J. Geodynamics Vol. 19, No. 2, pp.

141-157, 1995
Coovrieht 0 1994Elsevier Science Ltd
Pergamon Printed’& &ea~Britain. All rights reserved
02663707(94)EOOO4-E 02&t-3707/95$7.00+ 0.00

METHODS AND EXPERIENCES OF HIGH PRECISION

GRAVIMETRY AS A TOOL FOR CRUSTAL MOVEMENT
DETECTION

ERWIN GROTEN’* and MATTHIAS BECKER*

’ Institute of Physical Geodesy, Technical University of Darmstadt, Petersenstrasse 13, 64287 Darmstadt,
Germany and * Institute of Applied Geodesy, Richard Strauss Allee 11, 60598 Frankfurt, Germany

(Received for publication 7 February 1994; accepted 6 April 1994)

Abstract-Whenever vertical dislocation is of primary interest, purely geometric surveying

techniques, such as GPS, VLBI etc, are deficient in detecting and interpreting geodynamic
phenomena such as subsidence and uplift. Absolute gravimetry meanwhile became so accurate
that it can successfully be applied under usual conditions. However, for dense field observations in
regional and/or local studies relative gravimetry still cannot yet be fully replaced by absolute
gravimetry. The basic advantage of gravimetry, in combination with satellite or traditional
surveying techniques, lies in its “integrating effect”, which means that height and gravity changes
together enable, at least in principle, the solution of a four-dimensional boundary value problem,
so that a clearer interpretation of uplift and subsidence becomes available. Two examples are
outlined where gravity networks are intended to resolve vertical displacement or changes in
potential in relation to earthquake events. A local network in the Friuli seismic risk zone, where
highest accuracy of about 5 microgalt is obtained, is compared to a large scale gravity net in South
America, where lb-20ygal accuracy is demonstrated for gravity variations between consecutive
epochs. However, in both cases deficiencies still exist which make it necessary to carry out
additional observations before unique results can be expected. In case of South America extreme
environmental conditions prevented the accuracy usually achieved with 3 or 4 repeated campaigns
to be obtained. In case of Friuli the second measurement clearly reveals still unexplained
phenomena when compared with the first campaign. Thus possibilities as well as limitations are
pointed out.

1. INTRODUCTION

Geodetic techniques in monitoring recent crustal movements gained increased

interest within “zero-frequency-seismology” since it became clear that large scale
phenomena in plate tectonics could well be detected by precise geodetic survey
methods such as VLBI, GPS on the one hand (Shapiro, 1986) and local
measurements (Vyskocil et al., 1991) on the other hand. Nevertheless, gravi-
metric techniques (Groten, 1987) as well as topographic methods in exploring
* To whom all correspondence should be addressed.
t Throughout this paper the unit microgal, with 1 microgal = 10 nmls’, is used for convenience.

141
earthquake prone areas date back to work more than half a century ago as
discussed in Jeffreys (1924, 1976). With highly precise geodetic gravimetry even
more detailed phenomena than topographic features, Bouguer anomalies etc.
can be used; thus water table variations associated with earthquakes (Ma et al.,
1989) and recent crustal movement events become detectable by repeated
observations or continuous monitoring. If we take into account that interfero-
metric SAR (Synthetic Aperture Radar) satellite techniques. e.g. with the
ERS-1 satellite, reveal a spatial resolution of + 2.8 cm (Massonnet, 1993) then
even radar methods may in the future become a relevant monitoring technique in
detecting a larger area of recent crustal movements with respect to a neighbour-
ing zone of fixed crust. There is a variety of gravimetric techniques, part of them
was undergoing serious criticism, such as the arguments against dilatancy effects
(Roth, 1993) associated with micro-cracks which may occur only in the case of
shallow focus earthquakes. Such types of earthquakes exist. e.g. in the seismic
belt of Central China. in Ganzou Province where, consequently, still tidal
instruments are being used in the hope to detect secular tilts and uplift
phenomena associated with significant dilatancy phenomena. However, until
now such phenomena were not detected; there was only one major earthquake
in the mid-1980s, so chances were small in the recent past.
Most earthquake predictions and recent crustal movement phenomena
(Chorowicz and Deffontaines, 1993; Madariaga, 1986; Malone, 1986; Mantovani
and Boschi, 1986; Mukhopadhyay and Sen, 1986) are assumed to follow tectonic
plate motion theory, which is still based on rigid plate motion; alternatives
(Keith, 1993) in terms of viscous mantle flow were developed but are still not
often used even in large scale plate motion interpretation. It is, nevertheless,
obvious that rigid plate motion theory suffers from a variety of deficiencies.
The well known seismic gap concept of Fedotov and Mogi which was
meanwhile applied to a greater number of areas (Sykes, 1986), led us to establish
almost a decade ago gravimetric profiles in Chile, Argentinia and Bolivia.
East-West profiles are running from the Pacific coast across the Western
Cordillera up to Salta (Argentina) and North to South from La Paz to Santiago,
with ties to the aseismic areas in Argentina. Repeated high precision measure-
ments are taken at intervals of 3 yr. However, as discussed below, the observed
gravity changes as well as GPS position variations are still inconclusive and,
consequently, do not yet yield significant secular motion rates.
In contrast to the subduction process at the continental margin in South
America the collision at the nothernmost part of the plate boundary between the
Eurasian and African Plate leads to a rather complex plattern of tectonic
processes. In Friuli, where the lithospheric thickness varies significantly and
earthquake activities are well known (Martini and Searpa, 1986), a local
gravimetric testnet was established at the end of a GPS-chain across the western
Alps as major vertical and horizontal motion is expected there. As only one
repetition is available there, only preliminary information is at our disposal
which nevertheless gives some interesting insight to the gravimetric method.
 Gravimetry as a tool for crustal movement detection 143

