Anda di halaman 1dari 21

Archaeological Prospection

Archaeol. Prospect. 6, 85105 (1999)

The Determination of the Depth of


Magnetic Anomaly Sources
GUY DESVIGNES1*, ALAIN TABBAGH2 AND CHRISTOPHE BENECH1
1
UMR 7619 Sisyphe, Universite Paris 6CNRS, case 105, 4 place Jussieu,
75252 Paris cedex 05, France
2UMR 7619 Sisyphe, Universite Paris 6CNRS, Centre de Recherches

Geophysiques, Garchy 58150, France

ABSTRACT Magnetic prospecting is used mainly for horizontal mapping, whereas the determination of the
depth of features receives only limited attention. The inverse problem has an infinity of solutions
and it is not possible to define precisely either the exact shape or the exact depth of a given
feature. On the one hand, the depth of the source of an anomaly is limited and minimal
information about contrast, horizontal location and depth of the centre can be extracted from the
data. On the other hand, comparison between results obtained from different methods or by the
simultaneous interpretation of their results necessitates in the first place an evaluation of the
correspondence between the depth of the sources indicated by the different methods.
Details of a series of four techniques that deliver information about the depth or the range of
depths of magnetic anomaly sources are presented here. The search for the equivalent dipole
(i.e. a spherical body) is of very limited interest owing to the difference that exists between the
actual archaeological features and a sphere. In contrast, Euler deconvolution provides interesting
results when using a structural index approaching 2. The depth of the centre of magnetization
and the total magnetic moment also can be determined. The latter allows an estimation of
the product of the volume and the susceptibility contrast (the possible presence of non-induced
magnetizations being considered). The downward continuation of the equivalent stratum
susceptibility needs more empirical tests to define divergence criteria, but it delivers another
estimation of the susceptibility changes and information about its depth variation. Copyright
*c 1999 John Wiley & Sons, Ltd.

Key words: magnetic prospecting; depth determination; Euler deconvolution; equivalent


stratum.

Introduction advantageous, however, to be able to obtain


more information about the depth, or the range
The determination of the depth of features that of depths, of a given feature: it helps to evaluate
are the origin of magnetic anomalies constitutes a its limit in depth and thus the potential difficulty
difficult problem; it has theoretically an infinite (and cost) of a scheduled excavation, and to
number of solutions and, in archaeological establish if the observed magnetic contrast exists
prospecting, it has received limited attention inside or beneath the superficial cultivated layer.
because the efforts of interpreters mostly tend Beside these usual concerns, the new prospects
towards the restitution of the horizontal shape of that appear in the comparison of the different
the features, which often allows their ident- methods available and in the simultaneous
ification and sometimes their dating. It can be interpretation of different types of measurements,
make it necessary to take into account the differ-
ences that exist between the depths indicated
*Correspondence to: Dr G. Desvignes, UMR 7619 `Sisyphe',
Universite Paris 6, CNRS, Case 105, 4 place Jussieu, 75252 by the various methods of investigation. As
Paris cedex 05, France. magnetic properties' contrasts can be studied by

CCC 10752196/99/02008521$17.50 Received 28 July 1998


Copyright # 1999 John Wiley & Sons, Ltd. Accepted 22 February 1999
86 G. Desvignes, A. Tabbagh and C. Benech

