Recent surface displacements in the Upper Rhine Graben — Preliminary results from geodetic networks

Tectonophysics 602 (2013) 300–315

Contents lists available at SciVerse ScienceDirect

Tectonophysics journal homepage: www.elsevier.com/locate/tecto

Recent surface displacements in the Upper Rhine Graben — Preliminary results from geodetic networks Thomas Fuhrmann a, Bernhard Heck a, Andreas Knöpfler a,⁎, Frédéric Masson b, Michael Mayer a, Patrice Ulrich b, Malte Westerhaus a, Karl Zippelt a a b

Geodetic Institute, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany Institut de Physique du Globe de Strasbourg, CNRS Strasbourg University, Strasbourg, France

a r t i c l e

i n f o

Article history: Received 23 February 2012 Received in revised form 5 September 2012 Accepted 3 October 2012 Available online 12 October 2012 Keywords: Upper Rhine Graben GNSS Levelling Recent crustal motions Horizontal and vertical intraplate velocity field

a b s t r a c t Datasets of the GNSS Upper Rhine Graben Network (GURN) and the national levelling networks in Germany, France and Switzerland are investigated with respect to current surface displacements in the Upper Rhine Graben (URG) area. GURN consists of about 80 permanent GNSS (Global Navigation Satellite Systems) stations. The terrestrial levelling network comprises 1st and 2nd order levelling lines that have been remeasured at intervals of roughly 25 years, starting in 1922. Compared to earlier studies national institutions and private companies made available raw data, allowing for consistent solutions for the URG region. We focussed on the southern and eastern parts of the investigation area. Our preliminary results show that the levelling and GNSS datasets are sensitive to resolve small surface displacement rates down to an order of magnitude of 0.2 mm/a and 0.4 mm/a, respectively. The observed horizontal velocity components for a test region south of Strasbourg, obtained from GNSS coordinate time series, vary around 0.5 mm/a. The results are in general agreement with interseismic strain built-up in a sinistral strike–slip regime. Since the accuracy of the GNSS derived vertical component is insufficient, data of precise levelling networks is used to determine vertical displacement rates. More than 75% of the vertical rates obtained from a kinematic adjustment of 1st order levelling lines in the eastern part of URG vary between −0.2 mm/a and +0.2 mm/a, indicating that this region behaves stable. Higher rates up to 0.5 mm/a in a limited region south of Freiburg are in general agreement with active faulting. We conclude that both networks deliver stable results that reflect real surface movements in the URG area. We note, however, that geodetically observed surface displacements generally result from a superposition of different effects, and that a separation in tectonic and non-tectonic processes needs additional information and expertise. © 2012 Elsevier B.V. All rights reserved.

1. Introduction As central and most prominent segment of the European Cenozoic rift system, the seismically and tectonically active Rhine Graben is of continuous geo-scientific interest (Lemeille et al., 1999; Ziegler, 1992). Its southern part, called Upper Rhine Graben (URG), is located in the border triangle between Germany, France and Switzerland and extends from Basel to Frankfurt. The approx. 300 km long and 40 km wide segment is bounded on the French side by the Vosges Mountains and on the German side by the Black Forest. Preceded by volcanism (late Cretaceous), the rifting was initiated during late Eocene starting with broadly E–W resp. ENE–WSW extension and lasted until Aquitanian time (Villemin and Coletta, 1990). Today, the URG is characterised by small uplift and subsidence rates less than about 0.5 mm/a and by a quasi-compressive, left-lateral strike–slip tectonic regime. The ⁎ Corresponding author at: Geodetic Institute, Karlsruhe Institute of Technology (KIT), Englerstrasse 7, D - 76131 Karlsruhe, Germany. Tel.: +49 721 6084 2303; fax: +49 721 6084 6552. E-mail address: andreas.knoepfl[email protected] (A. Knöpfler). 0040-1951/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tecto.2012.10.012

maximum horizontal stress-axis is oriented NW–SE. The URG is considered to be the seismically most active region of northwest Europe with significant probability for the occurrence of destructive earthquakes (Meghraoui et al., 2001). The evolution and neotectonics of the URG have been intensively studied by a consortium of 25 universities and governmental agencies from Germany, France, the Netherlands and Switzerland for a 10 year period lasting from 1999 to 2009. This multi-disciplinary research and training programme, called URGENT, was initiated by the geo-scientific institutions of the European Confederation of Upper Rhine Universities (EUCOR), comprising Basel, Freiburg, Strasbourg and Karlsruhe. URGENT aimed at a better understanding of the seismic hazard, neotectonics and evolution of the Upper Rhine Graben and surrounding areas, as well as the management of water resources (see http://comp1.geol.unibas.ch). Within this paper, the focus will be on the recent crustal motions detected using observations of Global Navigation Satellite Systems (GNSS, e.g. GPS) and precise levelling in order to contribute to a better understanding of the processes that affect the URG region. Geodetic measurements allow pinpointing areas of active deformation, assessing the magnitudes and rates of displacements and, thus, delivering important

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

301

Table 1 URG-related residual velocities of Tesauro et al. (2005, 2006) with respect to a Eurasian plate rotation free ITRF2000 solution. Site

Time span

(Vnorth ± σVnorth) [mm/a]

(Veast ± σVeast) [mm/a]

Publication

KARL STRA KARL STRA

05/09/2001–06/25/2003 03/22/2000–11/06/2003 05/09/2001–05/19/2004 03/22/2000–05/19/2004

0.81 ± 0.68 0.27 ± 0.56 0.7 ± 0.6 0.3 ± 0.5

0.02 ± 0.68 −0.21 ± 0.56 0.2 ± 0.6 −0.2 ± 0.5

Tesauro Tesauro Tesauro Tesauro

boundary conditions needed to verify the predictions of geomechanical models. Due to significant increasing of exploration activity in the URG (e.g. geothermal energy, raw oil, groundwater management), the need for precise geodetic information with high spatial and temporal resolution – as fundamental basis for researchers and decision makers – is increasing. Accordingly, geodetic measurements have been an integral part of the EUCOR-URGENT activities. At that time, however, the geodetic efforts were not able to resolve active deformation in the URG area unambiguously, since the available networks and datasets were not dense enough, neither in space nor in time. A major problem was that many datasets were restricted to national use only. Therefore, most studies were based on a limited amount of data that was freely available. Meanwhile, the situation has improved considerably. In this paper, we report first results from two scientific projects that focus on a comprehensive processing of unified datasets from Germany, France and Switzerland. We will show that the resolution of the transnational space geodetic and terrestrial geodetic networks in the URG area is sufficient to detect even small displacement rates well below 1 mm/a, and we will present case studies where tectonic processes might be reflected in GNSS and levelling results. The paper will close with an outlook concerning a rigorous fusion of geodetic methods planned for the future. 2. GNSS-derived horizontal velocity field of the Upper Rhine Graben area Within this section, previous investigations related to GNSS data are reviewed (Section 2.1) in order to enable a comparison with the new results gained within the GNSS Upper Rhine Graben Network GURN. The GURN initiative is described in Section 2.2. Section 2.3 is dedicated to the data evaluation strategy used to derive a preliminary horizontal velocity field (Section 2.4).

et et et et

al. al. al. al.

