Sie sind auf Seite 1von 19

Resistivity anomaly imaging by probability

tomography
1
Paolo Mauriello
2
and Domenico Patella
3
Abstract
Probability tomography is a new concept reecting the inherently uncertain nature of
any geophysical interpretation. The rationale of the new procedure is based on the fact
that a measurable anomalous eld, representing the response of a buried feature to a
physical stimulation, can be approximated by a set of partial anomaly source
contributions. These may be given a multiplicity of congurations to generate
cumulative responses, which are all compatible with the observed data within the
accuracy of measurement. The purpose of the new imaging procedure is the design of
an occurrence probability space of elementary anomaly sources, located anywhere
inside an explored underground volume. In geoelectrics, the decomposition is made
within a regular resistivity lattice, using the Frechet derivatives of the electric potential
weighted by resistivity difference coefcients. The typical tomography is a diffuse
image of the resistivity difference probability pattern, that is quite different from the
usual modelled geometry derived from standard inversion.
Introduction
In a recent paper, Mauriello, Monna and Patella (1998) extended to geoelectric
measurements the principles of probability tomography, previously proposed by
Patella (1997) for the self-potential (SP) method. This extension involved the charge
occurrence probability concept, and the geoelectric tomography in effect consisted of
an image reconstruction of the most probable location of electrical charges induced by
the primary source over buried resistivity discontinuities. Numerous tests performed
on synthetic examples and near-surface eld data have demonstrated the efciency of
the tomography algorithm and the physical reliability of charge occurrence probability
imaging in geoelectrics.
In this paper, we deal again with geoelectric probability tomography, considering
now the resistivity parameter. With this new formulation we propose a pattern
recognition method to investigate the most probable distribution of underground
1999 European Association of Geoscientists & Engineers 411
Geophysical Prospecting, 1999, 47, 411429
1
Received March 1998, revision accepted December 1998.
2
Institute of Technologies Applied to Cultural Heritage, National Research Council, Rome, Italy.
3
Department of Physical Sciences, University Federico II, Naples, Italy. E-mail: domenico.patella@
na.infn.it
resistivity anomalies. The present contribution thus complements the previous
approach by Mauriello et al. (1998). For simplicity, we limit the analysis to the 2D
geoelectric pseudosection technique.
Resistivity probability tomography
The resistivity anomaly occurrence probability function
We consider a set of apparent resistivity data r
a
(, n) ( = 1, 2 . . , L; n = 1, 2 . . , N),
measured by any electrode device (polepole, poledipole, dipoledipole, etc.) along a
straight-line prole located on the free surface of an inhomogeneous, isotropic
resistivity structure. Using the standard rules for pseudosection tracing, we assign the
r
a
(, n) values to the nodes of a vertical 2D grid across the prole. At each node is the
position along the x-axis, dening the prole, and n is the pseudodepth along the
vertical z-axis, positive downwards (see Fig. 1).
We assume the whole portion of ground contributing to all of the measured r
a
(, n)
data to be composed of Q elementary cells with a sufciently small volume DV, each
identied by a true resistivity r
q
(q = 1, 2 . . , Q). Expanding r
a
(, n) in a Taylor series,
we obtain
Dr
a
(; n) = r
a
(; n) r
a; 0
(; n)
=
X
Q
q =1
r
a; 0
(; n)
r
q
Dr
q

X
(higher order derivatives); (1)
where Dr
a
(, n) represents the difference of the measured apparent resistivity r
a
(, n)
from an apparent resistivity r
a,0
(, n), computed at the same (, n) node using a
reference model indicated by mod
0
. Dr
q
is the difference between the actual resistivity
in the qth cell and the resistivity in the same cell obtained using mod
0
.
Adopting the same scheme that was used to develop the electric charge occurrence
412 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 1. The dipoledipole pseudosection proling method.
probability tomography (Patella 1997; Mauriello et al. 1998), we now introduce a
power function W, related to the apparent anomaly data function Dr
a
(, n) as follows:
W =
X
L
=1
X
N
n=1
[Dr
a
(; n)]
2
: (2)
Using the expansion (1), we extract from W the main contribution W
/
, referred to as
the rst-order derivative, given by
W
/
=
X
L
=1
X
N
n=1
Dr
a
(; n)
X
Q
q =1
r
a; 0
(; n)
r
q
Dr
q
; (3)
which, rearranging the sums and putting
I
q
(; n) = r
a; 0
(; n)= r
q
; (4)
becomes
W
/
=
X
Q
q =1
Dr
q
X
L
=1
X
N
n=1
Dr
a
(; n)I
q
(; n): (5)
Applying Schwarzs inequality to a generic qth double sum in (5), we obtain
X
L
=1
X
N
n=1
Dr
a
(; n)I
q
(; n)
" #
2

