Long-term crustal deformation monitored by gravity and space techniques at Medicina, Italy and Wettzell, Germany

Journal of Geodynamics 38 (2004) 281–292

Long-term crustal deformation monitored by gravity and spacetechniques at Medicina, Italy and Wettzell, Germany

B. Richtera,∗, S. Zerbinib, F. Matontib, D. Simona

a Bundesamt für Kartographie und Geodäsie, Frankfurt, Germanyb Dipartimento di Fisica, Università di Bologna, Italy

Received 30 January 2004; received in revised form 24 May 2004; accepted 9 July 2004

Abstract

Series of gravity recordings at the stations Medicina (Italy) and Wettzell (Germany) are investigated to separateseasonal gravity variations from long-term trends in gravity. The findings are compared to height variations monitoredby continuous GPS observations. To study the origin of these variations in height and gravity the environmentalparameters at the stations are included in the fact finding. In Medicina, a clear seasonal signal is visible in the gravityand height data series, caused by seasonal fluctuations in the atmosphere including mass redistribution, the ocean,groundwater but also by geo-mechanical effects such as soil consolidation and thermal expansion of the structuresupporting the GPS antenna. In Wettzell, no seasonal effect could be clearly identified, and the long-term trend ingravity is mainly caused by ground water variations. The successful combination of height and gravity series withthe derived ratio of gravity to height changes indicates that the long-term trends in height and gravity are most likelydue to mass changes rather than to tectonic movements.© 2004 Elsevier Ltd. All rights reserved.

1. Introduction

Independently from space geodetic techniques like GPS, SLR or VLBI, another way to constrainvertical deformation is to use precise gravity measurements with ballistic absolute gravimeters and super-conducting (cryogenic, SG) relative gravimeters. The value of�g/�H can exhibit large variations but ismostly between−15 and−35 nm s-2/cm (Jachens, 1978). The free-air relation of−30 nm s-2/cm, which

∗ Corresponding author. Fax: +49 69 6333425.E-mail address:[email protected] (B. Richter).

0264-3707/$ – see front matter © 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.jog.2004.07.013

282 B. Richter et al. / Journal of Geodynamics 38 (2004) 281–292

mostly occurs locally, corresponds to a vertical surface movement without mass change. For large areavariations, the Bouguer relation−20 nm s-2/cm is frequently found (Torge, 1989). This relation allowsthe conversion from gravity to height changes and vice-versa. The precision of the most accurate absolutegravimeters is of about 10–30 nm s-2, equivalent to 10-9 fraction of the Earth gravity field for instantaneousmeasurements. The SG’s, the best realisation of today’s relative gravimeters, can measure continuouslychanges of gravity to a precision better than 0.5 nm s-2 using 1-min averages. Therefore, long-term gravityfield variations can be determined with precision better than 10-10. Consequently, gravimeters are sen-sitive to height changes of a few millimetres. Gravity variations converted into height changes obtainedthrough the combination of absolute and relative (SG) gravity measurements, can therefore, be comparedto those derived from the analysis of space geodetic data (Zerbini et al., 2001).

Continuous monitoring and the combination of vertical height and gravity changes allow the separationof the gravity potential signal due to the mass redistribution from the geometric signal due to heightchanges. This ensemble of different observational capacities constitutes a powerful tool for the realizationof integrated multipurpose networks suitable for high-precision monitoring of ground deformation and,more in general, for environmental research.

2. Instruments and location of the experiments

One experiment takes place at Medicina, Italy, a fiducial reference station of the space geodesy net-work near Bologna (Fig. 1a). The station is located in a low-lying sector of the middle-lower Po Plain,where the surficial sedimentary sequences are mostly fined-grained clays and silty clays. A shallow un-confined (phreatic) aquifer is present in the first few meters of the ground. Since the beginning of the

Fig. 1. Location of the stations Medicina, Italy (a) and Wettzell, Germany (b), source:http://sopac.ucsd.edu/.

http://sopac.ucsd.edu/

B. Richter et al. / Journal of Geodynamics 38 (2004) 281–292 283

observations in 1996, the maximum measured fluctuations of the water table are of the order of 2.5 m,according to the seasonal variability in terms of climatic conditions (rainfall, temperature, and relativehumidity).