(2005) (2005) (2006) (2006)

expected tectonic displacements accumulated within a few years (Rozsa et al., 2005a, 2005b). The results thus remained inconclusive. Nevertheless, it was decided to continue these studies with the strongly increasing number of permanent GNSS stations in the region and the continuous improvement of the processing strategies. Besides the GNSS campaigns mentioned above, in particular two research teams dealt with velocity fields based on permanent GNSS measurements. In order to derive horizontal intraplate velocities of central Western Europe, Tesauro et al. (2005) analysed and compiled heterogeneous datasets of velocities of permanent GPS sites stemming from various network solutions and analyses (ITRF2000, EPN, 1 AGNES, 2 REGAL, 3 RGP 4). In the URG only two sites, KARL (Karlsruhe, Germany) and STRA (Strasbourg, France; identical with STJ9), were taken into account. See Table 1 for residual ITRF2000-based velocities with respect to the European Plate. The displacement rates were slightly reduced with an extended database (Tesauro et al., 2006). Nocquet and Calais (2003) contributed to the understanding of intra- and interplate deformations of Central Europe with respect to processes that drive continental deformation and control associated seismicity. Therefore, datasets of ITRF2000, EPN, RGP, and REGAL (weekly resp. daily solutions) were rigorously compiled on normal equation basis. Again, only the aforementioned two stations were taken into account. The time series cover approx. 2–3 years. Jumps within coordinate time series were handled solving for different positions and constraining velocity values to be identical for the entire time series. The derived velocity rates are shown in Table 2. Nocquet and Calais (2003) carried out detailed studies focussing on the representativeness of published as well as calculated Euler pole values. Therefore, the strategy of calculating intraplate velocity values differs compared to Tesauro et al. (2005, 2006) resulting in reduced velocities. 2.2. GNSS Upper Rhine Graben network — GURN

2.1. Previous investigations Space geodetic measurements have been used since many years to detect tectonic displacements in the area of the URG. At the beginning of EUCOR-URGENT, only the American Global Positioning System (GPS) was available. Nowadays, also the Russian system GLONASS is in full operational capability, while the European initiative Galileo as well as the Chinese GNSS Compass is currently developed. In the following, we will use the general term GNSS that comprises all available satellite positioning systems. The early geodetic GPS measurements suffered from a little number of occupied sites and a low data volume. Due to the sparsity of permanent stations, three campaign measurements were performed in 1999, 2002 and 2003, each one covering a time period of a few days. Average RMS errors of the derived coordinates were of the order of 2 mm for the horizontal and 4 mm for the vertical component, respectively, which are an order of magnitude larger than the Table 2 URG-related residual velocities of Nocquet and Calais (2003) with respect to a Eurasian plate rotation free ITRF2000 solution. Site

Time span [years]

(Vnorth ± σVnorth) [mm/a]

(Veast ± σVeast) [mm/a]

KARL STJ9

1.9 2.7

0.12 ± 0.55 −0.03 ± 0.67

0.27 ± 0.41 0.40 ± 0.53

The transnational cooperation GURN was established in September 2008 by the Institut de Physique du Globe de Strasbourg (Ecole et Observatoire des Sciences de la Terre, EOST, Strasbourg, France) and the Geodetic Institute (GIK) of Karlsruhe University (now Karlsruhe Institute of Technology, Karlsruhe, Germany). Within GURN, the two institutions cooperate in order to carry out geo-scientific research in the framework of the transnational project TOPO-WECEP (Western and Central European Platform; http://www.topo-wecep.eu/, Cloetingh et al., 2007), which is the successor of the above mentioned EUCORURGENT initiative. GURN was established as a long-term project in order to derive site velocities resp. displacement rates with respect to time series of daily estimated site coordinates based on a highly precise and highly sensitive network of permanently operating GNSS sites. When GURN started, the network consisted mostly of permanently operating sites of SAPOS ®5 Baden-Württemberg and different data providers in France (e.g. EOST, Teria, 6 RGP). In 2009, GURN was extended to the south by cooperating with swisstopo (Switzerland) 1

EPN: EUREF Permanent Network (link: http://www.epncb.oma.be) AGNES: Automated GNSS Network for Switzerland (Brockmann et al., 2011) REGAL: Réseau GPS permanent dans les Alpes, France (link: http://kreiz.unice.fr/regal) 4 RGP: Réseau Géodésique Permanent Français, France (link: http://rgp.ign.fr) 5 SAPOS®: Satellite Positioning Service of the German surveying agencies (link: http://www.sapos.de) 6 Teria: GNSS Network for France (Gaudet and Landry, 2005) 2 3

302

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

51˚00'

Sites of GURN SAPOS® Baden−Württemberg SAPOS® Rheinland−Pfalz French sites Swiss sites Other sites

50˚30'

50˚00'

49˚30'

49˚00'

48˚30'

48˚00'

47˚30'

47˚00'

km 0

50

100 46˚30'

5˚

7˚

6˚

8˚

9˚

10˚

11˚

Fig. 1. Current status of GURN; sites with two collocated monumentations (two GNSS sites at one locality, with a distance of several tenths of meters) are marked by one symbol.

and to the north by cooperating with SAPOS ® Rheinland-Pfalz. Actually, GURN consists of approx. 80 permanently operating GNSS sites, most of them equipped with individually calibrated antennas. The recent status and the spatial distribution of the network are shown in Fig. 1.

80 2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

70

Number of sites

60 50 40 30 20 10 0 Fig. 2. History of GURN concerning the number of available sites, starting in the year 2002. Blue: GPS-only tracking sites, red: GPS and GLONASS tracking sites, green: total number of sites.

The main goal of GURN is a first coordinated scientific processing and analysis of data of all permanently operating GNSS sites in the whole URG area. In addition to the determination of 3D surface displacements, the database of GURN is used for research in different fields of GNSS positioning, e.g. new data processing techniques (e.g. multipath reduction using stacking techniques (Fuhrmann et al., 2010), extending the GNSS stochastic model (Luo et al., 2011)) and GNSS-based water vapour exploitation (e.g. Alshawaf et al., 2012). Due to the heterogeneous tasks of the GURN sites, which reach from cadastral purposes for real-time positioning applications up to sophisticated research in different fields, the monumentation and distribution of the sites is varying. In the German part of GURN, most of the sites are operated by the surveying agencies of Baden-Württemberg and Rheinland-Pfalz. The average distance between the sites is 50–60 km. The locations of the sites were chosen according to logistic aspects: accessibility to power supply and communication as well as security reasons for the site equipment. Originally, these sites were established for cadastral purposes, which aim for accuracies of a few centimetres in real-time. Nowadays, the data of these sites is also dedicated to be used within geodynamic investigations (Faulhaber, 2007). SAPOS® sites are mainly installed on public buildings, which differ in structure, height and age. An ‘a priori’ exclusion of monumentation effects like thermal expansion, tilting or groundwater induced displacements is not possible, therefore the GNSS-based time series of coordinates have to be inspected carefully for these kinds of signals. At some sites, tiltmeter records and various quality-related parameters (e.g. multipath) were analysed in order to gain insight into site stability and GNSS data quality (Knöpfler et al., 2010). Only a few German sites, which are

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

303

55˚00'

Reference Frame of GURN EOST Processing Reference sites GURN sites

POTS

DELF

52˚30' JOZE

DENT

50˚00' WTZR

47˚30'

GRAZ

45˚00' TLSE

MEDI

GRAS

42˚30' MATE VILL

40˚00' CAGL

37˚30' SFER

NOT1 LAMP

35˚00'

32˚30'

km 0

350˚

355˚

0˚

5˚

10˚

15˚

200

400

20˚

30˚00' 25˚

55˚00'

Reference Frame of GURN GIK Processing Reference sites GURN sites POTS

52˚30'

WARE KLOP

50˚00' WTZR MLVL

47˚30'

GRAZ

SJDV

45˚00'

GENO

GRAS MARS

42˚30'

km 0

100

200

40˚00' 0˚

5˚

10˚

15˚

20˚

Fig. 3. Integration of GURN into the regional reference frame (here: EPN). Top: EOST realisation, bottom: GIK realisation.