X
L
=1
X
N
n=1
[Dr
a
(; n)]
2

X
L
=1
X
N
n=1
[I
q
(; n)]
2
: (6)
Finally, dividing the square root of the left-hand term of (6) by the square root of the
corresponding right-hand term, we introduce the function
h
q
= C
q
X
L
=1
X
N
n=1
Dr
a
(; n)I
q
(; n); (7)
where
C
q
=
X
L
=1
X
N
n=1
[Dr
a
(; n)]
2

X
L
=1
X
N
n=1
[I
q
(; n)]
2
" #
1=2
: (8)
It is easy to prove that the function h
q
satises the condition
1 h
q
1: (9)
Each h
q
value is heuristically interpreted as the probability that a resistivity anomaly
located in the qth cell, deviating frommod
0
, is responsible for the whole set of measured
apparent resistivities, within the rst-order expansion. Positive and negative values of
h
q
result from increments and decrements of resistivity in the qth cell with respect to
mod
0
, respectively. We refer to h
q
as the resistivity anomaly occurrence probability
(RAOP) function.
The RAOP function can be computed easily provided that the Jacobian matrix term
I
q
(, n) is assigned an explicit form, completely independent of grid resistivities. The
Resistivity anomaly imaging 413
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
only possibility of obtaining such a type of I
q
(, n) function is to assume mod
0
to
be a homogeneous, isotropic half-space. In this case I
q
(, n) can be determined
analytically, using the Frechet derivative of the electrical potential f, introduced
by Loke and Barker (1995) following the theory developed by Park and Van
(1991).
The resistivity probability tomography imaging
Using the RAOP function, the probability tomography imaging consists of a scanning
procedure operated by the test function I
q
(, n) on the eld data set Dr
a
(, n), along
the vertical section through the measuring prole.
Ideally, the procedure consists of moving a resistivity anomaly element of unit
strength through the cells of the lattice in order to recognize where resistivity deviations
frommod
0
are located in a probabilistic sense. In each position of the scanning element,
i.e. for each value of q, we calculate, by means of (7), the relative h
q
value, which is
attributed to the centre of the cell where the scanning element is temporarily located. A
resistivity probability tomography image can nally be drawn by contouring or colour
ranging the whole set of computed h
q
values.
Of course, the most general approach is 3D, but the simplied 2D data acquisition is
the most frequently used procedure. Obviously, the 2D approach lacks off-prole
resolution and the signicance of the RAOP function decreases rapidly outside the
prole section.
The scanning function
For simplicity, we limit our analysis to the 2D pseudosection proling technique,
performed using a dipoledipole electrode conguration (see Fig. 1). As stated above,
the straight-line prole is assumed to be the x-axis of a plane rectangular coordinate
system, where the vertical z-axis is positive downwards. The origin can be placed
anywhere along the prole, e.g. coincident with the rst current electrode position (see
Fig. 1). Moreover, the dipole amplitude a is equal, in current practice, to the advancing
step along the prole.
The dipoledipole apparent resistivity r
DD
a
(, n) is dened as
r
DD
a
(; n) =
K
DD
I

Df(; n); (10)


where (see Fig. 1) I

is the intensity of the primary current injected into the ground


through the current electrodes A and B in the th and ( 1)th positions, respectively,
f is the electrical potential function and K
DD
is the dipoledipole geometrical factor,
computed by (Hanneson 1990) as
K
DD
= pan(n 1)(n 2): (11)
414 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
The potential difference Df(, n) is expanded as
Df(; n) = f(; n ) f( 1; n ) f(; n 1) f( 1; n 1):
(12)
Using the denitions (4), (10) and (12), the dipolar scanning function I
DD
q
(, n) can
be written in extended form as
I
DD
q
(; n) =
K
DD
I

r
q
[f(; n ) f( 1; n )
f(; n 1) f( 1; n 1)]: (13)
The four Frechet derivatives are taken from Loke and Barker (1995) as follows:
f(; n )
r
q
=
I