both magnetic and electromagnetic methods (Tite We first present these methods, apply them to a
and Mullins, 1969; Tabbagh, 1974; Desvignes synthetic example and then consider their appli-
and Tabbagh, 1995) it is important to be able to cation on actual data corresponding to total field
separate in the results of the two methods the measurements.
differences originating in the depth of investiga-
tion from those originating in the presence of
viscous or thermoremanent magnetizations. In Determination of the characteristics
other words, as the depth of investigation of the of the equivalent dipole
magnetic method is higher than that of electro-
magnetic methods, one must know, before imple-
Principle
menting a simultaneous interpretation, if the Magnetic anomalies observed in archaeological
magnetic anomaly under consideration may or prospecting may be classified roughly in two
may not originate at a depth out of range of the categories: the ones that correspond to elongate
electromagnetic apparatus used. features (ditches for example), and those that
Since the first use of magnetic prospecting ( for have limited lateral extension in both directions.
investigation of structures other than those associ- For the first category, a line of dipoles must be
ated with magnetic ore, magnetic prospecting used rather than one dipole alone, which is
began in 1915 when Adolf Schmidt constructed convenient for the second category. As in both
his precision vertical-field balance), numerous cases the process that must be followed to fit the
techniques or `rules of thumb' have been devel- theoretical anomaly to the experimental data is
oped for approximating the depth to sources from the same, we shall limit our discussion to the
a measure of the observed anomalies; the major single dipole case. As a survey normally shows a
principle is that there is no lower limit. The most series of elementary anomalies, each must be
well-known method is the half-width, which considered separately; a moving window is thus
evaluates the source centre depth (Telford et al., displaced over the area surveyed, the size of this
1976). Smith (1959) derived techniques for maxi- window needs to be sufficiently wide to cover a
mum depth estimates to the top of the source significant part of the anomaly (7 or 8 m).
determination, based on horizontal derivatives of Four parameters are sought using the least-
magnetic anomalies. Spectral investigations also square method: the magnetic moment, M, and
can be used, where the mean depth is derived the three coordinates, x0 , y0 , and z0 , of the dipole.
(Bhattacharyya, 1978). A maximum depth would The expression of the anomaly at any measured
correspond either to a small body exhibiting a point (x, y), DT(x, y), is linear in M but not linear
high and unlikely contrast or to a spherical body. in x0 , y0 and z0 , thus it has to be linearized using
All these methods are potentially erroneous the Taylor series expansion:
depending upon the degree to which the simplify-
ing assumptions they use concerning the shape of DTx; y DTx; yjx ;y ;z
a a a
the sources are met by the geological or archaeo- @DT @DT @DT 1
logical features. Thus it would be better for a good dx0 dy0 dz0
@x0 @y0 @z0
depth determination to avoid such simplifying
assumptions about the geometry of the sources. An iterative scheme is then applied: (i) a set of a
In the present paper we will consider four priori estimated coordinate values, xa , ya and za is
methods: first chosen; (ii) the moment M is determined; (iii)
the three increments dx0 , dy0 , dz0 are calculated
(i) determination of the equivalent dipole; and added to the a priori values; and the entire
(ii) Euler deconvolution; cycle is reiterated until a convergence parameter,
(iii) location of the centre of magnetization and
determination of the total magnetic moment dx20 dy20 dz20
of a feature; x20 y20 z20
(iv) downward continuation of the equivalent
stratum magnetization. reaches a given threshold.

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 87

Figure 1. Total field anomaly of a prismatic magnetic feature (2  2  0.5 m3) having its top at 0.3 m depth. The
susceptibility contrast is 1  10 3 S.I. and the sensor height 0.3 m.

Application to a synthetic case z 0.3 m). The magnetic inclination is 648 and
the declination 08. The anomaly is presented
This process is applied to a simple synthetic case in Figure 1. The calculated equivalent dipole
previously presented (Desvignes and Tabbagh, has a location of x0 5 m, y0 0.15 m and
1995), in which a magnetic prismatic feature of z0 0.69 m and a magnetic moment of
2  2  0.5 m3 size has a susceptibility contrast of 0.091 A m2, where the centre of the square feature
1  10 3 S.I. with its surroundings. Its top is is at x0 5 m, y0 0 m, z0 0.55 m and the
located at 0.3 m depth and the measurements product of its volume by its magnetic
achieved using a total field magnetometer with the susceptibility contrast and by the Earth's mag-
sensor at 0.3 m above ground (z 0 being the netic field (37.5 A m 1) is 0.075 A m2. The differ-
ground surface, the height of the measurement is ence in the strength of the moment is slightly

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
88 G. Desvignes, A. Tabbagh and C. Benech

