Many effects influence gravity measurements; they can be :
§ instrumental
effects such as :
– temperature
and/or relative humidity,
–
atmospheric
air pressure if it is not well compensated by the gravimeter.
§ geophysical
effects such as :
–
Earth tides,
–
atmospheric
air pressure,
–
rainfall and snow,
–
water table,
–
geology of the area.
§ volcanic effects
through variations of elevation or changes in mass/density.
§
a coupling effect tilt/apparent gravity : the tilt
variations induce an instrumental
effect
on the
gravimeters producing apparent
gravity changes.
II.
Order of magnitude
Volcanic activities produce gravity changes of some tens of microgals. If
we want to detect those effects, we have to correct the data with a precision
of at least 10 µgal. This corresponds to :
Ÿ
4% of the Earth Tides,
Ÿ
30 mbar of air pressure variations,
Ÿ
3 cm of elevation,
Ÿ
attraction of a 10 cm thick rock layer.
On what concerns the
temperature and the humidity, they cause an instrumental effect on spring
gravimeters which can reaches an amplitude of about one mgal as a mean value on
yearly period. The tilt effect is negligible if the gravimeter is properly
adjusted; if not, the tilt can have a non linear effect.
III.
Earth tides effect
We first
analyse the Earth tides effect (Eterna, Wenzel 1996) which can be thus
modelised with high accuracy and removed efficiently from the gravity
measurements. If we are only interested in long period gravity variations, we
can use a low-pass filtering.
IV.
Air pressure effect
Considering
the fact that, for the final residuals,
we are looking for a precision of about 1 µgal for a period of one day, a
simple linear regression using the local pressure (see table 1) is sufficient
to correct the air pressure effect. We notice that the regression coefficient
is near to –0.3 µgal/mbar except for the LCR3 and LCR8 where
it is very high indicating that these gravimeters are not well compensated any
more for air pressure variations and this probably because the pressure seals
are not more airtight (El Wahabi et al, 1997; 1998).
Table
1 :
Pressure regression coefficients in µgal/mbar (computed using Eterna Software).
They are around
–0.3 µgal/mbar except for the LCR3 and LCR8 which present obvious anomalies for
air pressure.
V.
Temperature and humidity
effects
The relative
humidity is closely dependent on the temperature. Therefore,
it is difficult to separate their effects on gravimeters. Moreover, these
effects depend on the frequency band which can be divided in three parts :
short, medium and long periods.
1.
Short periods
By short periods,
we mean the periods from some minutes to a few hours. A typical example of a
quick change of the temperature can be the switching on or off of the viewing
lamp of gravimeters. This induces
some perturbations due to the fact
that the thermostat of the
gravimeter needs a certain time to adjust itself. Fortunately, this takes only
some minutes and so has no significant influence.
2.
Medium periods
By medium
periods, we mean the periods from some days to a few weeks, and for this band
of frequencies, unfortunately, we have no example. For this reason, we decided
to make an experiment in the fundamental station of the Royal Observatory of
Belgium in order to study the temperature and humidity influences on spring
gravimeters. The station, which is in the cellar, is thermostatised
while the humidity is not controlled. Some exchanges of air masses are
possible, especially, when opening the station door. We have at our disposal, a
LaCoste & Romberg G8 and high precision pressure and temperature/humidity
sensors. Our objectives are the following :
*
the determination of the transfer function of the
couple temperature/humidity between inside and outside the station,
*
the study of the gravity changes with the
temperature variations in a stable humidity environment,
*
the study of the gravity changes with the humidity
variations in a temperature controlled environment,
*
the study of the correlation between temperature and
humidity for each frequency band to decide which parameter is more suitable to
determine the transfer function for the gravimeter responses.
As first results, we can notice in
the figure 1 a
large temperature perturbation (an increase of about 4°C) during
three days (around 18 November 1998) to which corresponds a gravity residuals variation with a change in the slope following
immediately after the steep fall of the temperature. This suggests a link
between the gravity change and the corresponding temperature perturbation.
Another
example given on the figure 2 shows a large perturbation in the temperature, a
decrease of about 5°C,
which is also correlated to the gravity residuals
with a phase lag. Comparing the temperature and the relative humidity records,
there is no clear correlation
between these two parameters. We notice indeed, during
the beginning of March 1999,
a large increase in the relative humidity of about 11%
in one day while the temperature remains stable and no significant change
appears in the gravity residuals.
However, in the middle of March, a decrease of relative humidity appears to be
anti-correlated with the man made temperature perturbation. As a conclusion, it
seems that the temperature has an effect on spring gravimeters for medium
periods while the humidity might need more time to influence the gravimeter response.