In the two latter cases similar reasoning as in Caputo et al. (1986) may be
appropriate in view of maximum and minimum Bouguer anomalies. However,
presently the interpretation of recent crustal movements in terms of predictable
seismic and other phenomena is mostly required. Insofar direct deformation
rates, several cm per year in California, as discussed in Sykes (1986), are of
interest for disaster prevention and related topics.

2. THE ROLE OF GPS-MEASUREMENTS

In general GPS-measurements are presently preferred in monitoring recent

crustal movements for two reasons: (1) contrary to very local stationary geodetic
monitoring by tiltmeters or extensometers GPS measurements can be applied
over quite a large range of distances so that very local effects along ruptures,
fissures etc. can be averaged out. On the other hand, GPS-dislocations represent
variations in discrete points or stations whereas gravimeters (stationary or in
repeated campaigns; absolute as well as relative) measure an integrated effect
which contains the local uplift together with a regional potential change caused
by mass dislocations; the separation of both effects is difficult unless a combi-
nation of gravity variation and station-dislocation in the sense of a time-
dependent Boundary Value Problem is being solved for. Moreover, environ-
mental effects (such as water-table variations) and other undesired perturbations
(such as barometric pressure and associated non-tectonic height variations), may
“pollute” the results.
Nevertheless, also the purely geometric height variations as obtained, e.g.
from GPS are not free from biases because they are affected by the transforma-
tion from space-fixed to earth-fixed reference frames (mainly polar motion) and
the time dependent variations in the geocenter. Therefore, and to establish a
proper relation to an equipotential surface it makes sense to use a multi-
instrument approach. In addition, even the purely geometric position and height
variations, as obtained from GPS, VLBI and Laser-Ranging, are not free from
“corruption” such as barometric height changes which may perturb tectonic
interpretation. For the solution of the aforementioned time dependent
Boundary Value Problem, the combination of geometrical and physical methods
is a must. Only by a combination of the modern satellite techniques and
gravimetry, levelling or monitoring of ocean tide gauges the relation to the
physically meaningful reference surface of the geoid (or some other equipoten-
tial surface) can be established.

3. THE VERTICAL COMPONENT

The discrepancy between recent and long-term secular dislocation rates leads
to the use of long-time records-such as tidal recording-and to the relation of
144 EKWIN GKOIFN and MAIIHIAS BECKER

tide gauge to GPS or VLBI or SLR networks, also in order to separate the
variations of geocentric tide gauge positions from mean sea level and associated
loading variations. This again illustrates the transition to rather complex moni-
toring systems where the aforementioned potential changes again are taken into
account. Typical cases of this kind are investigations in active areas such as
Fennoscandia and the Mediterranean Basin, where recent GPS-campaigns led to
the first results which even include loading effects. These areas are characterized
by significant postglacial and tectonic motion, respectively, whereas the sea level
changes are relatively small and uncomplicated in modeling processes.
The introduction of permanent global orbit prediction services of high accur-
acy by the International Association of Geodesy such as IGS, GIG etc. led to a
substantial improvement of GPS surveys. To coordinate and combine regional
surveys like pieces of a puzzle game makes it possible to refer all such
observations to a single reference frame and, consequently, infer large-scale
phenomena, too.
The fact that, over longer distances, the errors in the frames of reference make
it often impossible to clearly separate vertical from horizontal dislocation vector
components in space geodesy, led to the tendency to express the results in terms
of rotation-invariant distance variations. This fact underlines the need for
extremely precise reference frames (Shapiro, 1986) in modern space geodesy
with subcentimeter accuracies over large distances.
Tendency goes away from simple presentation of recent crustal movements
without futher modeling, modeling becomes more complex in order to detect
mechanisms behind the pure kinematical effects. Thus purely kinematic are
replaced by kinematic-dynamic studies. Only in this way deterministic and/or
stochastic prediction can become available. Things are too complicated in most
cases for single linear interpolation and extrapolation of geometric phenomena.