DV
4p
2

[x
q
a][x
q
(n ) a] y
2
q
z
2
q
|[x
q
a]
2
y
2
q
z
2
q

3=2
|[x
q
(n )a]
2
y
2
q
z
2
q

3=2
;
(14)
f( 1; n )
r
q
=
I

DV
4p
2

[x
q
( 1) a][x
q
(n )a] y
2
q
z
2
q
|[x
q
( 1) a]
2
y
2
q
z
2
q

3=2
|[x
q
(n )a]
2
y
2
q
z
2
q

3=2
;
(15)
f(; n 1)
r
q
=
I

DV
4p
2

[x
q
a][x
q
(n 1)a] y
2
q
z
2
q
|[x
q
a]
2
y
2
q
z
2
q

3=2
|[x
q
(n 1)a]
2
y
2
q
z
2
q

3=2
;
(16)
f(1; n1)
r
q
=
I

DV
4p
2

[x
q
( 1)a][x
q
(n 1)a] y
2
q
z
2
q
|[x
q
(1)a]
2
y
2
q
z
2
q

3=2
|[x
q
(n1)a]
2
y
2
q
z
2
q

3=2
;
(17)
where x
q
, y
q
, z
q
are the coordinates of the barycentre of the cubic element with volume
DVand resistivity r
q
.
Resistivity anomaly imaging 415
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Discussion
The reference model
Even though the scanning function I
q
(, n) is independent of the resistivity of the
reference homogeneous half-space, we still need to assign a value to this resistivity in
order to compute the data function Dr
a
(, n). Avery reliable uniform resistivity can be
either the background true resistivity, if known, or simply the average apparent
resistivity.
In this context, the resistivity anomaly concept has a purely relative meaning, since it,
in effect, represents a departure from the selected reference resistivity. Therefore, the
RAOP tomography is a mod
0
-dependent method. None the less, we expect to nd, in
principle, a general consistency among all of the images deriving from different,
plausible choices of mod
0
. These images would actually represent different, but
equivalent, ways of looking at the same geoelectric structure. This aspect will be
claried below, during the presentation of a synthetic example.
The rst-order contribution
The use of only the rst-order contribution W
/
to the total power W, a function of the
apparent resistivity anomaly function Dr
a
(, n), deserves further explanation for the
versatility and potentiality of this new method to be fully appreciated.
Basically, the theoretical foundations of the probability tomography (Patella 1997)
do not allow, in resistivity anomaly imaging, more than one derivative order of the
r
a,0
(, n) function to be used simultaneously. If the Taylor series (1) of r
a
(, n) had
been truncated at a derivative order higher than the rst, it would have been impossible
for the unknown resistivity departure coefcients Dr
q
to disappear from the expression
of the h
q
function. The automatic cancellation of the unknown Dr
q
multipliers is the
most relevant characteristic of resistivity probability tomography imaging and is the
reason why this new imaging method requires no rm guess about the resistivity
departure coefcients Dr
q
to be made in advance.
Of course, we are aware that any inuence of one resistivity deviation over another is
completely neglected in a rst-order tomography and that a rather smooth image is
likely to be the nal result. In further work, we shall develop, separately, the power term
W
//
related to the second-order derivatives. We will investigate the relevant information
which W
//
adds to the W
/
tomography and how, in a W
//
tomography, structural
information can be distinguished from high wavenumber noise.
Probability tomography versus inversion
From the conceptual point of view, a big difference exists between any of the current
resistivity inversion methods and our new probability tomography imaging.
Resistivity inversion is basically a deterministic approach, because ideally its strategy
is to retrieve the true resistivity structure from a measured apparent resistivity data set.
416 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
However, in practice, a large number of equivalent solutions can be derived because of
data inaccuracy and noise contamination. A priori information is essential for the
choice of the base model (Oldenburg, McGillivray and Ellis 1993). The a priori
information is also expected to reduce the drawbacks of non-uniqueness.
The tomography imaging described here is basically a linearized probabilistic
approach because its aim is to obtain the occurrence probability of a resistivity
anomaly anywhere in the volume under investigation. As demonstrated previously, the
algorithm does not require constraints on the unknown parameters. Moreover, any
hypothetical resistivity model, matching a measured apparent resistivity data set, is
potentially included, to a rst-order approximation, within the probability tomography
image corresponding to the given set. In our strategy, the available a priori information
is of no direct methodological value. It can eventually be used in a following phase to
help interpret a geoelectric physical tomography in geological terms, if necessary.
Applications
Synthetic examples
In order to test the resolution power of the resistivity probability tomography, we rst
deal with the response of a synthetic model.
The model consists of three prismatic blocks having resistivities of 5, 10 and 500 Qm,
buried in a uniformhalf-space of resistivity 100 Qm. Figure 2 shows the plan and cross-
section of the three-prism model. Initially, we analyse the response of each prism