Figure
1 :
Illustration of the LCR G8 drift with temperature variations in Brussels.
From top to
bottom : gravity residuals after
removing the Earth tides and air pressure effects, gravity residuals reduced
by a polynomial fit and temperature records with a big man made perturbation.
Figure
2 :
Illustration of the LCR G8 drift with temperature and relative humidity
variations in
Brussels. From top to bottom : gravity residuals
after removing the Earth tides and air pressure effects (also reduced by a polynomial fit), temperature records and
relative air humidity measurements.
3.
Long periods
Usually by long
periods, we mean the periods of several months and especially the yearly cycles.
We shall hereafter call the drift
the non tidal apparent gravity variations after removing the pressure effect.
Generally, the recordings of spring gravimeters present an annual oscillation
of the drift which is correlated with the temperature and the relative humidity
variations. The question is now to determine whether
both of the two parameters are responsible of this drift or only one of them
and, in this latter case, which one. The answer will be illustrated by means of
the following examples.
The first one came from Centuripe,
a station in the Etna area (Sicily), where
the LaCoste Romberg (LCR) G3 was installed in September 1992. The obvious
annual periodicity, that we see in the gravity residuals,
is perfectly correlated with the temperature records (see figure 3). However, during the beginning of the measurements, the
gravity residuals show a large
perturbation (see figure 3 and 4) which coincides with the setting of a door in
the station room. We think that the door could have limited the air circulation
between inside and outside so that a new regime set in the air and heat
exchanges. Looking at the station room temperature, no significant perturbation
was recorded for the same period proving the fact that the drift perturbation
was not due to the temperature
effect (El Wahabi et al., 1997; 1998).
Figure
3 :
Illustration of the LCR G3 annual oscillation with temperature variations in
Centuripe. From
top to bottom : temperature and gravity residuals
after removing the Earth tides and air pressure effects.
Figure
4 :
Illustration of the LCR G3 drift perturbation with temperature variations in
Centuripe
(initial part of figure 3). From top to bottom : temperature and gravity residuals
after removing the Earth tides and air pressure effects.
Now, comparing
the annual oscillation of the same gravimeter, the LCR G3, in two different
stations (Centuripe in Sicily and Lanzarote in the Canary Islands), we see in
the figure 5 that the drift amplitude is more than twice bigger in Centuripe
than in Lanzarote. Therefore, the
annual periodicity is due to the
environment where we put the
gravimeters.
Figure
5 :
Illustration of the LCR G3 annual periodicity in Centuripe (Top) and in
Lanzarote
(Bottom). In Centuripe
the gravimeter had more than twice the drift amplitude recorded in Lanzarote.
In the other
station in the Etna area, Serra-la-Nave, the LCR G8 was installed in January
1992. Due to a lack of heating power, the station room was imperfectly thermostatised. We decided later, in June 1994, to
lower the thermostat. The result was
a drop of the mean temperature in the station from 23 °C to 19 °C (see figure 6). This
strong variation was not noticeable in the drift curve of the LCR G8, so that
the room temperature was not the responsible parameter of the annual
oscillation (El Wahabi et al., 1998). A cross-correlation shows a phase lag of
about 44 days between the temperature outside the station (in the open air) and
the temperature inside the station which itself is leading the gravity signal
by about 25 days. The station is thus acting like a low-pass filter between the
temperature outside and inside the station, as well as the gravimeter which is
also acting like a low-pass filter between the temperature excitation in the
station and the gravity response.
Looking now at
the residuals of the gravimeters
LCR/ZLS G385 (see figure 7) and LCR D2 (see figure 8) which are installed in
Potsdam station, we see clearly an anti-correlation with the relative humidity
while the temperature is stabilised in a range of ±0.1 °C. The phase difference
between gravity and relative humidity signals is close to 180 degree.
Figure
6 :
Illustration of the LCR G8 annual periodicity with temperature variations in
Serra-
La-Nave. From top
to bottom : gravity residuals after
removing the Earth tides and air pressure effects, the temperature recorded in
the station room and the temperature recorded in the open air. The phase lag
between the temperature outside and inside the station is about 44 days; and
between the inside temperature excitation and the gravimeter response is about
25 days.
Figure
7 :
Illustration of the LCR/ZLS G385 annual drift (Top) with humidity variations
(Bottom) in
Potsdam. The temperature is stabilised at 25°C ± 0.1°C. The lower
curve on the top graph shows the annual drift reduced
by a linear fit.