304

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Fig. 4. Time series of the coordinate components for the EOST (left) and GIK (right) solutions (blue) for the baseline OFFE–RAVE (OFFE kept fix). In addition, estimated regression lines (red) and velocities for intervals >2 years are shown.

operated by other institutions, have stable monumentations like deeply grounded pillars or metal tripods. Other sources of unwanted signals are equipment changes (e.g. antenna, receiver, receiver firmware), that often induce jumps in the coordinate time series; in addition, velocity rates often change significantly after such site-specific events. Due to the commercial task of SAPOS ®, equipment changes are done without paying attention to long and undisturbed coordinate time series. In France, most of the GURN sites are operated by private companies, which offer GNSS data for real-time applications. Therefore accuracy requirements are aiming for centimetre level and most sites are installed on buildings. The data quality of these sites is similar to the SAPOS ® sites. However, in the last years EOST set up pillar sites, especially for geodynamic purposes. These sites have proper foundations and they are realised as double-walled pillars. The time series of these geodynamic sites are still rather short (2–3 years). Swisstopo, the national surveying agency of Switzerland, also uses various types of monumentation. Some of the sites are located on buildings, similar to the German SAPOS ® sites, other sites are installed using robust steel masts or pillars, which are well-founded and connected to the bedrock. At some sites, installations on buildings and the masts or pillars are collocated. On the mast and pillar sites, equipment changes are only performed, when the current components are damaged. Thus, jumps in the coordinate time series induced by hardware changes are reduced to a minimum. The main part of the GURN database starts in 2002. Almost the complete data of the sites of SAPOS® Baden-Württemberg is available since then. A major increase of available sites in GURN happened in 2004, when SAPOS® Rheinland-Pfalz started to archive their raw data. Fig. 2 shows the history of GURN concerning the number of available sites. This plot illustrates the growing number of sites as well as the successive hardware changes from GPS-only tracking sites to combined GPS and GLONASS tracking sites. Starting mainly in 2008, SAPOS ® Baden-Württemberg and SAPOS® Rheinland-Pfalz changed the antennas and receivers due to an increasing number of available GLONASS satellites; the Russian global positioning system GLONASS was extended stepwise to a full operational capability in 2011. 2.3. GNSS data processing At the present state of the project, subsets of the available GNSS data are independently processed by EOST and GIK according to their individual processing strategies. Various case studies were carried out and numerous scenarios were calculated by both cooperating institutions. This enabled the quantitative comparison of results and will guarantee

a consistent coordinate time series database. Since the expected surface displacements of tectonic origin are very small, it is crucial to assess the impact of systematic errors resulting from a specific processing schema. The current test phase shall result in refined processing strategies for the whole dataset. The test phase comprises also an investigation of various approaches to assess and to remove meteorological, hydrological and anthropogenic disturbances from the coordinate time series. At EOST, the GNSS software package GAMIT/GLOBK (Dong et al., 1998) is used, which focuses on daily network solutions and combinations to generate site velocities. The GIK working group uses the Bernese GPS Software (Dach et al., 2007). Both scientific software packages offer various processing options. In order to derive consistent ITRF2005-related results for the complete time series, specific care was taken to additional data like precise ephemeris of the GNSS satellites and Earth Rotation Parameters, which are necessary for the processing of GNSS data. At the GIK, products provided by the Centre for Orbit Determination in Europe (CODE, Berne, Switzerland) are used, that are based on reprocessing campaigns according to the latest models and strategies (Lutz et al., 2011). According to their independent processing strategies, the EOST and GIK working groups use different sites to integrate GURN into the regional reference frame. The applied network configuration of both institutions is given in Fig. 3. The detailed analysis of the effects of the reference frame realisation is currently under research. 2.4. Preliminary horizontal intraplate velocity field As a representative GURN scenario example, the results for a sub-network of selected sites in the southern part of the URG area are presented. In order to improve the determination of linear trends, especially for relatively short time series, only sites almost unaffected by seasonal signals have been taken into account. Based on the ITRF2005 (Altamimi et al., 2007) network solution and according to Nocquet and Calais (2003), residual velocities with respect to an Euler pole, estimated purely from the geodetic dataset, are calculated site-related using local topocentric coordinates (northing, easting, up). To avoid a contamination with artificial jumps due to hardware changes at the GNSS sites, the estimation of linear velocities is restricted to periods of more than two years between known antenna changes. Thus, at some sites more than one velocity value is calculable (e.g. site RAVE). For the estimation approach, a robust linear regression is used, which is non-sensitive to outliers (DuMouchel and O'Brien, 1989). Fig. 4 shows a representative time series example of a baseline (OFFE–RAVE) in order to illustrate the high data quality and

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

305

49˚00' 1 mm/year STJ9 ENTZ

EOST OFFE

TUE2 48˚30'

BFO1 SIGM 48˚00'

MAKS RAVE BLFT

47˚30'

km 0

50 47˚00'

7˚

6˚

8˚

9˚

10˚ 49˚00'

1 mm/year STJ9

EOST

ENTZ OFFE

BFO1

48˚30'

TUE2

VILL SIGM MAKS

48˚00'

FRE2

Data volume of velocities > 2 .. ≤ 4 years > 4 .. ≤ 6 years > 6 .. ≤ 8 years > 8 years

RAVE

BLFT

47˚30'

km

LUCE 0

50 47˚00'

6˚

7˚

8˚

9˚

10˚

Fig. 5. Recent GURN-based URG-related horizontal residual velocity field with respect to ITRF2005 after subtraction of the Eurasian plate rotation; top: EOST, bottom: GIK.

the conformity of the results of both institutions. Note that the EOST solution takes into account only data between 2003 and 2009 while the GIK solution makes use of GNSS data within the time span 2002–2010. In order to remove the Eurasian trend, a best-fitting Euler pole has been derived from the coordinates and the velocities of the EPN sites used as reference frame for the GIK solution. Subtraction of the respective site velocities from GIK as well as EOST solutions results in residual horizontal velocity fields that reflect (i) displacements that are not compatible with rigid plate motion, and (ii) effects emanating from the different processing strategies chosen by EOST and GIK (Fig. 5, Tables 3 and 4). Fig. 5 highlights current achievements and persistent problems that are summarized as follows: • With few exceptions, the horizontal velocity components are less than 0.5 mm/a. The formal error is well below 0.1 mm/a which constitutes a substantial improvement compared to earlier work (Nocquet and Calais, 2003; Tesauro et al., 2005, 2006).

• Replacement or change of equipment often leads to significant changes in the velocity estimates before and after the event (e.g. RAVE). This still unresolved problem clearly shows that the current accuracy of the network is substantially worse than the formal errors indicate. A bias of estimated trends due to short period disturbances may be part of the problem, especially for short time series (e.g. 2nd time segment of RAVE, compare Table 3). It is assumed that this situation will improve with time. However, it is also possible that this effect reflects some kind of instrumental noise induced by a varying adaption of electronic components which would place a technical limit to the precision of long term GNSS measurements. If we take the standard deviations of the difference between the two present solutions (σdiffnorth = 0.36 mm/a, σdiffeast = 0.38 mm/a) as a measure of the true sensitivity of the network, our test scenario shows that the lower limit for detecting horizontal movements in the URG area is still above 0.4 mm/a.