separately, and then the response of the combined three-prism model. The segment
AA
/
, indicated in Fig. 2, is the straight-line prole, across which the dipoledipole
pseudosection simulation was performed, using a dipolar spread and a sampling step,
each of 5 m. A set of 200 apparent resistivity values were contoured to draweach of the
synthetic pseudosections shown in Fig. 3. Random noise (5%) affects every simulation.
The results from the resistivity probability tomographic imaging are shown in Fig. 4.
The 100 Qm background resistivity was taken as the uniform resistivity of the reference
homogeneous half-space. The inverted V-shaped perturbing effects, typical of this
technique, completely vanish with this tomographic ltering.
The rst three tomographic sections in Fig. 4 show the three prisms analysed
separately. The tomographic images for these three simple cases are self-evident. A
well-contoured RAOP nucleus of congruent algebraic sign is obtained corresponding
to the actual position of each block. Considering nowthe combined three-prismmodel,
the following features can be seen in the RAOP tomography in Fig. 4d: (i) an intense
negative nucleus to the left, corresponding closely to the 5 Qm conductive body (Fig.
4a); (ii) a central positive nucleus, easily ascribable to the 500 Qm resistive prism (Fig.
4c); and (iii) a weak negative nucleus to the right, corresponding closely with the
location of the 10 Qm conductive prism (Fig. 4b). In terms of resolution, this last
feature is probably the most interesting result. Indeed, on the right-hand side of
the original pseudosection (Fig. 3d), where the 10 Qm prism is located, the only
Resistivity anomaly imaging 417
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
418 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
experimental evidence is a diffuse, weak apparent resistivity halo that can hardly be
ascribed to a buried feature.
We now consider the effects due to a change of the reference half-space uniform
resistivity. As stated above, besides the background true resistivity, the average
apparent resistivity can also be used as a reliable reference resistivity. Of course, this is
the most obvious choice for the analysis of actual eld data, when there is no indication
of which value should be attributed to the background true resistivity. In principle, we
could also take any apparent resistivity value as a reference half-space resistivity. We
analyse all of these possibilities in order to verify to what extent an apparent resistivity
difference function Dr
a
(, n) can still be signicant in this newprobability tomographic
method. Outside the interval of the measured apparent resistivity values, any other
reference resistivity would be an unrealistic assumption.
Figure 5 shows a sequence of RAOP tomographies, resulting from 13 different
reference values, selected every 10 Qmin the full apparent resistivity range 20130 Qm,
including the average apparent resistivity of 103.5 Qm. The previously discussed
tomography, using the background resistivity of 100 Qm, is shown again in Fig. 5i. We
Figure 2. A synthetic three-prism resistivity model: (a) plan view, (b) cross-sectional view.
Resistivity anomaly imaging 419
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 3. Dipoledipole apparent resistivity pseudosections for (a) the prism with 5 Qm
resistivity, (b) the prism with 10 Qm resistivity, (c) the prism with 500 Qm resistivity and (d) the
combined three-prism model. For location and size of the prisms, refer to Fig. 2.
420 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 4. Tomography images of the resistivity anomaly occurrence probability for (a) the
prism with 5 Qm resistivity, (b) the prism with 10 Qm resistivity, (c) the prism with 500 Qm
resistivity and (d) the combined three-prism model. For location and size of the prisms, refer
to Fig. 2.
observe that the RAOP tomography pattern of Fig. 5i is roughly the same as that
obtained using the average apparent resistivity (Fig. 5j), owing to the very small
difference existing between the two resistivity values.
The RAOP tomography changes signicantly if we consider the calculated
Resistivity anomaly imaging 421
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 5. Tomography images of the resistivity anomaly occurrence probability for the three-
prism model of Fig. 2, using as reference resistivity: (a) the minimum apparent resistivity 20 Qm,
(b) 30 Qm, (c) 40 Qm, (d) 50Qm, (e) 60Qm, (f) 70 Qm, (g) 80 Qm, (h) 90 Qm, (i) the
background true resistivity 100 Qm, (j) the average apparent resistivity 103.5Qm, (k) 110 Qm, (l)
120 Qm, and (m) the maximum apparent resistivity 130 Qm.
minimum and maximum apparent resistivities of 20 Qm and 130 Qm, respectively. As
can be seen in Fig. 5a, in the rst case the 10 Qm prism is the least resolved body
because of the vanishing resistivity contrast with the reference resistivity. The innitely
extended background volume of resistivity 100 Qm becomes a positive resistivity
anomaly that dominates the whole section. The appearance of the central, positive
RAOP nucleus demonstrates that the prism of resistivity 500 Qm is the best resolved