greater than that in the depth: if we adopt the geometrically from those considered in mining,
same susceptibility contrast, the equivalent sphere oil or structural prospecting and the experience
would have a 0.83 m radius, and if we adopt acquired in these domains is unfortunately of
the same volume of the sphere as that of the true limited interest, because two-dimensional fea-
body, the susceptibility contrast would reach tures are most often considered (Thurston and
1.21  10 3 S.I. Thus obtaining exact results from Smith, 1997). Despite the fact that most archae-
this method seems limited to a very narrow range ological features are of small lateral extent, they
of shapes of the body source. cannot be described as dipoles (spheres) because
they are usually of limited thickness and are
larger at their top than at their bottom. In the
Euler deconvolution preceding section, for a feature with a width that
is four times its thickness and which has vertical
Principle
limits, the difference between the true depth and
Euler deconvolution was introduced quite the depth of the equivalent dipole is noticeable.
recently (Thompson, 1982; Reid et al., 1990) in Thus, in further calculations we shall have to test
order to identify quickly the location of elemen- for the best value of the structural index that can
tary sources that may correspond to the actual be considered for any type of archaeological
data. It is based on the fact that the magnetic feature. It is also possible to consider N as a
anomaly DT(x, y) verifies the Euler homogeneity fourth unknown and to determine it together
equation: with the coordinates.
@DT @DT @DT
x x0 y y0 z0 h
@x @y @z
Application to a synthetic case
NDTx; y 2
If we consider the same synthetic case, i.e. the
where x0 , y0 and z0 are the coordinates of the response of the prismatic 2  2  0.5 m3 feature,
source, h the height of the measurement and N is we first apply the algorithm where four unknowns
the `structural index' that governs the decrease of are determined. For several sizes of the window
the field with the distance from the source. For a ranging from 8  8 m2 to 3  3 m2, we obtain
dipole (a sphere), N 3, the anomaly of which very stable results: x0 5.0 m, y0 0.01 m,
is proportional to 1/r3; N 2 for a circular z0 1.51 m and N 3.68. The values of the
section cylinder, the anomaly of which is two first coordinates are very close to the true
proportional to 1/r2; and N 1 for a sheet-like ones, 5.0 and 0 m, whereas z0 is far from it (0.55 m)
body, the anomaly of which is proportional to and N surpasses 3, which is physically unaccep-
1/r. table.
The values of the gradients of DT can be When we apply the algorithm where N has a
determined by linear filtering (Gun, 1975; fixed value, we observe that x0 and y0 are again
Tabbagh et al., 1997) over the surveyed area. correctly determined and that z0 only changes
After having chosen N, the three coordinates with N. Figure 2 presents the values of z0
values can be deduced directly from equation (2) obtained for different values of N ranging from
using the values of the anomaly and of its 1.2 to 3.7: the curve increases quite linearly. The
gradients at three different points. In practice, it correct value of z0 corresponds to N 2.0. This
is better to consider more than three points and to value is not physically surprising; it results from
apply the least-squares method to determine the the fact that the width of the feature is four times
three coordinates. This deconvolution is purely greater than its thickness. Other synthetic data
geometrical and does not deliver the magnitude presented in previous papers (Desvignes and
of the source, which will be determined by Tabbagh, 1995; Tabbagh et al., 1997) give similar
another method (see below). The main difficulty values and, even for complex features, the values
encountered in its use is the choice of N. Features of N corresponding to the correct z0 value stay in
sought in archaeological prospecting, differ the interval 1.85 to 2.2. We can thus consider that

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 89

equivalent dipole method. It must be noted,


however, that:

(i) the exact values of (x0 , y0) and the exact


value of the orientation of the Earth's mag-
netic field were introduced;
(ii) the sampling interval is small (0.5 m).

If these conditions are not fulfilled, the theory


and numerical tests show that the correct value of
M is still obtained and its estimation is robust,
but that the value of z0 is sensitive to these
parameters.

Downward continuation of the


equivalent stratum magnetization
Figure 2. Variation of the depth of the centre of the square
magnetic feature calculated by Euler deconvolution, versus Principle
the structural index value N.
For potential field methods, downward continua-
tion constitutes a well-known solution for enhan-
cing the location of the sources. This approach,
the range of N is around these values for archaeo- however, requires several assumptions, the main
logical features (see also the field data below). ones being that there are no sources between the
measurement level and the continuation depth,
and that it is possible to define criteria for the
Location of the centre of divergence of the transformed data indicating
magnetization and of the total that the level of the sources is reached. Down-
magnetic moment ward continuation can be applied to any com-
Principle ponent of the field or its derivatives, or to the
potential using the same continuation function.
The determination of the three coordinates of the In magnetic prospecting, the continuation of
centre of magnetization x0 , y0 and z0 , and of the the potential would be preferred to that of any
total magnetic moment, M, is possible whatever field component, because it corresponds in
the exact shape of the feature. They can be magnetism to the equivalent stratum magnetiza-
deduced from complex integrals of the variations tion (see Appendix 2), which can be transformed
of the three components of the magnetic anomaly easily into susceptibility values of the equivalent
(Andersen and Pedersen, 1979). This method was stratum.
adapted recently for the determination of z0 and If one considers an anomaly DT(x, y) obtained
M, when using only the total field anomaly (see at z 0 m for a given inclination I, the effect of I
Appendix 1). must be first corrected by a reduction to the pole
that gives DTv(x, y). It has been established (see
Appendix 2) that this anomaly can be generated
Application to a synthetic case
by a horizontally varying vertical magnetization
By again considering the anomaly generated by sv(x, y) at z 0 m, the magnitude of which is
the simple prismatic feature we obtained with deduced from the potential U(x, y), defined by
this method z0 0.545 m and M 0.0751 A m2. DTv x; y @U
@z . Therefore, sv(x, y) 2U(x, y).
These determinations are very good (51%) This equivalent stratum constitutes one possible
and much better than those obtained by the solution of the inverse problem. By downward