3. GLOBAL CHANGE

The prediction of geodynamic phenomena and the ability of disaster preven-

tion becomes more and more a criterion for the usefulness of methods and
techniques. In studying long-time processes this often leads to the need to apply
highest accuracy in order to get maximum reliability for long-term prediction. In
the case of the problems outlined in section 3 this means that global high
precision reference frames are needed for spatial and temporal extrapolation.
This leads to combined efforts in space and geodynamic research as geodynamic
processes affect orbit prediction (in the form of ocean tides etc.) and, vice versa,
precise orbit enables better accuracy in positioning in the terrestrial reference
frame. This interrelation necessitates the use of reference frames such as ITRF
as clearly defined and determined time-dependent systems instead of systems
such as WGS 84, even if, at present, such frames still appear to be appropriate
for local or small scale surveys.
 Gravimetry as a tool for crustal movement detection 145

5. GLOBAL GENERAL GEODYNAMICS

Okubo and others have outlined procedures for using high-precision (ocean)
surface information in getting improved information on underground processes
using potential theory. Such models can be formulated independently of any
assumption on (rigid) plate tectonics. With increasing accuracy of space radar
and altimetry technology, as mentioned in the introduction, this type of infor-
mation may become increasingly valuable in the aforementioned complex
prediction processes. Insofar the following few numerical examples may just be
considered as small isolated components to be integrated in a larger complex of
geodynamic considerations. This is the reason why, at the beginning of this
paper, we mentioned a number of apparently independent approaches, carried
out at different parts of the earth at different times. The aim must be to finally
combine and relate them in a uniform modal in order to be able to get prediction
abilities.
The possibility to monitor the mean sea level with precision of a few
centimeters by combining, e.g. Topex-Poseidon and ERS-1 altimetry can at least
lead to solutions within regional areas. It is sometimes questioned whether these
solutions are really accurate enough over larger areas to determine large scale
phenomena such as ocean tides with sufficient accuracy for global interpretation.
Insofar the number of “still open questions” is significant.

6. GEODETIC MODELING

Geodetic modeling of aseismic as well as pre-, co- and postseismic motion is,
in general, not following the schemes of seismic source mechanisms in terms of
strike-slip, normal and other faulting types; there are a few exceptions such as in
Okubo (1992); other seismically oriented geodetic modeling is found in several
papers by Okubo and Sun. More often geodetically oriented models such as
point masses or other simple geometries are preferred. There is a great variety of
mass and structure modeling based on different types of observations. To infer
deformation or structures from observables then leads to well known inverse
problems (Zhao and Chao, 1993a, 1993b); modeling of gravity and/or potential
(geoid) changes associated with dislocation and deformation was considered by a
great variety of authors such as Okubo (1991), Groten (1991) (see also Chao et
al., 1993; Savage, 1984; Walsh and Rice, 1979; Wong and Walsh, 1991) where
specifics were considered, such as volcanic deformations. Varga and co-workers
focused on strain and stress fields for specific cases (Varga, 1992; Varga et al.,
1993). Also variations of deflections of the vertical were considered by various
authors, based on astronomical measurements where Hu et al. (1988) even
considered earthquake prediction based on it. Dilatancy modeling (Groten,
1978) was mentioned already earlier. Consequently, at least in principle, com-
plex modeling based on different types of observations is possibly leading to
combination results, where strain, potential and gravity differences and disloca-
tion fields are derived, even including oceanic areas and sea surface displace-
ments.
The geodetic treatment and evaluation of deformations and structures is
hampered by two basic deficiences: (1) the well known infinity of solutions of the
inverse gravimetric problem where external information is necessary to limit the
variety of possible solutions: the assumption of harmonic or bi-harmonic density
distributions is one of the (more or less arbitrary) methods in that case; (2) the
Datum problem associated with relative measurements where repeated obser-
vations are related to each other by more or less arbitrary assumptions (inherent
in free adjustment, keeping one or several points of a network fixed); (3) secular
drift in observations which mainly affects stationary observations so that only
periodic or aperiodic phenomena can be derived whereas events linear in time
(or similar) can seldomly be detected. The example discussed in detail in the
following section is affected by ( 1) and (2).
The datum problem for dislocations can be avoided by the transition to strain,
but in case of vertical displacement the treatment in terms of dislocation is often
preferred.
In this paper we focus on the Friuh area and the Southern Central Andes
which are active seismic areas as outlined before and repeated gravimetry is used
as a particular case even though the length of the observation series is not yet
sufficient to get conclusive results.

7. BASIC‘ IMPLICATIONS OF PRECISE GRAVIMETRY

Looking at the measurement of gravity and its minute variations, three groups
of techniques and associated instruments have to be considered. These are the
stationary recordings of tidal and other gravity changes, the absolute gravity
apparatus and the classical relative gravimeters.
In the first field of tidal recordings the introduction of the superconducting
gravity meters by Goodkind (1986) led to a giant step forward. Gravity
variations are monitored with a resolution of one-thousandth of a ,ugal (or one
part in 10~” m/s’) and with an accuracy of about 0.01-0.1 pgal. This instrument
led to a great improvement of tidal models, modeling of atmospheric pressure
induced gravity variations and all kinds of transient effects in gravity up to polar
motion research (Richter, 1986). However, due to the limited number of about
15 instruments worldwide and due to the fact that they cannot be used for field
applications up to now. the implication for geodynamical research is very limited
and constrained to basic modeling of environmental effects,
The measurement of absolute gravity (Timmen et al., 1992) has experienced a
great deal of technological improvement. State of the art is now demonstrated by
the FG.5 Absolute Gravimeter (Carter et al., 1994). This instrument, developed
by the U.S. National Institute of Standards and Technology, the U.S. National
Oceanic and Atmospheric Administration NOAA and the Institute of Applied
 Gravimetry as a tool for crustal movement detection 147