Continuous GPS observations acquired with a Leica GPS system 399 receiver have been performedsince July 1996. The antenna is installed on a specifically designed steel pole screwed into the referencebenchmarks. The system and the set-up are unchanged since the installation. To monitor continuouslyvariations of the gravity field, the SG GWR C023 was installed about 600 m apart from the GPS system atthe end of October 1996. The instrument is operated in a temperature-controlled laboratory on a concretepier, which is founded 1 m deep into the ground. On another pier, next to the SG, absolute gravity (AG)measurements have been regularly performed by means of AXIS FG5 absolute gravimeters to checkthe SG data series. These observations are used to eliminate offsets in SG data series as well as theinstrumental drift, taking advantage of the AG long-term stability. In addition, relevant meteorologicalparameters such as air pressure, temperature, relative humidity, precipitation and surficial water tabledata are collected on a continuous basis. To estimate the atmospheric mass redistribution, data from12-h balloon radio sounding from a nearby regional meteorological station (approximately 3 km apartMedicina) are also collected.

Another case for long-term studies of the gravity field variations are the observations performed atWettzell, in the Bavarian Forest in Germany, a fiducial reference station for space geodetic applications(Fig. 1b). Here, data of the dual SG GWR DC029 are available since 1999 until 2002. This instrumentwas also episodically checked by absolute gravity observations. Besides the common meteorologicalparameters, the ground water table is recorded continuously. It should be pointed out that the groundwater gauge is located approximately 400 m apart from the gravimeter building. The geological for-mation in Wettzell is deeply weathered bedrock covered by a thin layer of soil. The ground water ismainly stored in cracks, which have a random distribution in size and depth. So there is no free watertable.

3. Data analysis

The SG gravity observations acquired at Medicina during the time period October 1996–November2003 were analysed. For the Wettzell station, the data records are relevant to the period June1999–December 2002. Both stations collect the data with a 10-s sampling rate. The data are then filteredto 1-min samples. The TSoft software package (Vauterin, 1998) is adopted to pre-screen the observations.This allows to remove spikes and offsets between the data blocks and to fill small gaps by means of in-terpolated data. The ETERNA, version 3.30, software package (Wenzel, 1998) is used for the analysis ofthe SG gravity data. Details concerning the data analysis procedure can be found inZerbini et al. (2001).

The AG measurements, regularly performed at both stations, allowed to remove the offsets betweendata blocks of the SG data series. Also the annual instrumental drift of 16.5 nm/s2 for SG C023 (Medic-ina) and 19.3 nm/s2 for the SG D029 (Wettzell) was estimated by these comparisons. The offsets oc-curred in conjunction with liquid helium refilling and once during an earthquake, so all mechanicalshocks.

For a better comparison between the GPS height daily solutions and the gravity residuals (tides—solidEarth and ocean, polar motion and air pressure effects have been removed from the records), the gravitydata are averaged to daily values.


The GPS analysis follows standard procedures using the Bernese Software package, version 4.2(Beutler et al., 2001) and the IGS orbits to compute daily 3D coordinates. The coordinates are deter-mined in the frame of five fiducial IGS stations (ITRF2000) in the region. Details are provided byZerbiniet al. (2001).

4. Comparison of long-term gravity and height variations

The combination of the height variation measurements provided byGPS and of the gravity potentialchanges observed by the SG allows the separation between the geometric signal from that induced by masschanges embedded in the gravity records. This is most important for understanding the signal-producingmechanisms in both the GPS and gravity time series. To study the observed seasonal oscillations, theloading effects in the case of GPS and the loading and Newtonian attraction effects in the case of gravity,due to seasonal variations in the atmospheric pressure and surficial hydrology as well as to non-tidaloceanic processes have been accounted for and modelled.

The observed signal in the GPS and gravity time series results from a superposition of various effects(seeTable 1, for example). Because only one of the parameters involved is the target of the investigation,the effects of all the others should be eliminated or, at least minimized, by means of theoretical or numericalmodels. Here, the trend in both series, height and gravity, caused by long-term crustal deformations and/ormass changes will be the focus of the study.

The main problem with observed gravity variations is to be able to discriminate in the observed changeshow much is due to mass redistribution and how much is due to actual vertical displacements. AlreadyJachens (1978)derived the bounds for the ratio of gravity to height changes corresponding to differentphysical phenomena (Fig. 2). For comparison, the ratio obtained for the body tide is added to the figure(Vanıcek and Krakiwsky, 1986).

Two stations will be investigated, one, Medicina, where seasonal signals are relevant, and a second one,the Wettzell station, where seasonality has not been detected. Both stations, however, are characterizedby a remarkable interannual variability. The seasonal signals in the GPS and gravity observation seriescan be modelled to a high degree of accuracy. The background of the models, specifically for Medicina, isgiven inZerbini et al. (2002)andRomagnoli et al. (2003). Here, only a short update concerning modellingof the non-tidal oceanic effects is provided.