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
90 G. Desvignes, A. Tabbagh and C. Benech

continuation of the potential, an infinite series of bility value is comparable with the true volu-
other solutions can be proposed, but as the depth metric susceptibility contrast; it is difficult to
increases the maximum of sv increases and the suggest a physical explanation for the value of
lateral variations of the magnetization change. A this, but it has to be noted for further comparison
maximum acceptable value of the depth could be with experimental data.
defined according to two criteria: (i) by fixing a From this first example we conclude that the
maximum acceptable value for sv and/or (ii) by separation of the sources into a group of point-
checking the lateral changes of sv , for instance like maxima can be considered as a criterion
when negative peaks appear in alternation with a indicating that the top of the feature is reached,
series of high positive peaks. whereas the regularity of the amplitude increase
The calculation is achieved in the spectral makes it difficult to define a threshold.
domain, where u and v are the spatial frequencies It must be noticed that these synthetic data are
corresponding to x and y directions respectively. noise free; in case of considerable measurement
With F(u, v) being the Fourier transform of errors the downward continuation may be
DT(x, y), the calculation of the potential and its unstable and the interpretation difficult.
downward continuation over a vertical distance h
are achieved simultaneously using (Tabbagh et al.,
1997): Application to field data
p p
1 u2 v2 2ph u2 v2 Two different examples are considered for testing
Fu; v p e
2p iu cos I u2 v2 sin I2 the preceding processes on field data. The first
3 one corresponds to a pottery workshop, which
produced amphorae from the Hellenistic period
when the declination is null. The magnetization on the Datca peninsula (Resadiye, Turkey)
is thus obtained after an inverse Fourier trans- (Empereur et al., in press): it presents both kilns
form. By dividing it by the amplitude of the and waste heaps; the magnetic contrast can be
Earth's magnetic field, an equivalent stratum high and the depth of the features may exceed
susceptibility can be restored. 1 m. The second is a Merovingian cemetery,
where burials correspond to pits filled with
topsoil. It is located at Garchy just beside the
Application to a synthetic case
`Centre de Recherches Geophysiques'. Here, the
If we apply the process to the synthetic case size of the features and the magnetic contrast is
( prismatic 2  2  0.5 m3 feature, where the more limited, the bottom of the features never
measurement step is 0.5 m) we observe first exceeding 1 m.
(Figure 3a) that the susceptibility of the equival- Two different parcels of land at the first site are
ent stratum has smooth lateral variations at the considered: they were surveyed in 1991 and 1992
measurement level (z 0.3 m). It is symmetric with a Geometrics 816 proton magnetometer
and its amplitude reaches 0.55  10 3 S.I. A with 1 nT sensitivity. The sensor was held 0.7 m
downward continuation increases the amplitude above ground surface and the measurement
and improves the correspondence between the mesh was 2  2 m2. The first selected zone,
true square shape of the body and the lateral zone I, shows (Figure 4) a burnt brick feature
changes of the susceptibility (Figure 3b and c). At ( partly explored by excavation but most probably
0.3 m below ground surface, a depth which a kiln) in the eastern part and a group of features
corresponds to the level of the top of the feature, (kiln or waste heaps) in the western part. The
we obtain a clear square of 1  10 3 S.I. ampli- second zone, zone Q, shows (Figure 5) a group of
tude. At lower levels (Figure 3d), inside the body, five isolated positive anomalies, the northern-
the susceptibility image exhibits different poles most of which was recognized as a kiln, and two
(each being inside the prism but corresponding to other features, an elongated one crossing the
a corner); such an aspect can be interpreted as a entire map and a concentrated one just at its
divergence effect. At 0.3 m depth, the suscepti- northernmost limit.

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 91

Figure 3. (a) Equivalent stratum susceptibility at measurement level (0.3 m above the ground surface) for the prismatic
magnetic feature.

The Merovingian cemetery has been studied magnetic susceptibility (more than 0.7  10 3
over a long time (Hesse, 1966), as its location S.I.), the burial pits dug in the Oxfordian lime-
facilitates the testing of new prospecting methods stone exhibit an equivalent or slightly greater
or new instruments. The data considered susceptibility.
(Figure 6) were recorded in 1979 (Banteaux et al.,
1979) using two synchronized Geometrics 816
Amphorae workshop at Resadiye (Turkey)
proton magnetometers (at 1 nT sensitivity) in
differential mode. The mobile sensor was held at Zone I
0.3 m height above ground and the measurement In this zone the search for the equivalent dipole
mesh was 1  1 m2. On this site the superficial parameters results in x0 9.51 m, y0 8.74 m,
cultivated layer is thin (0.2 m) but of significant z0 4.3 m and M 93 A m 2 for the eastern

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
92 G. Desvignes, A. Tabbagh and C. Benech