306

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Table 3 Recent GURN-based URG-related residual velocities with respect to ITRF2005 after subtraction of the Eurasian plate rotation, GIK solution. Site

Time span

(Vnorth ± σVnorth) [mm/a]

(Veast ± σVeast) [mm/a]

OFFE OFFE RAVE RAVE SIGM SIGM VILL TUE2 TUE2 FRE2 BFO1 BLFT ENTZ EOST LUCE MAKS STJ9

01/01/2002–09/17/2008 09/17/2008–11/08/2010 02/15/2002–11/20/2008 11/20/2008–04/16/2011 03/12/2002–11/18/2008 11/18/2008–04/16/2011 12/01/2002–01/23/2008 04/28/2006–08/07/2008 08/07/2008–04/16/2011 07/17/2006–04/16/2011 11/12/2006–04/16/2011 05/14/2007–09/27/2010 12/02/2004–08/06/2009 04/12/2007–04/16/2011 11/01/2008–04/16/2011 05/09/2007–04/16/2011 01/01/2002–04/16/2011

0.02 ± 0.01 −0.36 ± 0.05 0.01 ± 0.01 0.44 ± 0.04 −0.15 ± 0.01 0.50 ± 0.04 0.24 ± 0.02 0.50 ± 0.05 −0.03 ± 0.03 0.07 ± 0.03 −0.89 ± 0.02 −0.29 ± 0.03 −0.29 ± 0.01 −0.21 ± 0.03 0.25 ± 0.04 −0.45 ± 0.03 −0.15 ± 0.01

0.06 ± 0.01 0.88 ± 0.05 −0.50 ± 0.01 0.28 ± 0.04 −0.04 ± 0.01 0.52 ± 0.04 −0.40 ± 0.02 −0.22 ± 0.04 0.44 ± 0.04 0.46 ± 0.03 −0.25 ± 0.03 0.58 ± 0.02 0.26 ± 0.02 0.69 ± 0.02 0.34 ± 0.04 0.64 ± 0.03 0.05 ± 0.01

• As is usually the case with GNSS, the uncertainty of the vertical component is larger by a factor of 3 as compared to the horizontal component. It is not expected that height changes obtained by GURN are sensitive to surface displacements less than 1.5 mm/a (see Section 3). As will be discussed later, we aim to replace GNSS heights by levelling results for an investigation of recent tectonic surface movements. • Inspection of Fig. 5 suggests a systematic difference between the GIK and EOST solutions. Subtracting the EOST from the GIK velocity vectors and averaging these values v
dif f north ¼ 0:25  0:11 mm=a; v
dif f east ¼ −0:08  0:12 mm=aÞ results in a differential horizontal motion of 0.27 mm/a in northerly direction. We attribute this systematic effect mainly to the different reference frames used in both approaches. We suspect that effects induced by the choice of the reference have only a minor influence on the relative velocities within the network (i.e. strains). Further investigations will have to prove this assumption. Apart from the inconsistencies mentioned above, the preliminary GURN results in general fit well into the overall tectonic picture. The EOST velocity vectors west of the river Rhine point towards N145°E, closely matching the general trend of the maximum horizontal stress axis in the URG area (Heidbach et al., 2008; Lopes Cardozo and Behrmann, 2006). Almost no displacement is observed in the eastern part of the investigation area. The GIK solution contains the above mentioned additional N-trend of the order of 0.3 mm/a; the relative displacement of the western versus the eastern part, however, essentially resembles the EOST solution. Resolved on the general trend of the Southern Rhine Graben (N20°E), the results indicate that the segment between Strasbourg and Basel currently accommodates compression and sinistral shear of an order of magnitude of 0.5 mm/a, Table 4 Recent GURN-based URG-related residual velocities with respect to ITRF2005 after subtraction of the Eurasian plate rotation, EOST solution. Site

Time span

(Vnorth ± σVnorth) [mm/a]

(Veast ± σVeast) [mm/a]

OFFE RAVE SIGM TUE2 BFO1 BLFT ENTZ EOST MAKS STJ9

01/01/2003–09/17/2008 01/01/2003–11/20/2008 01/01/2003–11/19/2008 04/27/2006–08/07/2008 11/15/2006–12/31/2009 05/14/2007–12/31/2009 12/02/2004–08/06/2009 04/12/2007–12/31/2009 05/09/2007–12/31/2009 01/01/2003–12/31/2009

−0.21 ± 0.02 −0.12 ± 0.01 −0.27 ± 0.02 −0.01 ± 0.06 −0.40 ± 0.04 −0.40 ± 0.06 −0.39 ± 0.02 −1.00 ± 0.06 −0.43 ± 0.07 −0.40 ± 0.01

0.15 ± 0.02 −0.20 ± 0.02 0.13 ± 0.02 −0.26 ± 0.07 0.71 ± 0.04 0.39 ± 0.04 0.37 ± 0.02 0.33 ± 0.06 0.28 ± 0.07 0.12 ± 0.01

respectively. The direction as well as the magnitude of motion is in agreement with general geological concepts (Illies et al., 1981; Lopes Cardozo and Behrmann, 2006). The GNSS derived horizontal displacement rates, having the same order of magnitude as the vertical rates (Section 3), may be slightly larger than average rates during the Pleistocene (Illies, 1979; Illies et al., 1981). It has to be stated clearly that these results are preliminary and deserve further refinement. The next steps within the GURN initiative are directed to • continued studies on reference frame effects and final decision about the processing strategy; • further reduction of site-specific effects (e.g. multipath, antenna modelling); • improved preprocessing of the coordinate time series (e.g. removal of steps, jumps and outliers due to equipment changes or instrumental malfunctions); • inclusion of GLONASS data (until now, only GPS has been used in order to avoid apparent effects on velocity caused by artefacts from combination of GPS and GLONASS (King and Watson, 2010)); • determination of a unified horizontal velocity field for the whole GURN region; • determination of strain field for the whole GURN region; • expansion and densification of the GURN network by implementation of additional GNSS datasets (e.g. AXIO-NET).

3. Levelling This section presents the strategy of the determination of vertical displacements using the precise levelling technique. The fundamental principle of levelling is simple and has not changed since its beginning. The basic observations of levelling are height differences between fixed benchmarks. Starting at a benchmark with a well-known absolute height (e.g. tide gauge), heights of other benchmarks can be estimated by consecutively adding the measured height differences to the height of the starting point. In this way, geodetic levelling networks are built using closed loops of levelling measurements for internal checking of the observations. These networks were constructed about 100 years ago by the national survey authorities and have been remeasured several times. This allows the calculation of vertical displacements. Section 3.1 focusses on previous investigations, while Section 3.2 describes the database of the current research. In Section 3.3, the data evaluation strategy is presented in order to derive a preliminary vertical velocity field including a comparison to GNSS-derived velocities (Section 3.4). 3.1. Previous investigations Mälzer (1967) and Mälzer et al. (1983, 1988) published maps of height changes in the URG and the Rhenish Massif based on repeated precise levelling data on 1st order lines of the German surveying authorities and data from the national agency for nuclear waste deposit in Switzerland (Nationale Genossenschaft für die Lagerung radioaktiver Abfälle (Nagra), link: www.nagra.ch). They found several indications for active tectonic deformation, but presented also examples for man-induced subsidence due to extraction of petroleum and groundwater. Demoulin et al. (1998) carried out a local study of fault movements in the southern Black Forest and concluded that the faults in this region are creeping aseismically at rates varying between 0.2 mm/a and 1.1 mm/a. Zippelt and Dierks (2007) continued the work of Mälzer et al. (1988). They analysed data of up to five levelling campaigns between 1922 and 2004 in southern Baden-Württemberg and northern Switzerland, including also 2nd order lines which are shown to contribute useful information about vertical movements in the interior of the wide 1st order loops.