Table 1Signals affecting GPS height and gravity measurements

Height Gravity

Deficiencies in the reference system Polar motionTidal deformation Tidal effects

Short and long-term environmental effects Short and long-term environmental effectsAtmospheric signal (regional, local) Atmospheric signal (regional, local)Hydrological signal (regional and local) Hydrological signal (regional and local)Ocean signal (global/regional) Ocean signal (global/regional)Thermal effect of structures (local) Thermal effect of structures (local)

Long-term crustal deformation Long-term crustal deformation


Fig. 2. Ratios of gravity to height changes for different physical phenomena (taken fromVanıcek and Krakiwsky, 1986).

5. The Medicina station

5.1. Comparison between the observed and modelled seasonal oscillations

The GPS and gravity time series used for this work include data from the beginning of the observations,which started in mid 1996 for GPS and in late 1996 for the SG, till the end of 2003. The gravity datacollected during 1996 and 1997 were not used because of a gravity anomaly occurred in mid 1997. Apossible explanation of this event is given inZerbini et al. (2002).

As regards gravity, the effect of the local atmospheric air pressure variations is removed from theSG observations in the course of the data analysis by using an empirically derived transfer functionbetween gravity and the local air pressure values of−2.94 nm s-2/(h Pa). This is a common procedure inthe treatment of the SG data. However, the use of a single transfer function may leave components in


the data, which are not accounted for by using a mean value. An approach using frequency-dependentatmospheric corrections should lead to a better understanding of seasonal variations and would improvethe modelling capability.

In order to compare observed and modelled seasonal oscillations, loading and mass effects havebeen calculated. The gravity variations due to the vertical air mass attraction effect induced byseasonal rising of warm air and sinking of cold air have been estimated by using 12-h radiosonde data acquired by the San Pietro Capofiume meteorological station nearby Medicina (Simon,2003). This effect does not include any loading component because the pressure at ground is notaffected.

The seasonal non-tidal loading and mass attraction components have been estimated (van Dam, 2003personal communication) by using the ocean bottom pressure data from the Ocean General CirculationModel used in the ECCO project (http://www.ecco-group.org). The Green’s functions fromFarrell (1972)were used to estimate the loading and the mass attraction components. The adoption of the ocean bottompressure data from the ECCO project has the advantage with respect to the previous used non-tidal oceanmodels (Zerbini et al., 2002; Romagnoli et al., 2003) that data are available twice daily on a 1◦ × 1◦ gridand this allows to model the short and very short period variations.

The combined hydrological loading and mass effect influencing the SG time series was esti-mated by correlating the simplified climatic hydrological balance, available for the Casola Caninastation in the vicinity of Medicina (Zinoni, 2003 personal communication), with the gravity recor-dings.

For the GPS height seasonal variations, the loadings due to air pressure, surficial hydrology, non-tidaloceanic effects and thermal expansion of the structure, which supports the antenna, have been considered.Details concerning the air pressure and surficial hydrological effects and the thermal expansion can befound inZerbini et al. (2002)andRomagnoli et al. (2003). As pointed out in the preceding discussionconcerning gravity, the non-tidal ocean loading has been estimated by means of the ocean bottom pressuredata from the ECCO project (van Dam, 2003 personal communication).

In this study, a different approach with respect to previously published papers has been chosen tocompare observed and modelled seasonal variations. Average observed and modelled seasonal cycleshave been computed by stacking the observations as well as the modelled data for the individual years1998–2003 for gravity and 1996–2003 for the GPS height series. The scatter of the individual years versusthe mean value is used to estimate the variability of the seasonal cycles of gravity and height (Fig. 3). Theyare displayed as error bars inFig. 3a and b. The seasonal cycle in gravity, about 50 nm s-2 peak-to-peak(Fig. 3a), is prominent in the observations and in the modelled data; its deformation component, the heightseasonal cycle, is displayed in the lower panel (b), the amplitude of the signal is about 1 cm peak-to-peak(Fig. 3b). The observed and modelled gravity seasonal cycles show an excellent agreement in amplitudewith maximum differences in the order of 10 nm s-2 (Fig. 3a, yellow line). It is recognizable, however, ashift in phase of about 55 days between the two minima during the summer period.Fig. 3a, shows thestacked observations and the seasonal model, the yellow line is the difference between the two. A cross-correlation of the gravity residuals (linear trend removed from the data) with the simplified hydrologicalbalance data shows a similar time lag. A backward shifting of 55 days of the hydrological balance contri-bution to the modelled seasonal cycle would likely bring the observed and modelled cycles to coincide andto further reduce the residuals between the observed and computed seasonal cycles. InFig. 3b the observedand modelled height seasonal cycles are presented as well as the difference between the two (yellow line).This residual has a peak-to-peak amplitude of about 2 mm, a period of about 40 days and it shows a mod-