Figure 3. (b) Downward continuation of the equivalent stratum susceptibility at 0.1 m below the ground surface.

kiln, using a 16  16 m2 window. The depth of z0 and M, are definitely overestimated. Thus this
the dipole was clearly too deep, excavations process has no further purpose in the present
having shown that the top of the burnt clay was case.
at 0.8 m depth. In the western part similar results In contrast, Euler deconvolution showed very
were obtained, x0 12.7 m, y0 45 m and coherent and interesting results. For the eastern
z0 9.3 m with a very high moment of kiln we obtained for an 8  8 m2 window,
866 A m2; this huge value is the consequence of x0 11 m, y0 7.6 m, z0 1.73 m and a struc-
the very high z0 value. Such incorrect deter- tural index N 2.15 (medians of 11 independent
mination drastically limits the interest in the determinations); for a 12  12 m2 window, we had
equivalent dipole determination: whereas x0 and x0 12 m, y0 8.7 m, z0 1.2 m and N 1.90.
y0 are easily estimated by simply looking at the Both these determinations are in agreement with
map, the results obtained for the two unknowns, the location of the top of the first layer of burnt

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 93

Figure 3. (c) Downward continuation of the equivalent stratum susceptibility at 0.3 m (the top of the feature) below the
ground surface.

clay at 0.8 m and of the burnt bricks of the kiln at For the eastern kiln the determination of the
1.60 m. The structural index values confirm those depth of the centre of magnetization using
obtained with synthetic data. For the western x0 8 m, y0 12 m and a 16  16 m2 window
group of features the automatic determination of is z0 2.15 m; the total magnetic moment equals
N leads to very low structural indices: the median 108 A m2. By considering three types of magne-
value is N 0.56, this corresponds to z0 0.1 m, tization, thermoremanent, viscous and induced,
which would imply the presence of a magnetic and by estimating the volume to 20 m3, we obtain
medium just below the ground surface, which is a susceptibility contrast between 144  10 3 S.I.
not in accordance with the subsurface suscepti- (if the magnetization is an induced one only) and
bility map obtained form electromagnetic 48  10 3 S.I. (if the thermoremanent and the
measurements (EM15). If we fix N 2, z0 remains viscous magnetizations are equivalent to the
around 1.3 m. induced one). For the anomaly at the centre of

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
94 G. Desvignes, A. Tabbagh and C. Benech

Figure 3. (d) Downward continuation of the equivalent stratum susceptibility at 0.55 m (the centre of the feature) below the
ground surface.

the western group, starting with x0 13 m, kiln. It must be noticed also that the lateral
y0 43 m we obtain for a 10  10 m2 window: variation of susceptibility clearly differs from the
z0 1.15 m and M 30 A m 2, in good agree- total field anomaly map (Figure 4). At 0.8 m
ment with the Euler deconvolution results. Using below ground level (Figure 7b) this anomaly is
a 10 m3 volume we obtain a susceptibility con- dissociated into a group of point maxima. The
trast between 80 and 27  10 3 S.I. depending on increase of the susceptibility value between 0.2
the nature of the magnetization. and 0.8 m depth is more marked for the eastern
The downward continuation of the equivalent kiln, from 70 to 100  10 3 S.I., than for the
stratum at 0.2 m below ground level (Figure 7a) western group, from 100 to 130  10 3 S.I. This
gives higher values of the susceptibility for the can be an indication for a difference in the depth
western group of features than for the eastern of the magnetic anomaly sources. The magnitude

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 95

Figure 4. Total field magnetic anomaly map on the pottery workshop of Resadiye (Turkey), zone I (measurements by A.
Hesse). The isolated anomaly at the southeastern corner probably corresponds to a kiln, the top of which is at 0.8 m depth
below the ground surface.

of the susceptibility (which in this calculation x0 7.5 m, y0 7 m, z0 1.5 m and N 2.96; it


includes all types of magnetization) is consistent differs from the four other features, which give z0
with the strength of the total magnetic moment. greater than 1.3 m and N 2.15. The elongated
feature and the concentrated one correspond to
Zone Q low structural indices if N is determined auto-
If we consider directly the Euler deconvolution, matically: they give, respectively, z0 0.4 m and
the kiln among the five features that outcrops N 1 and z0 1.2 m and N 1.13. These
just below the top soil (0.25 m) exhibits a values would correspond to superficial features.
high structural index and too great a depth: If we fix N 2, z0 becomes greater than 1.3 m

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
96 G. Desvignes, A. Tabbagh and C. Benech

Figure 5. Total field anomaly map on the pottery workshop of Resadiye (Turkey), zone Q (measurement by A. Hesse).