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

307

Fig. 6. Levelling networks of French and German surveying agencies close to the URG region.

3.2. Levelling database Repeated levelling data from the URG and its surroundings dating back until 1922 are currently analysed at GIK within a scientific research project. The archived data has been provided by the Institut National de l'Information Géographique et Forestière (IGN) and the surveying agencies from Baden-Württemberg, Rheinland-Pfalz and Saarland. Fig. 6 shows the levelling networks of France and Germany in the vicinity of the URG which are considered within this project. The networks are divided in different orders, the 1st order measurements being more precise than 2nd or 3rd order measurements. The French part consists of one large 1st order loop building a frame for the 2nd order measurements. The German 1st order measurements (east of the river Rhine) build a denser network. Also 2nd order measurements are available for this region, which are going to be analysed in future work. In the following we will present results for the German part of the 1st order levelling network, which has already been analysed. The analysis of the levelling lines west of the river Rhine is still in progress. Within any closed loop of the levelling lines, a loop misclosure can be calculated, which should be zero neglecting measurement errors and the influence of gravity. As geodetic levelling is path-dependent, there is always a theoretical loop error caused by non-parallelism of equipotential surfaces of the gravity field. This error depends on the anomalous variations of the gravity field and is larger in mountainous areas and negligible in flat areas. Fig. 7 presents the loops of the Baden-Württemberg levelling network including lengths and theoretical loop misclosures. To correct the raw measurements for the influence of gravity, a normal height correction is applied using point-wise gravity measurements (Heck, 2003; Hofmann-Wellenhof and Moritz, 2006). The levelling lines shown in Fig. 7 were measured at least four times.

The measurement epochs of every loop and the residual loop misclosures – after subtracting the theoretical loop misclosure from the total loop misclosure – are presented in Table 5. All residual loop errors in all epochs are smaller than 30 mm. As the distance between adjacent benchmarks is normally less than 1 km and the loop lengths are 185–334 km, these levellings show a high accuracy, because the loop errors are composed of the measuring errors of all measured height differences along a loop. The computation of distance-dependent standard deviations of the presented loops from the loop misclosures yields the average values for every epoch displayed in Table 6. Regarding the standard deviations, the levelling data of these four measurement epochs have a similar precision. 3.3. Calculation of vertical displacements The calculation of vertical displacements from levelling data can be subdivided into two principle strategies (see Fig. 8). If the period of the measurements within a levelling network is short with respect to the vertical motions in the area, all levelled height differences can be attributed to a specific epoch, and a static adjustment of the network at any epoch can be performed. Fixing a vertical datum valid for all epochs yields “absolute” heights of any benchmark at various epochs. Afterwards the adjusted heights of each remeasured benchmark are compared which yields the vertical displacement between two or more epochs (D'Anastasio et al., 2006). We will use this approach for a discussion of local height changes where we assume that displacements within one epoch are zero (Section 3.4.2). This assumption does not hold for large levelling networks, where even between the first and the last measurement of one loop, time differences of about 15 years are observed (see Table 5). In addition,

308

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Fig. 7. Analysed levelling loops east of the river Rhine, with loop number, length and theoretical loop misclosure.

a connection of the French and German sub-networks is not possible using static adjustment, due to large temporal differences. To account for different measurement times a kinematic adjustment approach was developed which takes the dates of the levellings into account (Ghitau, 1970; Hohldahl, 1975; Mälzer et al., 1983). The basic equation of kinematic adjustment is given by

i

0

ti

H j ¼ H j þ ∫ α j dt t0

which means that the height Hji of a benchmark j at the observation epoch ti is modelled using its height Hj0 with respect to a reference

time t0. The integral of the benchmark movement αj can be approximated using a Taylor expansion which yields i

0

Hj ≈ Hj þ

1 0 1 dvj 2 v Δt þ Δt þ … : 1! j i 2! dt t0 i

The first term represents a linear motion (velocity vj0), the second an accelerated motion with the period Δti, and so on. Δti is calculated using Δt i ¼ t i −t 0 : In a first analysis step, only linear motions are assumed in the adjustment, afterwards terms of higher order are added at benchmarks where a linear model is not appropriate (Zippelt, 1988).

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315 Table 5 Measurement epochs and residual loop misclosures for all analysed levelling loops east of the river Rhine. No.

Epoch

608

1922–1935 1959–1961 1982–1984 2004–2011

−0.7 3.1 23.2 −2.8

mm mm mm mm

609

1924–1939 1952–1968 1980–1987 2005–2010

10.9 20.0 3.6 3.5

mm mm mm mm

614

1924–1939 1952–1969 1981–1985 2007–2010

10.2 4.3 −4.6 −3.0

mm mm mm mm

610

1924–1939 1956–1960 1982–1987 2005–2008

−6.4 −4.8 11.4 −9.9

mm mm mm mm

1937–1939 1957–1969 1981–1986 2005–2008

8.1 11.4 −11.2 −9.9

mm mm mm mm

1923–1939 1959–1962 1982–1984 2004–2009

−3.7 −14.7 1.3 24.0

mm mm mm mm

1923–1938 1959–1969 1981–1987 2006–2010

17.6 10.9 6.0 −25.9

mm mm mm mm

601

1922–1935 1959–1961 1982–1984 2004–2011

−4.8 6.5 0.7 12.1

mm mm mm mm

602

1922–1935 1959–1962 1982–1984 2004–2011

−2.8 11.1 9.5 4.3

mm mm mm mm

603

1938–1941 1962–1967 1981–1982 2006–2010

5.2 3.5 2.6 −8.2

mm mm mm mm

613

611

⁎ 612

⁎

Misclosure

309

Epoch 1

Epoch 2

...

Epoch i

Adjustment 1

Adjustment 2

...

Adjustment i

Vertical displacement

Epoch 1

Epoch 2

...

Epoch i

Adjustment (t)

Vertical displacement Fig. 8. Adjustment strategies for the determination of vertical displacements; top: static adjustment, bottom: kinematic adjustment.

a single stable reference point should be used in order to derive geodynamically easily interpretable results. As a consequence, the following assumptions are valid for the reference point: • no (or well-known) displacement; • measured in as many epochs as possible; • located in the centre of the network. As ‘a priori’ displacements are often unknown, additional information (e.g. from geology) is needed for the selection of stable sites. The third requirement is reasonable, if a comparable precision for all adjusted parameters is desired, because the adjusted precision decreases with the distance from the reference point. For the analysis of the URG network (Fig. 7) a benchmark at a church in Freudenstadt, a stable region of the black forest, is chosen. This benchmark is a nodal point situated in the centre of the network and was measured in each campaign (see Fig. 9).

⁎ Large loop misclosure caused by measurement errors in a part of the line bordering the loops 611 and 612. The erroneous part will be remeasured in 2012.