http://www.ecco-group.org/


Fig. 3. Observed and modelled gravity and height seasonal cycles and relevant differences. The seasonal cycles were obtainedby staking of the observations and of the relevant models. Upper panel (a): Gravity seasonal cycle (observed blue, modelled red),difference between the two cycles (yellow) shifted by 40 nm s-2 for graphical purposes. Lower panel (b): GPS height seasonalcycle (observed blue, modelled red), difference between the two cycles (yellow) shifted by 4 mm for graphical purposes.


Fig. 4. Gravity observations (SG C023 and absolute gravity data FG5 101) at Medicina corrected for the seasonal effect. Theremaining linear trend (red line) is 4 nm s-2/year, converted to height changes−1.2 to−2 mm/year.

erate linear trend. It is likely that this residual difference mostly results from modelling of the air pressureand hydrological loadings. A check on the air pressure loading time series has shown that the bumpsrecognizable in the seasonal cycle and occurring in April and November are, to a large extent, attributableto the modelling of this loading effect. Additional studies are underway to better understand and model therelevant contributions of the local hydrology and of the air pressure to both gravity and GPS height vari-ations.

5.2. Long-term trends in heights and gravity

After subtracting the average modelled seasonal cycle from the gravity observations, a long-termtrend in gravity of 4± 1 nm s-2/year can be determined (Fig. 4). However, the signals in the resid-uals and all the little wiggles on them are in fact not modelled gravity signals—and since the SGnoise level is less than 0.5 nm s-2, the signal to noise ratio is more like 20:1. It is for this reason thatno error bars on the SG data are shown. The error bars are too small to be seen on the SG residualsignals.

Allowing that the gravity change is due also to mass change, it would be equivalent to−2 mm/year,assuming a conversion of 50 nm s-2/cm. Latest results from the GPS analysis providing an observedtrend in height of−2.57± 0.06 mm/year underline the findings with a completely different technique(Fig. 5).


Fig. 5. GPS height series at Medicina corrected for the seasonal effect. The overall linear trend is−2.57 mm/year (red line).

Notwithstanding the existence of natural land subsidence in the area in the order of 1 mm/year (Danteet al., 1997), a possible interpretation could be that the observed subsidence rate in the Medicina area isdue to a long-term change in the regional water regime.

6. The Wettzell station

At the Wettzell station no seasonal effect is recognizable in the observations or could be separatedby stacking. The large non-seasonal gravity variations are mainly driven by ground water variations(Fig. 6a). By fitting the SG time series as well as the absolute gravity observations to the observedwater table variations an admittance factor of 67 nm s-2/m is derived. Assuming the Bouguer plate the-ory, the admittance factor would lead to a porosity of 17%, which is possible in deeply weatheredbedrock.

By taking the ground water admittance into account, the gravity variation is reduced by half. Theremaining variations could be explained by either near field hydrology by using a closer ground watergauge, not available at the moment, or by other environmental effects not yet modelled for this station.Although, the long-term trend in the water table is representative for the area, it is especially this effectwhich reduces the trend in the gravity registration significantly. The final calculations lead to an increase ingravity of about 4± 1 nm s-2/year, which can be interpreted as a subsidence of 1.2–2 mm/year (Fig. 6b).Results from space geodetic techniques confirm this rate, e.g. the rate provided by the EUREF timeseries for Wettzell (−2.3 mm/year). The gravity to height change ratio of 1.7 nm s-2/cm would affirm theobserved deformation (subsidence).


Fig. 6. Gravity observations at Wettzell. Upper panel (a): Blue is observed gravity, light blue are water table variations with atrend equal to 18 nm s-2/year (admittance factor 67 nm s-2/m). Lower panel (b): Dark yellow is gravity corrected for ground watereffects and remaining trend in gravity of 4 nm s-2/year.

7. Conclusions

The combination of continuous SG measurements and episodic absolute measurements are the rightand only procedure to ensure a high resolution, continuous and instrumental drift free monitoring of local


gravity variations. To interpret the observed variations additional information derived from space geodesydata and environmental parameters is necessary to set up high-precision seasonal models and long-termregression functions.