Figure
8 :
Illustration of the LCR D2 annual drift (Top) with humidity variations
(Bottom) in
Potsdam. The temperature is stabilised at 25°C ± 0.1°C. The lower
curve on the top graph shows the annual drift reduced
by a linear fit.
Some examples of Askania
gravimeters come from Pecny, Potsdam and Walferdange stations. In Pecny, where the station is thermostatised
with a precision of ±0.1 °C,
the gravity residuals of the Askania
228 show a correlation -instead
of an anti-correlation like in the case of LCR gravimeters- with the air humidity and a large apparent phase lag of about three
months (see figure 9).
Figure
9 :
Illustration of the Askania Gs 228 annual periodicity with temperature and
humidity
variations in Pecny. From top to bottom : gravity residuals
after removing Earth tides, temperature and humidity records. The lower curve
on the top graph shows the annual drift reduced
by a polynomial fit.
The Askania
222 in
Potsdam presents the same feature as the previous one (see figure 10). In 1989,
an experiment has been made in Potsdam by putting the Askania 222 in an air-sealed
container. Afterwards, the annual oscillation disappeared completely (see
figure 11).
Figure 10 : Illustration of the Askania
Gs 222 annual periodicity (Top) with humidity
Variations
(Bottom) in Potsdam. The temperature is stabilised at 25°C ± 0.1°C. The lower curve on
the top graph shows the annual drift reduced
by a polynomial fit. The temperature is controlled with an accuracy of ±0.1 °C.
Figure
11 : Illustration of the Askania Gs 222 annual oscillation (Top) with
humidity
variations
(Bottom) in Potsdam. The temperature is stabilised at 25°C ± 0.1°C. After putting the
gravimeter in an air sealed container (14 July 1989), the annual drift
disappeared.
In Walferdange,
the Askania 191/BN01 was installed in the mine where
the temperature and the humidity are very stable around 11°C and 80% respectively. The
gravity residuals show only a linear
drift (see figure 12) proving that one of these two meteorological parameters
or both of them are responsible for the annual oscillation.
Figure 12 : Illustration of the Askania
Gs 191/BN01 drift in the mine of Walferdange.
The Earth tides
and the air pressure effect are removed. No annual component is visible.
Bastien and
Goodacre (1990) made an experiment with the LCR ET12. They put the gravimeter
in a temperature controlled environment with low and high humidity level. The
result is that the annual drift apparently disappeared during
the period of the experiment with opposite quasi-linear drift rates for low and
high humidity level (see figure 13).
Figure
13 : Illustration of the LCR G12 annual periodicity in Ottawa. The drift
behaviour between
1985/1987 disappears after putting the gravimeter in controlled environment.
VI.
Conclusions
With these examples, we see
clearly that the humidity has an effect on spring gravimeters for long period
variations. It should be noticed that all the gravimeters are equipped with the
capacitive transducers and, therefore, the humidity can indeed change the
dielectric constant of the air. However, Askania gravimeters are completely sealed
against pressure variations and the LCR gravimeters are sealed with dry inert
gas inside. There should thus be no
direct influence of humidity on the capacitance, while most of the electronic
parts are exposed to the humidity changes.
When monitoring the gravity on
volcanoes, we have to record in addition at least three other parameters : the air pressure, the temperature
and the relative humidity.
Acknowledgements
We are first very
grateful to Prof. P. Pâquet for his help and encouragements. We thank Prof. B.
Ducarme as well for his valuable advice and support. We acknowledge also Mr. M.
Hendrickx for his help. Many thanks go to Mr. N. d’Oreye for providing the data
of Askania 191.
References
Bastien R. and Goodacre A.K.,
1990, The effect of humidity variations on long-term tidal gravity recordings,
Bull. Inf. Marées Terrestres, 106, p. 7506-7510.
El Wahabi A., Ducarme B.,
Van Ruymbeke M., d’Oreye N. and Somerhausen A. , 1998a, Continuous gravity
observations at Mount Etna (Sicily) and correlations between temperature and
gravimetric records, Cahiers du
Centre Européen de Géodynamique et de Séismologie, Vol.14, p.105-120, 1997.
El Wahabi A., Ducarme B.,
Van Ruymbeke M., d’Oreye N. and Somerhausen A. , 1998b, Four years of
continuous gravity observations at Mount Etna (Sicily), Proceedings of the 13th
International Symposium on Earth Tides, p.477-479, 1998.
Wenzel H.-G., 1996, The
nanogal software : Earth tide data processing package : Eterna 3.3, Bull. Inf.
Marées Terrestres, 124, p. 9425-9439.