for all. It has to be noted that although the index to be considered in the Euler deconvolution
automatic determination of N takes into account would have to be between 1 and 2 rather than 2.
all the information, fixing the value of N The downward continuation at 0.2 m below
may advantage the lower part of the magnetic the ground surface (Figure 8a) clearly shows the
body; in other words the choice of N may act as a difference between the kiln and the four other
filter. features in the western corner of the survey. At
The determination of the depth of the centre 0.8 m (Figure 8b) the anomaly of the kiln stays
of magnetization and of the total magnetic coherent, in agreement with the fact that the
moment produces the following results: for the depth of its centre is below 0.8 m. The magnitude
body centred at x0 27 m and y0 29 m, of the equivalent stratum susceptibility of the
z0 0.85 m and M 10 A m2, for the body elongated and concentrated waste heaps
centred at x0 7 m and y0 7 m, z0 0.75 m increases slightly from 0.2 m to 0.8 m continua-
and M 9.7 A m2. Both depth results are lower tion depths ( from 40 to 60  10 3 S.I.). This
than 1.3 m, which suggests that the structural suggests that it would be interesting to consider

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 97

Figure 6. Total field anomaly map on the Merovingian cemetery at Garchy (France) (measurement by A. Hesse).

the rate of increase with continuation depth as a 0.6, to which correspond small values of z0 , from 0
possible criterion for source location. Again these to 0.3 m. This suggests that part of the magnetic
susceptibility values are in agreement with the contrast originates in the topsoil layer (this corres-
total moment strength and a likely volume. ponds to the observation), owing to the fact that
the magnetic material is partly scattered in this
layer by agricultural work and erosion processes.
Merovingian cemetery at Garchy (France)
By fixing the structural index at N 2, the depth
A group of a dozen burial pits can be observed on of the magnetic sources is increased to an average
the total field anomaly map (Figure 6). If we apply value of 0.55 m in agreement with the observa-
Euler deconvolution, the automatic determination tions realized during excavations (Hesse, 1966).
of N results in very low values, between 0.3 and The calculation of z0 and M was performed for

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
98 G. Desvignes, A. Tabbagh and C. Benech

Figure 7. Resadiye, zone I, downward continuation of the equivalent stratum susceptibility: (a) at 0.2 m below the ground
surface.

three anomalies centred at (x 40 m, y 67 m), Downward continuation of the equivalent


(x 34 m, y 68 m) and (x 20 m, y 78 m) stratum susceptibility suggests for a 0.2 m con-
respectively. For each we obtained z0 0.75 m; tinuation depth below the ground surface
this value is larger than that obtained from Euler (Figure 9a), that the pits are surrounded by a
deconvolution, thus the features are more com- zone of greater susceptibility, which was not
pact than expected. The total magnetic moments evident on the total field anomaly map. For
are, respectively, 2.1 A m2, 0.98 A m2 and 0.49 A greater continuation depths, 0.8 m in Figure 9b,
m2. For a volume of several cubic metres they the maximum susceptibility value reaches twice
correspond to several 10 3 S.I. susceptibility the known susceptibility contrast and a lot of
values. peaks appear in the anomalies: this suggests that

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 99

Figure 7. Resadiye, zone I, downward continuation of the equivalent stratum susceptibility: (b) at 0.8 m below the ground
surface.

the level of the sources is overstepped. In that for a possible viscous magnetization, and thus to
case, however, it is difficult to define either by the establish that the archaeological features have (or
magnitude of the susceptibility or by its spatial have not) been disturbed by agricultural work.
variation a precise depth for the magnetic
anomaly sources. Nevertheless it will be interest- Conclusion
ing to compare, in magnitude and in lateral
changes, the susceptibility variations obtained Four techniques were considered here for the
for small continuation depths, here 0.2 m, to a determination of the depth of magnetic anomaly
susceptibility map supplied by an electro- sources. The first one, the determination of the
magnetic apparatus; in order to provide evidence depth of the equivalent dipole is definitely of no

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
100 G. Desvignes, A. Tabbagh and C. Benech

Figure 8. Resadiye, zone Q, downward continuation of the equivalent stratum susceptibility: (a) at 0.2 m below the ground
surface.