The data is adjusted using a Gauss–Markov model, wherein the meai sured height differences are introduced as observables (hj,k = Hki − Hji) and used to determine heights and displacements. In the context of the stochastic model it is assumed that there are no correlations between the measurements, and the precision is mainly depending on the distance between the corresponding benchmarks. Since the normal equation system has a rank deficiency, the datum of the network adjustment has to be fixed. Several methods of datum definition are discussed in Zippelt (1988), concluding that

Table 6 Average standard deviations (loop misclosure/√length) for the four measurement epochs. Epoch no.

Standard deviation

1 2 3 4

0.44 0.56 0.49 0.63

⁎ Without loops 611 and 612.

mm/√km mm/√km mm/√km mm/√km (⁎0.44 mm/√km)

Fig. 9. Levelling lines at the reference point with respect to geology (source: OneGeology, 2011). The reference point was observed four times along line 632 (in 1933, 1956, 1987 and 2007), five times along line 631 (in 1938, 1957, 1968, 1986 and 2007) and five times along line 635 (1933, 1960, 1968, 1987 and 2007).

310

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Table 7 Global standard deviations of kinematic adjustment and standard deviations of adjusted parameters. Standard deviation a priori (σ0) Standard deviation a posteriori (σ0) Mean standard deviation of heights Mean standard deviation of linear coefficients

1.00 0.71 6.44 0.23

mm mm mm (min: 0.71, max: 8.82) mm/a (min: 0.04, max: 0.38)

3.4. Preliminary results This section is dedicated to the results of the adjustment of the levelling data. Sections 3.4.1 and 3.4.2 focus on the kinematic and the static approaches. A preliminary comparison between levellingand GNSS-based results is presented in Section 3.4.3

3.4.1. Kinematic adjustment Analysing the levelling network (Fig. 7), the kinematic adjustment approach was applied using a total number of 10,361 observations at 3967 benchmarks. The unknowns consist of 3966 heights, 3966 linear coefficients and 32 accelerated coefficients, which means that a statistical testing refused a linear model for these 32 benchmarks. Some statistical numbers describing the precision of the adjustment are displayed in Table 7. The standard deviations of the adjusted parameters are small close to the reference point and increase with distance from the reference point (see Fig. 10). The values of the estimated linear coefficients are in the range of − 21.6 mm/a and + 10.2 mm/a with a mean value of − 0.1 mm/a. There is no significant correlation between estimated heights and linear coefficients. Outliers in the linear coefficients are removed by statistical testing of the linear coefficient with respect to the coefficients

Fig. 10. Standard deviation of linear coefficients (velocities) resulting from kinematic adjustments.

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

in the neighbourhood (Zippelt and Dierks, 2007). In total 234 linear coefficients behave significantly different compared to the linear coefficients of the surrounding region. In most cases the outliers are linear coefficients with large absolute values (large subsidence or large uplift), e.g. caused by a subsidence of the support of the benchmark with respect to the ground. As the individual movement of the corresponding benchmarks is not relevant to our study, these linear coefficients are excluded from the dataset. Based on the remaining 3732 linear coefficients a displacement map was generated (Fig. 11). The linear displacement rates derived from kinematic adjustment are small. This corresponds to expected geological rates for the URG area. More than 89% of the vertical displacement rates shown in Fig. 11 are in the range of −0.3 to 0.3 mm/a, more than 75% in the range of −0.2 to 0.2 mm/a. These small rates are hardly significant,

311

particularly for benchmarks with a larger distance from the reference point (see Fig. 10). We conclude that the major part of the investigated area behaves stable. In contrast, two regions show a significant subsidence pattern: the wider surroundings of Stuttgart and an area southwest of Freiburg. The first one is supposed to be caused by geochemical processes in the subsurface (Gipskeuper) and studied in detail in cooperation with the Department of Hydrogeology at KIT. The latter one is particularly interesting for this paper because of its location in the URG, and is investigated in detail within the next sub-section. 3.4.2. Static analysis of the region close to Freiburg The 1st order levelling lines and the geological setting close to Freiburg are shown in Fig. 12. For a detailed analysis of this limited

Fig. 11. Vertical displacement rates from kinematic adjustment of repeated precise levellings. Values for mean displacement rates and mean standard deviations are approximative.

312

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Fig. 12. Levelling lines and selected benchmarks close to Freiburg with respect to the geological setting (source: OneGeology, 2011), MBF: Main Border Fault; RRF: Rhine River Fault; WSF: Weinstetten Fault; SBB: Schönberg Block. Geological layers which are not explained in this figure are explained in Fig. 9.

region the static approach is used which is based on raw levelling measurements only corrected for the gravity effect. Starting at point A (right-hand side, Fig. 12) heights are calculated for the benchmarks along levelling lines 603 and 601 and along levelling lines 603 and 639, respectively. Different measurement dates within one epoch are neglected. Due to its small displacement rate resulting from kinematic adjustment (0.06 mm/a) and its geological situation, point A is considered to be stable. In addition, point A was measured in all four epochs and is therefore chosen as a reference point for the static analysis. Between benchmarks B–C–D–Y, lines 601 and 639 cross several fault strands. Cumulative height changes between the first epoch and the later epochs are shown in Fig. 13. Between benchmark B, located in front of the Main Border Fault (MBF) near Freiburg, and benchmark C, located close to the western border fault of the Schönberg Block (SBB), a significant subsidence is clearly visible during all measurement epochs. Following line 601 towards south, a strong anomaly caused by mining activities is observed close to benchmark E. From 1928 to 1973 potash was excavated close to the village Buggingen located in the URG (Röhr, 2012). The mining activities lead to a subsidence of more than 6 cm between the first and the second measurement epoch at point E. Between the second and the third measurement epochs the subsidence accumulates to about 10 cm. Finally, between the last two measurement epochs, and thus after potash mining was stopped, only a small additional subsidence is visible (Fig. 13 top). This example shows that anthropogenic influences may significantly superpose potentially existing tectonic signals in the URG, which are supposed to be less than 1 mm/a

(Illies, 1979) in vertical direction. However, it is hard to believe that local mining effects are responsible also for the negative height changes near Freiburg (benchmarks B–C) which is located at a distance of more than 20 km. Following levelling line 639 (Fig. 13 bottom) a sudden negative height change is visible at benchmark Y, which is close to the crossing between the Weinstetten Fault (WSF) and the important Rhine River Fault (RRF). Towards benchmark Z the displacements are decreasing ending up close to zero. A mining effect, having its maximum between the first two campaigns (Fig. 13 top) is not obvious along these sections. Fig. 14 shows a comparison of the last two campaigns, where a larger number of revisited benchmarks allows for a more detailed study. It turns out that the already mentioned significant height change between benchmarks B and C coincides with the location of the MBF (see also Fig. 12). The subsidence accumulates to 1.0–1.5 cm in about 22 years and prevails until benchmark Y. Close to this benchmark, the levelling line crosses the RRF which is marked by sudden offset of about −0.7 cm. Afterwards, the displacements are decreasing, reaching the zero level within three kilometers. There are some arguments that the levelling results on lines 603 and 639 reflect ongoing tectonic processes. The subsiding area is situated in or at least near to a releasing bend regime caused by a ~ 20° trend change of the URG from N to NNE (Buchmann and Connolly, 2007). According to Behrmann et al. (2003), the levelling line crosses actively moving fault segments between B and C (MBF) and near Y (RRF). Fault activity is deduced from recent earthquake clusters whose hypocenters fit well to the downdip projections of the faults. Taking into account that the fault planes near the eastern

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Height changes [m]

0.02 0

B

A

C

D F

−0.02 −0.04

E

−0.06 −0.08

1959/11 minus 1927/08 1983/09 minus 1927/08 2004/09 minus 1927/08

−0.1 0

10

20

30

40

50

60

70

Levelling track [km]

Height changes [m]

0.02 0

B

A

D

−0.02 −0.04 −0.06 1959/10 minus 1929/07 1983/08 minus 1929/07 2005/04 minus 1929/07

−0.1 0

10

20

30

40

50

60

Levelling track [km] Fig. 13. Height changes between the first epoch and later epochs along levelling lines 603 and 601 (top) and along levelling lines 603 and 639 (bottom). Point A is kept fixed; dates correspond to the average of observation times.