Geodesy IFAG, Germany, based on ideas of one of the pioneers of absolute

gravity, J. E. Faller, is reported to have an worse case instrumental uncertainty
of 1-2pgal (Niebauer et al., 1993).
If this accuracy, being roughly 0.5 to one order of magnitude better than that
of the predecessor-line of JILA absolute meters (Peter et al., 1989; Torge et al.,
1987), will be confirmed in more field tests and repeated campaigns the
establishment of basic reference sites and base networks will be greatly
improved.
Nevertheless, similar reasoning as for the superconducting instruments holds
for the absolute meters. There are, up to now, only about 15 instruments
working worldwide (only eight of the latest generation FG5) and rather stringent
requirements for the site-selection apply, not to mention the cost of the
instruments. So besides the work now being initiated by National Institutions
and a few Research Institutes in a limited number of projects, there is still room
and need for relative gravity work in geodynamical areas.
Looking at relative gravimetry it is basically associated with LaCoste G- and
D-Meters. Although there is a new alternative propagated now with the
SCINTREX-quartz-spring gravimeter CG3 (SCINTREX, 1989), there is too
little experience and published research data to judge about this instrument.
The “LaCoste Gravimetry” as applied to field work, astonishingly enough, did
not experience any major technological improvement during the last 30 years or
so. The basic accuracy the LaCoste meters are capable of was demonstrated, e.g.
by Kiviniemi (1974) with the design and performance of the Fennoscandian land
uplift lines gaining an accuracy of about 5pgal. Since then, the introduction of
the Model D gravity meter allowed this accuracy to be obtained by a wider range
of users and in a wider area of applications by the possibility to control the
instrumental calibration errors (periodical errors, non-linearities) to a greater
extent (see Dragert et al., 1981).
The other technical improvement was introduced to a wider use by Harrison
and Sato (1984) with the development of electrostatic feedback systems for
LaCoste D- and G-Meters. These feedback systems, later improved and made
available to a greater number of users by Roder et al. (1984) provided higher
accuracy for laboratory and stationary measurements, like tidal recording or
indoor gradient measurements. However, for field applications the basic accur-
acy of something around 5pgal at best could not be improved due to a large
number of effects, like thermal or vibrational sensitivity, which could not be
adequately controlled or modeled up to now (see Becker et al., 1987).
Nevertheless there are improvements in relative gravimetry for geodynamical
purposes and they are due to advanced modeling of systematic and random
errors in the adjustment procedures and due to special designs of both the
gravity networks and the observation techniques. Today it is accepted that
sophisticated mathematical models using variance component estimation and
some kind of robust estimation or outlier detection for the stochastic model in
combination with a functional model taking into account drift, tares and
calibration function parameters have to be used (Groten, 1983; Becker, 1990;
Czucor, 1987).
On the other hand careful calibration of the instruments has to be performed
and maintained (Kanngieser et al., 1983) for a description of suitable calibration
lines, as the conversion of counter readings to mgal may change in time
(Atzbacher et al., 1993). In addition to these items, reliability and repeatability
of results can only be assured by using at least three instruments in parallel, but
as many as six-eight instruments are recommended, this is characterized with
the expression “randomization principle”. As a single instrument cannot be
perfectly controlled and modeled and furthermore a repair or accident with it
may destroy any hope for the one to one repetition of a campaign for the
detection of gravity variations instead of a change in calibration only, the
randomization is the only way to overcome these problems.
The need for sophisticated modeling also extends to the proof of gravity
changes between repeated epochs of measurements. Deformation analysis based
on statistical procedures is required to identify points with stable or with
changing gravity and for determining significance levels for the changes (Becker,
1990).

8. AN EXAMPLE: THE FRIULI GRAVITY MONITORING NET

The Friuli Gravity Net is an example for local monitoring of gravity changes in
a tectonically active area. The north of Italy is located at the margin of a complex
geological structure formed as a consequence of the push of the Adriatic
microplate under the European plate (Fig. 1). This leads to strong earthquakes
and horizontal and vertical movements. Besides other geodetic tools like
levelling and GPS, gravimetry is applied for the study of crustal movements and
gravity variations.
The net, as sketched in Figs 1 and 3 is almost ideally suited for high precise
gravimetry. Driving distances are short, points are well monumented and the
gravity differences are small, i.e. less than 20 mgal in the core network, and thus,
together with the proper use of the D-meter reset option, eliminating scale and
periodical error problems, the datum is fixed by the ties to an Absolute Gravity
Site repeatedly observed and controlled. Two observation campaigns separated
2 years in time, using the identical set of 3 LaCoste Model D and 1 LaCoste
Model G gravity meters, have been conducted (Becker et al., 1994b).
Figure 2 shows the gravity variations for the single sites and associated root
mean square errors. These errors are in the order of 5--8ygal and are quite
typical for this type of network (Demirel and Gerstenecker, 1990; Kanngieser,
1982).
Results are indicating significant gravity changes on at least six of 12 points.
This was quite surprising since GPS campaigns repeated over the same interval
in time do not indicate any geometrical variation of position or height. Even
though these results appear quite contradictory, the revision of the correlation of
 Gravimetry as a tool for crustal movement detection 149