interest; as the actual lateral variations of the provide a control on the Euler deconvolution
contrast do not correspond to this model, the results and thus can be used to define N. The
depth is largely overestimated and the corre- strength of the total moment provides a value for
sponding moment value totally unlikely. In the magnetic susceptibility contrast when the
contrast, Euler deconvolution is a light and total volume of a feature can be estimated. The
efficient process, it will be easily implemented strength of the total moment also may indicate
for in-field processing and will allow deter- the presence of different types of magnetization.
mination of an approximated depth. By fixing In addition to information about the depth of
the structural index it is possible to reduce (or to the sources of anomalies, calculation of the
increase) the influence of the superficial part of susceptibility of the equivalent stratum and its
the magnetic sources. We have established by downward continuation can provide an order of
both the synthetic model and actual case studies magnitude estimation of the susceptibility values
that N 2 is a fair estimation of this index for and reveal its lateral variations. For both objec-
most archaeological features. The determination tives it needs more studies on real cases in order
of the depth of the centre of magnetization can to define its full abilities. Moreover, as it looks to

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 101

Figure 8. Resadiye, zone Q, downward continuation of the equivalent stratum susceptibility: (b) at 0.8 m below the ground
surface.

be a very interesting tool for the comparison to some integrals of the anomalies (Helbig,
between magnetic and electromagnetic measure- 1963)
ments, such studies will be performed soon. Z Z
2 1 1
Mx xDZx; y dx dy
m0 1 1
Appendix 1 Z 1 Z 1
2
My yDZx; y dx dy
Some parameters of the disturbing magnetic m0 1 1
body can be calculated from measured magnetic Z 1 Z 1
anomalies, without any assumption about its 2
Mz xDXx; y dx dy
shape. m0 1 1
Z 1 Z 1
(i) The three components of the total magnetic 2
yDYx; y dx dy
moment of the disturbing body are related m0 1 1

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
102 G. Desvignes, A. Tabbagh and C. Benech

Figure 9. Merovingian cemetery of Garchy, downward continuation of the equivalent stratum susceptibility: (a) at 0.2 m
below the ground surface.

Here the International System of units (S.I.) is and Pedersen (1979) for the two-dimensional
used; m0 4p  10 7 is the vacuum permeability case to three dimensions. We obtain:
and DX, DY, DZ are the three components of the Z 1  Z 1 
disturbing magnetic induction. 2
x DXx; y dy 2Mx dx
1 1
(ii) The three components of the centre of m0 x0 Mz z0 M x
the magnetization also can be determined Z 1  Z 1 
2
uniquely from measurements, using higher x DZx; y dy 2Mz dx
1 1
integrals of anomalies. It is quite easy to
generalize the results obtained by Andersen m0 x0 Mx z0 Mz

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 103

Figure 9. Merovingian cemetery of Garchy, downward continuation of the equivalent stratum susceptibility (b) at 0.8 m
below the ground surface.

Z 1  Z 1 
2 outside of this square is not negligible; it can
y DYx; y dx 2My dy
1 1 be approached analytically by assuming that
m0 x0 Mz z0 My the lateral variations of the anomaly corre-
spond to those of a single dipole with a
magnetic moment (Mx , My , Mz) at the centre
In the practical case data exist only within a (x0 , y0 , z0) (Grant and West, 1965).
limited area, although all the preceding
integrals are to be carried over the entire Moreover, rather than the three components
horizontal plane. Suppose, for simplicity, DX, DY, DZ the measurements generally are
that this area is a square with sides 2a. For all limited to the total disturbing magnetic induction
these integrals the part corresponding to the DT, that is the component in the direction of the

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
104 G. Desvignes, A. Tabbagh and C. Benech

main terrestrial magnetic field (with inclination I and


and declination D).
4a2 z0 a2 2a Ra
C3 2
8 artg log
DT cos D cos IDX sin D cos IDY sin IDZ a2 z0 R z 0 R z0 Ra

Thus only the component M of the total magnetic So z0 can be determined from the variations of DT
moment in the same direction can be determined and the preceding knowledge of M. Again the two
uniquely from these measurements. We obtain: correction terms C2 and C3 involve z0 ; but in this
Z a Z a case z0 needs to be known with good precision.
4p Therefore, a somewhat complex, but stable,
M xDTx; y dx dy
m0 C1 cos D sin 2I a a iterative process has to be carried out. Also note
that these formulae assume that the coordinates x0
or and y0 are zero; thus the centre of each anomaly
Z Z must be determined carefully using reduction to
a a
4p the pole or Euler deconvolution.
M yDTx; y dx dy
m0 C1 sin D sin 2I a a
1
Appendix 2
with
The equivalent stratum was first used in mag-
a2 4a2 z0 netic prospecting interpretation by Peters (1949)
C1 4 artg 2
z0 R a z20 R in order to restore the depth of a basement, i.e.
the boundary between low susceptibility sedi-
and mentary rocks and the more magnetic basement
q rocks. It can be defined by following the same
R 2a2 z20 process as in gravimetry (Grant and West, 1965).
If we consider on the horizontal plane at z 0
These two formulae enable the evaluation of M a distribution of vertical magnetization sv(x, y), it
from numerical integration of the variations of generates, at a point located above the z 0
DT; indeed the correction term C1 involves z0 ; but plane (z is positive downward), a potential
this value need not be determined with great Z Z
1 1 1 @
accuracy here (one can use the half-width of the Ux; y; z sv x0 ; y0
4p 1 1 @z
anomaly)
1 0 0
 q dx dy
4p
z0 x x0 2 y y0 2 z2
m0Mcos2 I cos2 DC2 sin2 IC3
Z a Z a By locating the origin at the point (x, y) and
2
 x DTx; y dx dy adopting polar coordinates, r and y, we have:
a a
Z
1 1 sv r
or Ux; y; z z r dr
2 0 r2 z2 3=2
4p
z0 The integration domain can be divided in two
m0 Mcos2 I sin2 DC2 sin2 IC3 parts, [0, a] and [a, 1 ]. For a given value of a the
Z a Z a 2
2 integration over the second part vanishes when z
 y DTx; y dx dy tends towards 0, while the integration over the
a a
first part equals
with  
z
1 p
4a4 a2 4a Ra a2 z2
C2 2
8 artg log
2
z0 a z0 R z0 R z0 Ra and tends towards 1.