body. The use of the maximum apparent resistivity of 130 Qm leads to almost a reverse
conguration (Fig. 5m). The innitely extended background body of resistivity
100 Qm now becomes a negative resistivity anomaly that again dominates the whole
section. Nevertheless, two well-contoured, more negative, nuclei appear to originate
fromthe two smaller prisms having the lowest true resistivities. Moreover, the existence
of the central resistive prism can be inferred from the visible distortion of the negative
isolines on both sides of the prism.
All the other tomographies provide a discrete spectrum of intermediate patterns. We
observe a gradual transformation from one to the other extreme pattern, passing
through the more resolved tomographies, corresponding to the true background
resistivity and the average apparent resistivity. Intuitively, we may state that the best
resolution derives from all those reference values that induce the largest density of
vanishing values of the h
q
function. This is usually true when the choice is the
background resistivity of a large extended body, and is always true when the choice is
the average apparent resistivity.
In conclusion, we are now able to conrm the statement made previously about the
expected general equivalence of the various tomographies which can be processed
using different plausible resistivity values for the reference half-space. The images
shown in Fig. 5 demonstrate that the physical information deriving from the RAOP
tomographies has a basically invariant structure, despite the diversity of the images,
obtained by changing the reference resistivity. The resolution can, however, be
different, because it depends on the amount of contrast between the reference
resistivity and the resistivity of a target body. The analysis made so far ultimately
indicates that, within the limits of rst-order tomography, the average apparent
resistivity is the reference value leading to the most balanced resolution of all the
existing resistivity anomalies.
Field examples
We consider two eld examples. The rst example is an application to archaeology and
the second is an application to volcanology.
Archaeological application
We consider the same application dealt with in our previous paper (Mauriello et al.
1998). This was the study of the response due to a hypogeal dromos-chamber tomb
located in the Sabine Necropolis at Colle del Forno, along the Tiberine Valley 30 km
north of Rome. Two zero-offset parallel straight-line proles were planned there,
422 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Resistivity anomaly imaging 423
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 6. Archaeological application: (a) and (b) apparent resistivity pseudosections; (c) and (d)
resistivity probability tomographies.
crossing almost perpendicularly an elongated morphological feature on the surface,
very probably caused by the presence of a tomb. The two proles were assumed to
cross the entrance corridor (dromos) and the main body (chamber) of a tomb,
respectively.
Figures 6a and b showthe corresponding pseudosections, each obtained with a set of
152 apparent resistivities measured with a constant dipole spacing of 0.5 m. The eld
data are taken from Cammarano et al. (1998). The rst pseudosection (Fig. 6a) is
characterized by a pronounced inverted V-shaped low-resistivity anomaly, while in the
second pseudosection (Fig. 6b) a resistivity high appears to dominate the surveyed
area. A less well-dened inverted V-shaped effect and a resistivity low on the left-hand
side of the prole also characterize the second pseudosection.
Figures 6c and d show the probability tomographies across the two proles. A
uniform half-space with a background resistivity of 24 Qm has been assumed as
mod
0
. This gure is known to represent the true resistivity of the dominating
geological formation (lithoid tuff), hosting the Sabine Necropolis. In particular, the
tomography of Fig. 6c, which corresponds to the pseudosection of Fig. 6a, is
characterized by a well-contoured negative nucleus centred at about 1.2 m depth.
Thus the probability of occurrence of a resistivity decrease below 24 Qm is very
high. Accounting for the position, shape, extension and sign of the nucleus, the
interpretation is that we may have detected the suspected dromos, provided that we
assume its interior to be totally lled with conductive loose sediments. Actually, this
is a situation which has frequently been met during the exploration of many other
tombs in the Sabine district. Accordingly, the tomography of Fig. 6d, which
corresponds to the pseudosection of Fig. 6b, is characterized by an equally well-
contoured positive nucleus appearing in the depth range from 1 to 2 m below
ground level. Thus the probability of occurrence of a resistivity increase above
24 Qm is very high. The larger vertical and horizontal extensions of the nucleus
would now suggest the existence of an empty cavity representing the original
chamber. Given also that the two inversely signed nuclei occupy the same position