I Villach
t I
, I Dobratsb
0 IO 20 30 km . 0

8 GPS + gravity
Cl Gravity
0 Tolmezzo

l GPS

Osoppu E Cumieli

site)

Fig. 1. Friuli area with selected GPS and gravity points.

gravity variations with the local geophysical and geological structures is obvious
(Figs 1 and 3). Sets of points with a common trend of gravity increase or decrease
are situated in areas separated by the major disturbance zones. Repeated
levellings, which started earlier this century and therefore also cover the time of
the 6.4 magnitude earthquake in 1976, with their last repetition in 1993, confirm
a trend of uplift (gravity decrease) north of the Periadriatic Disturbance Zone

30 -

-10 -
Z
._ T
n Fixed station
-20 -

Fig. 2. Gravity changes 1991 to 1993 and their root mean square errors.
 ERWIN GKO I I:N and MAT-I HIAS BECK~K

46.32 -
- Far distance Ana .
stations z
-20 Braulins l Cumieli
\ O-13
l Glemine
46.21 - -16
GSOPPO l

46.22 - Solim;>
Peonis
l 16
0
7 . Artegna

IO

Codroipo
fl; l So’aris 1. o

46.17 I I I I 1 Abs. site 1

12.92 12.91 13.02 13.07 13.12 13.17 13.22

Longitude

Fig. .i. Ciravity change\ at the (iemona core gravity net @gal)

(Marson, private commun. 1994). Although further repetitions are required for
a confirmation of trends, gravity may show actual geophysical phenomena not
affecting the purely geometric quantities sensed by GPS. This is another example
that only a combination of different techniques, like gravimetry, GPS and
wherever possible geodetic levelling, is required to uncover the complete picture
of geodynamical effects.
Another interestings aspect in the Friuli network is the demonstration of the
still problematic modeling of the individual instruments. As seen from Fig. 4,
single instruments on a few sites show discrepancies of up to 20pgal. Up to now
there is no explanation for this and it shows, that even highly sophisticated
techniques of observation, corrections and adjustment do not assure conclusive
results in general. There still seems to be a kind of “hidden variable”, an
imperfection in the stochastic or functional model for precise gravimetry, not
fully understood and asking for the use of several instruments in parallel to make
use of the randomization effect.

‘l‘able I. Accuracies ofdittcrent adjustments for

the Friuli Net

II I, R.m.s. R.m.s.d

IYYI 29x 102 &7.X6 ?h.S

19Y.) 296 Y3 &X.00 f4.1
All Data so2 155 t7.0 5.5.X

n = number of observations; u = number of

unknowns.
R.m.s. = root mean quart error of unit weight
(single gravimeter reading) in pgal.
R.m.s.d. = average root mean square error of
adjusted gravity differences in pgal.
 Gravimetry as a tool for crustal movement detection 151

1 I I I 1 I I
_
152 ERWIN GKO.W.N and MATTHIAS BEC.KEK

r I
\

’
; ,-_ LaPaz
r -6

St cruz
d ONN
+4
-19

0 Huari

i-

24°C Ant

+I5

28”’

/
i o San Juan
32’S ‘I

\
+51 0 Mendoza
\ 0
Sytiago ,
I
72” w 68” W 64” W

Longitude

Fig. 5. Gravity at ABC Profiles I YW

9. LARGE SCALE GRAVIMETRY MONITORING

An attempt to monitor the subduction process of the Pacific Plate under the
South American Plate along a large portion of Chile’s Northern Pacific coast was
made with the installation of the “ABC-Profiles” in 1984 (Becker et al., 1989).
About 30 main stations and the same number of additional sites were established
 Gravimetry as a tool for crustal movement detection 153