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)
Depth Determination of Magnetic Anomaly Sources 105

Then we have: Gun, P. J. (1975). Linear transformations of gravity


and magnetic fields. Geophysical Prospecting 23:
sv x; y 300312.
Ux; y : Helbig, K. (1963). Some integrals of magnetic
2
anomalies and their relation to the parameters of
Any distribution of the potential can be con- the disturbing body. Zeitschrift fur Geophysik 29:
8396.
sidered, at the measurement altitude z, as the
Hesse, A. (1966). Prospections geophysiques a faible
effect of the corresponding vertical magnetiza- profondeur Applications a l'archeologie. Paris:
tion distribution at the same level. Dunod.
Peters, L. J. (1949). The direct approach to magnetic
interpretation and its practical application. Geo-
physics 14(3): 290320.
References Reid, A. B., Allsop, J. M., Granser, H., Millett, A. J. and
Somerton, I. W. (1990). Magnetic interpretation
Andersen, F. H. and Pedersen, L. B. (1979). Some in three dimensions using Euler deconvolution.
relations between potential fields and the strength Geophysics 55(1): 8091.
and center of their sources. Geophysical Prospecting Smith, R. A. (1959). Some depth formulae for local
27(4): 761774. magnetic and gravity anomalies. Geophysical
Banteaux, L., Hesse, A. and Jolivet, A. (1979). Prospecting 7: 5563.
A new apparatus of differential recording of the Tabbagh, A. (1974). Methodes electromagnetiques
magnetic field for archaeological prospecting. Nine- de prospection applicables aux problemes archeo-
teenth International Symposium on Archaeometry and logiques. Archaeo-physika 5: 349437.
Archaeological Prospection, London, 2831 March Tabbagh, A., Desvignes, G. and Dabas, M. (1997).
1979, Abstracts. Processing of Z gradiometer magnetic data using
Bhattacharyya, B. K. (1978). Computer modeling in linear transforms and analytic signal. Archaeological
gravity and magnetic interpretation. Geophysics 43: Prospection 4: 113.
912929. Telford, W. M., Geldart, L. P., Sheriff, R. E. and Keys,
Desvignes, G. and Tabbagh, A. (1995). Simultaneous D. A. (1976). Applied Geophysics. Cambridge: Cam-
interpretation of magnetic and electromagnetic pro- bridge University Press.
specting for characterisation of magnetic features. Thompson, D. T. (1982). EULDPH: a new technique for
Archaeological Prospection 2: 129139. making computer assisted estimates from magnetic
Empereur, J. Y., Hesse, A. and Tuna, N. (in press). Les data. Geophysics 41(1): 131137.
ateliers d'amphores de Datca, peninsule de Cnide. Thurston, J. B. and Smith, R. S. (1997). Automatic
Actes du colloque `Production et commerce des amphores conversion of magnetic data to depth, dip, and
anciennes en mer noire', Institut Francais d'Etudes susceptibility contrast using the SPI2 method.
Anatoliennes, Istanbul, 2528 May 1994. Geophysics 62: 807813.
Grant, F. S. and West, G. F. (1965). Interpretation Theory Tite, M. S. and Mullins, C. E. (1969). Electromagnetic
in Applied Geophysics. New York: MacGraw Hill, prospecting: a preliminary investigation. Prospezione
214216. archeologiche 4: 95102.

Copyright # 1999 John Wiley & Sons, Ltd. Archaeol. Prospect., Vol. 6, 85105 (1999)

Anda mungkin juga menyukai