along the two parallel proles, the probability that we may have detected a unique
well-preserved dromos-chamber system is very high.
Finally, to justify the RAOP negative nucleus on the left-hand side of the tomography
of Fig. 6d, we assume that a second deeper chamber tomb with collapsed material
inside may have been detected. Since no evidence of the entrance corridor appears in
the tomography of Fig. 6c, the interpretation is that the dromos of this new suspected
tomb is probably located on the opposite side, i.e. along the opposite slope of the
oblong hill, below which the whole necropolis was excavated.
Volcanological application
We consider nowan application of the dipoledipole geoelectric technique to the study
of the shallower plumbing system of La Fossa volcano on the island of Vulcano in the
Aeolian archipelago, Southern Italy.
424 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
The island of Vulcano was characterized in the past by intense explosive activity. To
date, the observed phenomena include the notable increase of temperature and
emission rate of La Fossa crater fumaroles, indicating an increase in the ux of hot deep
magmatic gases (Barberi et al. 1991). In order to locate the shallow fracture system
through which the hot gases migrate upwards, several geoelectric proles were
acquired at the base of La Fossa cone. We show the results obtained along one of these
proles, located as in Fig. 7, where a cyclic apparent resistivity time variation was
measured during approximately four years of observations, from1992 to 1996. During
this period the chosen prole was measured 14 times at irregular sampling intervals,
ranging from 1 to 10 months. Figure 8 shows the complete set of pseudosections,
redrawn after Di Maio et al. (1997).
Resistivity anomaly imaging 425
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 7. Volcanological application: the Vulcano island survey area.
426 P. Mauriello and D. Patella
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
In Fig. 8, an alternate sequence of apparent resistivity decrease and increase phases is
clearly evident, especially in the central sector of the prole. In this sector, unusual
variations in the apparent resistivity were recorded at some points, of up to ve orders
of magnitude, from values as low as 10 Qm to values as high as 10
5
Qm. To show the
resistivity variations more clearly, each pseudosection is drawn with its own colour
scale, which is not unique for the whole set, because the range of apparent resistivity
variation differs from one plot to another.
The observed apparent resistivity changes have led to the conclusion that a three-
step cyclic mechanism, say hot magmatic gas uprising/water-to-vapour transforma-
tion/inland sea water recall, would probably occur in this zone of intense secondary
volcanic activity (Di Maio et al. 1997).
In order to improve the probability tomography images, a mod
0
corresponding to the
average apparent resistivity has been adopted, case by case. Figure 9 shows the
complete set of tomographies corresponding to the pseudosections of Fig. 8. The main
evidence is the very neat migration dynamics of the RAOP high, which would
correspond to the variable position of a gas-vapour-rich bubble during its ascent
towards the free surface, according to the above volcanological hypothesis. Moreover,
the RAOP low, repeatedly occurring at the right-hand side of the prole, probably
indicates a sea-water recall inland, triggered by the pressure drop within the pores, left
empty behind the migrating bubble.
Conclusion
We have addressed the problem of geoelectric probability tomography by considering
the resistivity parameter. Our purpose was to provide a simple tool for image
reconstruction of the most probable location of resistivity anomalies underground, in
the most objective way. The important aspect of the analysis developed above is that
the resistivity signatures are considered only from a probabilistic viewpoint. This is
quite a new concept in geophysics, which conforms to the inherently uncertain nature
of the geophysical interpretation process.
To this end, it is worthwhile pointing out that the probability concept, introduced
here, is not the direct consequence of statistics performed on a set of repeated
measurements, representing the different responses of a simulated buried system in the
presence of varying sources of error. It is much more: it is the consequence of the
intrinsic non-uniqueness of the geophysical solution. Thus the statistical basis consists
of many models providing, within the accuracy of measurement, equivalent responses
that can in no way be distinguished from one another.
Resistivity anomaly imaging 427
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429
Figure 8. Volcanological application: apparent resistivity pseudosections, measured in (a) June
1992, (b) April 1993, (c) September 1993, (d) November 1993, (e) December 1993, (f) April
1994, (g) June 1994, (h) July 1994, (i) December 1994, (j) January 1995, (k) February 1995, (l)
May 1995, (m) October 1995, (n) May 1996.
4
2
8
P
.
M
a
u
r
i
e
l
l
o
a
n
d
D
.
P
a
t
e
l
l
a