in Argentina, Bolivia and Chile and repeatedly observed in three major cam-
paigns 1984, 1987 and 1990. Figure 5 shows the main gravity stations connected
by the profiles. Quite contrary to the situation in Friuli, the environmental
effects and design aspects of the network are almost the worst case possibly
imaginable. Gravity range is about 2 gal introducing all kinds of calibration
problems and the station separation of up to several hundred kilometers does not
allow daily or more frequent drift controls. Moreover stable reference sites or
absolute gravity sites are located rather distant from the zone of geodynamical
activity and the absolute sites were installed in Argentina prior to the third
campaign only. So from the beginning it was intended to use design and
observation schemes specially adapted to these difficult situations (Becker et al.,
1986) and to rely on the randomization effect by using six-eight gravity meters
identical to all campaigns.
Of special importance in the absence of several absolute sites covering the
complete gravity range is the scale control of the gravity meters used. As shown
in Atzbacher et al. (1993) the scale factor of LaCoste instruments is subject to
minute changes in time in the order of a few parts in 10e5. This introduces an
error of several tens of pgal in gravity differences of about 800 mgal as directly
observed, e.g. between La Paz and the Pacific Coast. Therefore it has to be
assured, that at least the sum of the scale changes of the set of gravity meters
used is zero or that larger scale factor changes are determined on calibration
lines. This also means, that the design of the network should be such that the
individual gravity values of the sites allow a discrimination of an overall scale
change in the network and of the set of gravity meters used based on the
geological and geophysical settings for the expected pattern of possible gravity
variations.
As for the ABC profiles the strategy worked quite well as can be inferred from
Fig. 5 and Table 3. Figure 5 shows the mean value for each major zone which was
to be computed from a set of l-4 individual sites at each area in order to check
the local stability. Table 3 shows the individual changes and confidence levels.
With the exception of some stations with probably local effects, most of the sites
seem to have been stable between 1984 and 1990. Gravity variations between
epochs are determined with errors of lO-15pgal and the variations itself are in
the order of lo-20pgal and so not significant up to now.
As can be seen from Table 2 and compared to Table 1 for the Friuli net the
accuracy, as far as represented in the errors of the adjustment, is inferior by a
factor of 2-3 only.
The errors of relative gravimetry and possible height variations as deduced out
of them at present are comparable to those of GPS and tide gauge records if used
for the analysis of time dependent variations in height in this special project
(Becker et al., 1994a). In conclusion it must be assumed that rates of change of
gravity and/or height are smaller than initially estimated or that changes do not
take place continuously but in descrete events related to major earthquakes,
which have not been observed close to the main gravity sites in the time up to
 ERWIN GKOTCN and MUTHIAS BIXKW

‘I‘ablc 2. Accuracies of different adjustments

for the ABC-Profiles

Year II u R.m.s. R.m.s.d.

lYU4 531 173 _I.

+73.0 + 16.5
1987 1252 377 217.5 ill.2
I990 1436 46X tw 6
___. + 14.5

II = number of observations; u = number of

unknowns.
R.m.s. =root mean square error of unit
weight (single gravimeter reading) in ugal.
R.m.s.d. = average root mean square error of
adjusted gravity differences in ugal.

now. Geodynamical interpretation still has to be based on geological information

covering much larger time spans. Large or global scale vertical deformations are
not as easily determined as horizontal rates, where geodetic measurements e.g.,
may well be used to constrain models of plate motions. In any case more
repetitions, especially in the case of seismic activity, is required and highly
desirable for the ABC profiles.

Table 3. Gravity changes and significance-tests for the ABC-profiles

1984-1990 (ugal)

Confidence levels for

Station Delta g R.m.s. for Fisher- and r-test

Africa I -16.X kY.h F-test. 99 I% r-test: 7X.2%

Africa 7 -40. I It IO. I F-test; lo&” r-test: 84.4%
Africa 4 3. I t9.x F-test: 34.9’Y
I 0 f-test: 19.7’Y
Iquique 1 -0.h 28.X F-test: S.l’Y” r-test: 4.1%”
Antofagasta I 4.5 +x. I F-test: 41 .6% r-test: 32.2’j6
Antofagasta 2 IO.6 k9.2 F-test: 74.2% f-test: 54.6%
Antofagasta 3 3.3 i8. I F-test: 31.3% f-test: 246%
Antofagasta 4 27.x k9.3 F-test: 99.4% I-test: 79.4%
Santiago I 46.3 rf-IO.8 F-test: 100.0% f-test: 85.4%
Santiago 2 56.1 t12.7 F-test: 100.0% f-test: 85.8%
X42023 -X.5 tx.4 F-test: 68.0% t-test: 50.4%
San Pedro I -6.5 2Y.S F-test: 50.2% r-test: 38.3%
X42025 7.9 +15.x F-test: 37.8% r-test: 29.5”/0
842026 IO.6 + 16. I F-test: 48.5% f-test: 37.1(X
Ollagur 39.7 + 17.5 F-test: 96.9% r-test: 73.6%
Calama -4.7 ?I 17.0 F-test: 21.7”/ r-test: 17.3%
Toconao -46.X 2 IS.0 F-test: 99.6”/1 t-test: 80.2%
Huatiquina -27.X 221.5 F-test: 79.3% r-test: 58.1%
Baquedano -23.3 &II.6 F-test: 94.6% t-test: 70.6%
X42032 10.5 215.X F-test: 48.7% f-test: 37.3%
Salta 1 16.1 f20.2 F-test: 56.9% f-test: 43.0%
Mendoza 0.4 + 16.8 F-test: 2.1% r-test: 1.7%
842101 -18.2 t IX..5 F-test: 66.6% I-test: 49.5%
842102 16.9 + 14.3 F-test: 75.3% f-test: 55.3%
La Paz --s.9 f9.0 F-test: 48.3”/, f-test: 37.0%
Oruru 4.4 * 17.3 F-test: 19.9% f-test: 15.8%
Uyuni -26.5 k23.3 F-test: 73.5% r-test: 54.1%
St. Cruz -18.X + 14.7 F-test: 78.6% r-test: 57.6%
 Gravimetry as a tool for crustal movement detection 155

