RD. 1996. Rapid least-square
inversion of apparent resistivity pseudosections
using a quasi-Newton method. Geophysical
Prospecting 44: 499–523.
Maillol JM, Moorman BJ, Glanzman WD, Walter FS,
Williams JL. 2002. Multi-method geophysical investigations at Mahram Bilqis, Yemen—Unveiling the temple of the Queen of Sheba. Proceedings
of the Symposium on the Application of Geophysics
to Environmental and Engineering Problems
(SAGEEP), Las Vegas, NV, February.
Manson JL. 1995. Starcevo pottery and neolithic
development in the central Balkans. In The Emergence of Pottery: Technology and Innovation in
Ancient Societies, Barnett WK, Hoopes JW (eds).
Smithsonian Institution: Washington; 65–77.
Neighbour T, Strachan R, Hobbs BA. 2001. Resistivity imaging of the linear earthworks at the
Mull of Galloway, Dumfries and Galloway.
Archaeological Prospection 8: 157–162.
Noel M, Xue B. 1991. Archaeological investigation
by electrical resistivity tomography: a preliminary study. Geophysical Journal International 107(1):
95–102.
Sambuelli L, Socco LV, Brecciaroli L. 1999. Acquisition and processing of electric, magnetic and GPR
data on a Roman site (Victimulae, Salussola,
Biella). Journal of Applied Geophysics 41(2–3): 189–
204.
Tabbagh A, Bossuet G, Becker H. 1988. A comparison between magnetic and electromagnetic prospection of a neolithic ring ditch in Bavaria.
Archaeometry 30: 132–144.
Whittle A. 1994. The first farmers. In The Oxford
Illustrated History of Prehistoric Europe, Cunliffe B
(ed.). Oxford University Press: Oxford; 136–166.
Zolyomi B. 1953. Die Entwicklungsgeschichte der
Vegetation Ungarns zeit dem letzen Interglazial.
Acta Biologica Academiae Scientarum Hungaricae 3–
4: 346–430.
Archaeol. Prospect. 11, 213–226 (2004)
Документ
Категория
Без категории
Просмотров
3
Размер файла
479 Кб
Теги
features, site, romania, archaeological, electric, early, response, magnetic, western, lui, neolithic, movila, deciov
1/--страниц
Пожаловаться на содержимое документа