Besides modelling the seasonal effects, long lasting observation series can be stacked and this workshows that the stacked mean value is close to the modelled one. Both results observed and modelled, canbe used to reduce the gravity variations for the seasonal variability. However, none of the environmentaleffects is strictly seasonal. Water table variations are affected by different factors both of climatologicaland anthropogenic nature and have most likely a long-term component, which is mapped into the gravityrecord. To interpret the long-term gravity variations, this mass and load effect shall be separated frompossible height induced gravity variations.

In both case studies the final “environmental free” trend in gravity can be confirmed by ob-served height variations. The combination of the height and gravity series and the resultinggravity–height ratio may lead to the conclusion that the observed effects are caused by groundwater.

Acknowledgements

This work has been developed under contracts MIUR 2002 from the Italian Ministry for Education,University and Research and l/R/204/02 from the Italian Space Agency. The authors are grateful to theBKG gravity team and to the staff of the Medicina Radioastronomy station.

References

Beutler, G., Bock, H., Brockmann, E., Dach, R., Fridez, P., Gurtner, W., Hugentobler, U., Ineichen, D., Johnson, J., Meindl, M.,Mervart, L., Rothacher, M., Schaer, S., Springer, T., Weber, R., 2001. In: Hugentobler, U., Schaer, S., Fridez, P. (Eds.), 2001Bernese GPS Software Version 4.2. Astronomical Institute, University of Berne, Switzerland, 515 pp.

Dante, A., Gonella, M., Teatini, P., Tomasi, L., 1997. Geographic Information System (GIS) and Data Management and RetrievalSystem (DMRS) in the CENAS Project. In: CENAS: Study on the Coastline Evolution of the Eastern Po Plain due to Sea-level Change Caused by Climate Variation and to Natural and Anthropic Subsidence. Environment Programme 1990–1994.Padova, Settembre 1997, pp. 297–321.

Farrell, W.E., 1972. Deformation of the earth by surface loads. Rev. Geophys. Space Phys. 10, 761–797.Jachens, R.C., 1978. The gravity method and interpretive techniques for detecting vertical crustal movements. In: Mueller, I.I.

(Ed.), IAG/IUGG and COSPAR, Columbus, USA, Proceedings of Ninth Geodesy/Solid Earth and Ocean Physics (GEOP)Research Conference, An International Symposium on the Application of Geodesy to Geodynamics. Department of GeodeticScience Report 280, The Ohio State University, Columbus, USA, pp. 153–155.

Romagnoli, C., Zerbini, S., Lago, L., Richter, B., Simon, D., Domenichini, F., Elmi, C., Ghiotti, M., 2003. Influence of soilconsolidation and thermal expansion effects on height and gravity variations. J. Geodyn. 35, 521–539.

Simon, D., 2003. Modelling of the Gravimetric Effects Induced by Vertical Air Mass Shifts [Mitteilungen des Bundesamtesfur Kartographie und Geodasie], Band 21. Verlag des Bundesamtes fur Kartographie und Geodasie, Frankfurt am Main,Germany, 100 pp.

Torge, W., 1989. Gravimetry. de Gruyter, Berlin, New York, 465 pp.Vanıcek, P., Krakiwsky, E., 1986. Geodesy: the Concepts. North-Holland, Amsterdam, New York, Oxford, 697 pp.Vauterin, P., 1998. TSoft: graphical and interactive software for the analysis of Earth Tide data. In: Proceedings of 13th Interna-

tional Symposium on Earth Tides, Brussels, Observatoire Royal de Belgique, Serie Geophysique, pp. 481–486.


Wenzel, H.-G., 1998. Earth tide data processing package ETERNA 3.30: the nanogal software. In: Ducarme, B., Paquet, P.(Eds.), Proceedings of the 13th International Symposium Earth Tides, 1997. Brussel, pp. 487–494.

Zerbini, S., Richter, B., Negusini, M., Romagnoli, C., Simon, D., Domenichini, F., Schwahn, W., 2001. Height and gravityvariations by continuous GPS, gravity and environmental parameter observations in the southern Po Plain, near Bologna,Italy. Earth Planet. Sci. Lett. 192 (3), 267–279.

Zerbini, S., Negusini, M., Romagnoli, C., Domenichini, F., Richter, B., Simon, D., 2002. Multi-parameter continuous observationsto detect ground deformation and to study environmental variability impacts. Global Planet. Change 34, 37–58.

Long-term crustal deformation monitored by gravity and space techniques at Medicina, Italy and Wettzell, Germany

Documents