10. CONCLUSION

Summarizing the experiences in precise gravimetric work it can be stated that

an old tool is still of interest. This holds in spite of the fact that GPS is certainly
more easily applied in many cases where geometrical deformation is of primary
interest. Todays resolution of GPS, where the global tracking network and up to
date receiver technology allows accuracies of l-3 cm over 1000 km baselines,
clearly is superior to the vertical resolution of gravity variations if reasonably
converted to height changes. Nevertheless the modeling of tectonic processes
requires the knowledge of potential changes and so gravimetry is a must. It is
clear that, especially with the advent of the latest absolute apparatuses, a
suitable combination of absolute and relative stations has to be used. Relative
gravimetry, LaCoste gravimetry, is required and is able to fill the gaps between
absolute sites and in those areas not suited for absolute measurements.
Quick-look and task-force applications on local profiles in case of seismic events
are, at least today, only possible by the use of relative instruments. The overhead
of organization and costs for the absolute apparatuses is still a limiting factor
especially for the less developed countries.
With better global surveys of topography, pattern recognition and radar
techniques may gain more interest besides GPS and other techniques which are
presently preferred. The use of superconducting and absolute gravimetry in
combination with more precise tide gauge results may substantially improve the
separation of tectonic from other phenomena and thus enable more exact
modeling. The results in terms of gravimetric and geodetic results should be
considered as isolated partial studies which need continuation over more than
one decade to lead to conclusive results and have to be seen as first attempts to
be integrated into complex interdisciplinary approaches. In these combinations
the overall scaling is of utmost importance; this concerns the dynamic (such as
gravimetric) as well as geometric quantities.
Recent crustal movements now appear closely related to seismological and
disaster prevention aspects and global change. The transition from typical short-
term solid earth models (elastic etc.) to long-term modeling (viscous etc.) still
poses problems, in general.

REFERENCES
Atzbacher Kai, Becker M., Gerstenecker C. and Miiller K. (1993) The scale factors of LaCoste and Romberg
gravimeters-accuracy and temporal changes, paper presented at Meeting of the European Geophysical
Society, Wiesbaden, April 1993.
Becker, M., (1990) Adjustment of microgravimetric measurements for detecting local and regional vertical
displacements. In Graoity Gradiometry and Gravimetry, (Rummel Hipkin, Ed.) pp. 149-161. IAG
Symposium No. 103, Edinburgh, August 8-10, 1989, New York.
Becker M., Gao X. M. and Groten E. (1986) First results of precise gravity measurements on the “A-B-C”-
Profiles. Tectonophysics 130, 33-47.
Becker M., Groten E., Lambert A., Liard J. 0. and Nakai S. (1987) An intercomparison of LaCoste and
Romberg Model-D gravimeters: results of the International D-meter Campaign 1983. Geophys. J. R. Astr.
Sot. 89, 499-526.
IS6 EKWINC;KOII.N and MAI~HIAS ~t<.htK

Becker M., Araneda M., Groten E., Hirsch O., Knoll ‘I. and Ostrau S. (1989)First results of repeated gravity
measurements in the Southern Central Andes. Zhl. C;eo/. faliionl. Ted 1. 19X9 5/6. 1151-l 161,
Becker M., Groten E.. Gehlich U. and Araneda M. (lYY4a) Precise gravimetric measurements at the
ABC-profiles 19X4 to lY90. Zhl. Grol. faliiont. Teil I, 1993 l/2, Stuttgart, 191-201.
Becker M.. Groten E.. Lewi E., Marson I. and Palmieri F. (lYY4b) Repeated precise gravity measurements in
the Friuli seismic risk area. In Proc,. of‘ f/r88th Intern. Symp. on Recent Crustal Movemenls, Kohe, Japan,
Dec. IYY.? (in press).
Caputo M., Milana G. and Rayhorn J. (IYXO) ‘l‘hc state of stress in the lithosphere caused by topography and
isostatic compensation. In Earthquake.\, Ob.wr~~a~ion.Theory and lnterpretafion, (Kanamori H. and Boschi
E. eds), pp. 569-575. North Holland Puhl. Co Amsterdam-New York-Oxford.
Carter W. E., Peter G.. Sasagawa G. S.. Klopping F. J.. Berstis K. A.. Hilt R. L., Nelson P., Christy G. L..
Niebauer T. M., Hollander W., Seeger H., Richter B.. Wilmes H. and Lothammer A. (1994) New gravity
meter improves measurements. EOS, Vol. 75, No. H. February 22.
Chao D.. Xu J. and Zhao S. (1093) Gravity and geoid changes due to multifault movements. IAG Gen.
Meeting. Peking.
Chorowicz J. and Deffontaines B. (lYY3)‘l‘ranster faults and pull-apart model in the Rhinegrahen from analysis
of multisource data. JGR, 98. 1433Y- I435 I.
Czucor E. and Lux P. (1987) Hauptschwcrenetz der Bundesrepublik Deutschland 1982 (DHSNX2) fertigges-
tellt, ZfV 112, 4St)-4%.
Demirel H. and Gerstenecker C. (IYYO) Secular gravity variations along the North Anatolian fault. In Graoity,
Gradiometry and Graoimetrie, IAG Sympo.Cum No. 10.3. Edinburgh, Augusr X-10. pp. 163-169. New
York 1989.
Dragert H., Lambert A. and Liard J. 0. IYXl)
 Gravimetry as a tool for crustal movement detection 157