1
9
9
9
E
u
r
o
p
e
a
n
A
s
s
o
c
i
a
t
i
o
n
o
f
G
e
o
s
c
i
e
n
t
i
s
t
s
&
E
n
g
i
n
e
e
r
s
,
G
e
o
p
h
y
s
i
c
a
l
P
r
o
s
p
e
c
t
i
n
g
,
4
7
,
4
1
1

4
2
9
Figure 9. Volcanological application: resistivity probability tomographies. For the sequence of the tomosections from(a) to (n) refer to Fig. 8.
References
Barberi F., Neri G., Valenza M. and Villari L. 1991. 19871990 unrest at Vulcano. Acta
Vulcanologica 1, 95106.
Cammarano F., Mauriello P., Patella D., Piro S., Rosso F. and Versino L. 1998. Integration of
high resolution geophysical methods. Detection of shallow depth bodies of archaeological
interest. Annali di Geosica 41, 359368.
Di Maio R., Mauriello P., Patella D., Petrillo Z., Piscitelli S., Siniscalchi A. and Veneruso M.
1997. Self-potential, geoelectric and magnetotelluric studies in Italian active volcanic areas.
Annali di Geosica 40, 519537.
Hanneson J.E. 1990. A model for interpreting IP/resistivity data from areas of steep dip and thin
overburden. In: Induced Polarization, Applications and Case Histories (eds J.P. Fink, E.O.
McAlister, B.K. Stenberg, S.H. Ward and W.G. Wieduwilt). Series Investigation in Geophysics
4, 128149. SEG, Tulsa.
Loke M.H. and Barker R.D. 1995. Least-squares deconvolution of apparent resistivity
pseudosections. Geophysics 60, 16821690.
Mauriello P., Monna D. and Patella D. 1998. 3D geoelectric tomography and archaeological
applications. Geophysical Prospecting 46, 543570.
Oldenburg D.W., McGillivray P.R. and Ellis R.G. 1993. Generalized subspace methods for
large-scale inverse problems. Geophysical Journal International 114, 1220.
Park S.K. and Van G.P. 1991. Inversion of polepole data for 3D resistivity structure beneath
arrays of electrodes. Geophysics 56, 951960.
Patella D. 1997. Introduction to ground surface self-potential tomography. Geophysical
Prospecting 45, 653681.
Resistivity anomaly imaging 429
1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 411429

Das könnte Ihnen auch gefallen