Height changes [m]

border of the URG dip towards west or southwest, the observed subsidence is compatible with normal faulting or oblique strike slip in a transtensional regime. This would fit to the conclusions of Lopes Cardozo and Behrmann (2006) who obtained a predominating normal faulting mechanism along the RRF south of 48°N from a compilation of fault plane/striation measurements. Thus, the levelling results may be preliminarily interpreted as being indicative for active faulting south of Freiburg. Although the temporal resolution of the levelling campaigns is rather coarse, the sequence shown in Fig. 13 bottom suggests that the observed subsidence is not due to episodic slip but might reflect continuous deformation. However, the rates seem to be largest between the 2nd and 3rd epochs (1959–1983), indicating that the deformation process is either non-linear in time, or that tectonic and non-tectonic processes are superimposed. Another important question is whether the inferred slip occurs on creeping or locked faults. For a creeping fault segment, one would expect sudden displacement offsets between adjacent benchmarks, as for example near Y. The gentle slopes of the displacement profile between km 30–35 and km 47–50 (Fig. 14), however, are more compatible with a fault model that is effectively locked in the upper part of the crust and stably sliding below. This conclusion is in contrast to the findings of

0.02 0

MBF

A

B

RRF

D

C

Z

Y

−0.02 2005/04 minus 1983/05

−0.04 0

10

20

30

Demoulin et al. (1998) who postulated that the faults adjacent to the south are creeping aseismically. The discussion, although not fully conclusive so far, demonstrates that the archived levelling data of the surveying agencies is able to deliver detailed information about fault movement and tectonic deformation in the URG area. However, a potential interference with non-tectonic processes always has to be kept in mind. Long lasting hydrological trends, changing the pore pressure, are likely candidates for vertical displacements at the surface. Provided that the faults create impermeable barriers for groundwater flow or pore pressure diffusion, a regional subsidence phenomenon delineated by faults like in Fig. 14 is conceivable. Additional information is necessary for a separation of effects and a proper interpretation of levelling data. Irrespective of the nature of a process under study, levelling delivers important boundary conditions for numerical models.

Z

C

Y

−0.08

313

40

50

60

Levelling track [km] Fig. 14. Height changes between the last two epochs along levelling lines 603 and 639 including the location of MBF and RRF, point A is kept fixed; dates correspond to the average of observation times.

3.4.3. Comparison of kinematic levelling results with GNSS results A comparison between the available results of the two geodetic techniques GNSS and levelling is possible so far only for a small test area in the south of Baden-Württemberg, where both displacement fields overlap. In Fig. 15 the levelling results are plotted together with the vertical displacement components derived from GNSS. At the present stage only a qualitative comparison between the two measurement techniques is possible, because the GNSS sites are not included in the levelling lines. Also the time span for the computation of displacement rates differs strongly (more than 80 years vs. less than 10 years). Inspection of Fig. 15 immediately shows that the two displacement fields do not have much in common. With exception of site 0397, the GNSS derived results are unrealistically large. We attribute this to the fact that the GNSS vertical component generally is less well-determined than the horizontal component. Due to this reason we strive for a rigorous combination of ground-based and space-based measurement techniques to derive the 3D velocity field with comparable reliability for all components (Heck et al., 2010). 4. Conclusions and Outlook We presented first results of a comprehensive analysis of crossborder geodetic datasets in the URG and surrounding regions. In contrast to earlier studies, we have access to unprocessed raw data contributed by different institutions and data providers from Germany, France and Switzerland. This opens the possibility to derive consistent solutions of recent crustal motions with improved accuracy and reliability. The data archives date back to 2002 for permanent GNSS data and to 1922 in case of repeated national levellings. At present, only a minor part of the available datasets has been processed, including analyses of vertical displacements on 1st order levelling lines in Baden-Württemberg and horizontal displacements within a small GNSS network covering the URG between Strasbourg and Basel. Results of the test scenarios, aiming at an improvement of processing strategies and analysis methods, are encouraging. It was shown that the levelling dataset is sensitive enough to resolve even the small tectonically induced vertical displacement rates expected for the URG area, while the current GNSS results place a new lower limit for the detection of horizontal displacement rates. A kinematic adjustment of repeated precise levellings reveals stable conditions for the major part of the investigated area with vertical displacement rates ranging between +/− 0.2 mm/a. Only two regions south of Freiburg and in the wider surroundings of Stuttgart show increased subsidence rates of 0.5 mm/a which are significant in terms of the formal adjustment error of 0.2 mm/a. While the former might reflect slip on seismically active faults near Freiburg (although locally augmented by mining activities), we do not attribute the wide subsidence bowl around Stuttgart to tectonic processes. Rather, we assume hydrologic changes or geochemical processes like subsurface erosion.

314

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Fig. 15. Comparison of levelling and GPS results of vertical displacements in the south of Baden-Württemberg.

Horizontal displacement rates obtained from a GURN sub-network south of Strasbourg vary around 0.5 mm/a. The formal error of rate estimation is well below 0.1 mm/a. Taking into account the actual differences of the EOST and GIK processing strategies as well as impacts of seasonal signals and hardware changes, we consider 0.4 mm/a to be a more realistic error estimate. Thus, levelling and GNSS results are comparable in magnitude of the observed displacements while the current uncertainty of the GNSS results is twice as high. Concerning GNSS this is a substantial improvement with respect to earlier work. Neglecting issues of the reference frames that are still to be solved, the GNSS-derived horizontal velocity vectors are aligned with the maximum horizontal compressive stress axis according to the world stress map, indicating that the observations are generally compatible with interseismic strain built-up. Provided a further refinement of the processing strategy, GURN might be able to deliver a consistent picture of horizontal tectonic movements. Further work will be done at GIK towards a rigorous combination of velocity fields derived from the different methods. Problems to be solved comprise: • Reliable interpolation approaches: The permanent GNSS stations are generally not included within the levelling lines. It is planned to use SAR interferometry (InSAR) in order to increase the density of observed points and to support the interpolation. • Different time periods: Levelling data cover a time period of 90 years compared to only 10 years related to GNSS. InSAR, dating back until 1992, will enhance the time period covered by space-borne geodetic methods. • Different height reference systems: Levelling delivers physical heights,

while GNSS delivers geometric heights. The difference may not be important considering height changes, but should be kept in mind. All these efforts are directed towards an inventory of current surface displacements in the trinational Upper Rhine Graben area with unprecedented detail and reliability. A regional map of surface motions will contribute to an improved understanding of intraplate deformation processes and will deliver important boundary conditions for numerical geomechanical models. It should be mentioned, however, that geodetically observed surface displacements will always reflect a mixture of processes acting on different spatial and temporal scales like mining, groundwater withdrawal, natural hydrologic changes, glacial isostatic adjustment, regional geochemical processes, tectonics and others. A proper separation of effects and an isolation of tectonic signals will need additional information and expertise. Acknowledgement We thank all our data providers for supplying GNSS and levelling data: RENAG (France), RGP (France), Teria (France), Orpheon (France), SAPOS® Baden-Württemberg (Germany), SAPOS® Rheinland-Pfalz (Germany), swisstopo (Switzerland), European Permanent Network, IGS, IGN (France), Landesamt für Geoinformation und Landmanagement (Baden-Württemberg, Germany), Landesamt für Vermessung und Geobasisinformation (Rheinland-Pfalz, Germany) and Landesamtes für Kataster-, Vermessungs- und Kartenwesen (Saarland, Germany). The levelling-related research described within this paper was funded through the German Science Foundation (DFG) under the grant HE 1433/17-1 (Vertical crustal motions in the Upper Rhine