Massonnet D. et al. (1993) The displacement field of the landers earthquake mapped by radar interferometry.
Nature 364, 138-142.
Mukhopadhyay A. and Sen S. (1986) On a system interacting strike-slip faults in a layered model of the
lithosphere-asthenosphere system. Proceedings International Symposium on Neotectonics in South Asia
Survey of India Dehradun, February 18-21.
Niebauer T., Sasagawa G., Hollander W., Nelson P. and Failer J. (1993) A new generation of absolute gravity
meters. Metrologia (in preparation).
Okubo S. (1991) Potential and gravity changes raised by point dislocations. Geophys. J. Intern. 105,573-586.
Okubo S. (1992) Gravity and potential changes due to shear and tensile faults in a half-space. JGR, 97, B5,
7137-7144.
Peter G., Moose R. E., Wessels C. W., Failer J. E. and Niebauer T. M. (1989) High-precision absolute gravity
observations in the United States. JGR 94, 5659-5674.
Richter B. (1986) The spectrum of a registration with a superconducting gravimeter. In Proc. of the J&h Intern.
Symp. on Earth Tides, Consejo. Sup. Invest. Cient. Madrid, Sept. 1985 (Vieira R., Ed.), pp. 131-139.
Roth F. (1993) Deformations in a layered crust due to a system of cracks-modeling the effect of dike injections
or dilatancy. JGR 98, B3, 4543-4551.
Roder R. H., Schtill M. and Wenzel G.-G. (1984) Ein elektronisches Regel- und Mebsystem fur
LaCoste-Romberg Gravimeter. Zfv 109, 494-497.
Savage J. C. (1984) Local gravity anomalies produced by dislocation sources. JGR 89, B3, 1945-1952.
SCINTREX (1989) The SCINTREX CG-3 Autograv. Scintrex Concord, Ontario, Canada.
Shapiro I. I. (1986) Use of space techniques in geodesy. In Earthquakes, Observation, Theory and
Interpretation (Kanamori H. and Boschi E., Eds.), pp. 530-567. North Holland Publ. Co.,
Amsterdam-New York-Oxford.
Sykes L. R. (1986) Predicting great earthquakes. In Earthquakes, Observation, Theory and Interpretation
(Kanamori H. and Boschi E. Eds).I DO. . . 398-435. North Holland Publ. Co.. Amsterdam-New
York-Oxford.
Timmen L., Roder R. H., Schnull M. and Torge W. (1992) Large scale absolute gravity control in South
America-the JILAG-3 campaigns 1988 to 1991. Proceedings, International Conference Cartography
Geodesy, Maracaibo, Venezuela, Nov. 24-Dee 3.
Torge E.. Roder, R. H., Schiill M. and Wenzel H. G., (1987) First results with the transportable absolute
gravity meter JILAG-3. Bull. Geod. 61, 161-176.
Varga P. (1992) Complete description of forced tangential elastic deformations. Phys. Earth and Planetary Int.
73, 199-205.
Varga T., Latynina L. A., Brimich L., Mentes G., Katona G. and Varga P. (1993) Study of the extensometric
records of the Pannonian Basin in the non-tidal frequency domain (reprint).
Vyskocil P., Zeman A., Tealeb A., Mahmond S. M. and El-Fiky G. S. (1991) Vertical movements around the
Kalabsha fault northwest of Aswan Lake, from precise leveling. J. Geodyn. 14, 1249-1262.
Walsh J. B. and Rice S. R. (1979) Local gravity changes in gravity resulting from deformation. JGR, 84, Bl,
165-170.
Wong T. F. and Walsh J. B. (1991) Deformation-induced gravity changes in volcanic regions. Geoph. J. Intern.
106, 513-520.
Zhao S. and Chao D. (1993a) Determination of preseismic fault movements in the Tangshan area by inversion
of levelling data. Dept. Geodesy, Wuhan Techn. Univ. Surv. and Mapping, monograph.
Zhao S. and Chao D. (1993b) Determination of current active segments at the Red River fault zone by
inversion of GPS baseline changes manuscript. Dept. Geodesy, Wuhan Techn. Univ. Surv. and Mapping.