T. Fuhrmann et al. / Tectonophysics 602 (2013) 300–315

Graben area from the analysis of repeated precise levelling data and SAR interferometry).

References Alshawaf, F., Fuhrmann, T., Heck, B., Hinz, S., Knöpfler, A., Luo, X., Mayer, M., Schenk, A., Thiele, A., Westerhaus, M., 2012. Fusion of InSAR Data and GNSS Measurements for the Determination of Atmospheric Water Vapour – A Basic concept. Proceedings, Earth Observation of Global Change, Springer, in print. Altamimi, Z., Collilieux, X., Legrand, J., Garayt, B., Boucher, C., 2007. ITRF2005: a new release of the International Terrestrial Reference Frame based on time series of station positions and Earth Orientation Parameters. Journal of Geophysical Research 112, B09401. Behrmann, J.H., Hermann, O., Horstmann, M., Tanner, D.C., Guillaume, B., 2003. Anatomy and kinematics of oblique continental rifting revealed: a three-dimensional case study of the southeast Upper Rhine graben (Germany). AAPG Bulletin 87, 1105–1121. Brockmann, E., Ineichen, D., Kistler, M., Marti, U., Schaer, S., Schlatter, A., Vogel, B., Wiget, A., Wild, U., 2011. 20 years of maintaining the Swiss Terrestrial Reference Frame CHTRF. In: Brockmann, E., et al. (Ed.), National Report of Switzerland 2011, swisstopo Report 11–11: Geodetic activities at swisstopo presented to the EUREF2011 Symposium. Buchmann, T.J., Connolly, P.T., 2007. Contemporary kinematics of the Upper Rhine Graben: a 3D finite element approach. Global and Planetary Change 58, 287–309. Cloetingh, S.A.P.L., Ziegler, P.A., Bogaard, P.J.F., Andriessen, P.A.M., Artemieva, I.M., Bada, G., van Balen, R.T., Beekman, F., Ben-Avraham, Z., Brun, J.-P., et al., 2007. TOPO-EUROPE: the geoscience of coupled deep Earth-surface processes. Global and Planetary Change 58 (1–4), 1–118. D'Anastasio, E., De Martini, P.M., Selvaggi, G., Pantosti, D., Marchioni, A., Maseroli, R., 2006. Short-term vertical velocity field in the Apennines (Italy) revealed by geodetic levelling data. Tectonophysics 418, 219–234. Dach, R., Hugentobler, U., Fridez, P., Meindl, M., 2007. Bernese GPS Software Version 5.0. User Manual of the Bernese GPS Software Version 5.0. Demoulin, A., Launoy, T., Zippelt, K., 1998. Recent crustal movements in the southern Black Forest (western Germany). Geologische Rundschau 87, 43–52. Dong, D., Herring, T.A., King, R.W., 1998. Estimating regional deformation from a combination of space and terrestrial geodetic data. Journal of Geodesy 72, 200–214. DuMouchel, W.H., O'Brien, F.L., 1989. Integrating a robust option into a multiple regression computing environment. Computer Science and Statistics: Proceedings of the 21st Symposium on the Interface. American Statistical Association, Alexandria, VA. Faulhaber, U., 2007. Geodätischer Raumbezug und Festpunktfelder in Baden-Württemberg. DVW Baden-Württemberg Mitteilungen, Heft 2, pp. 63–103. Fuhrmann, T., Knöpfler, A., Luo, X., Mayer, M., Heck, B., 2010. Zur GNSS-basierten Bestimmung des atmosphärischen Wasserdampfgehalts mittels Precise Point Positioning. KIT Scientific Reports 7561 / Schriftenreihe des Studiengangs Geodäsie und Geoinformatik 2010. KIT Scientific Publishing. ISBN: 978-386644-539-0, p. 2. Gaudet, A., Landry, J.C., 2005. TERIA: The GNSS Network for France. Proceedings, From Pharaohs to Geoinformatics, FIG Working Week 2005 and GSDI-8, April 16th-21th 2005, Cairo, Egypt. http://www.Fig.net/pub/cairo/papers/ts_08/ts08_03_gaudet_landry.pdf. Ghitau, D., 1970. Modellbildung und Rechenpraxis bei der nivellitischen Bestimmung säkularer Landhebungen. PhD Thesis, University of Bonn, 141 pp. Heck, B., 2003. Rechenverfahren und Auswertemodelle der Landesvermessung. Wichmann-Verlag, Heidelberg. Heck, B., Mayer, M., Westerhaus, M., Zippelt, K., 2010. Karlsruhe integrated displacement analysis approach — towards a rigorous combination of different geodetic methods. Proceedings of FIG Congress 2010, Sydney (13 pp.). Heidbach, O., Tingay, M., Barth, A., Reinecker, J., Kurfeß, D., Müller, B., 2008. The World Stress Map Database Release 2008. http://dx.doi.org/10.1594/GFZ.WSM.Rel2008. Hofmann-Wellenhof, B., Moritz, H., 2006. Physical Geodesy, 2nd ed. Springer, Vienna. 403 pp. Hohldahl, S.R., 1975. Models and strategies for computing vertical crustal movements in the United States. presented at International Symposium on Recent Crustal Movements, Grenoble. Illies, J.H., 1979. Rhinegraben: shear controlled vertical motions of the graben floor. Allgemeine Vermessungs-Nachrichten 86, 364–367.

315

Illies, J.H., Baumann, H., Hoffers, B., 1981. Stress pattern and strain release in the Alpine foreland. In: Vyskočil, P., Green, R., Mälzer, H. (Eds.), Recent Crustal Movements, 1979: Tectonophysics, 71, pp. 157–172. King, M.A., Watson, C.S., 2010. Long GPS coordinate time series: multipath and geometry effects. Journal of Geophysical Research 115 (B04403), 1–23. Knöpfler, A., Mayer, M., Heck, B., Masson, F., Ulrich, P., 2010. GURN (GNSS Upper Rhine Graben Network) — status and first results. FIG Congress 2010 — Facing the Challenges — Building the Capacity, 11.–16. April 2010, Sydney, Australien. online. http://www.Fig. net/pub/fig2010/papers/ts03c%5Cts03c_knoepfler_masson_et_al_4380.pdf. Lemeille, F., Cushing, M.E., Cotton, F., Grellet, B., Menillet, F., Audru, J.C., Renardy, F., Flehoc, C., 1999. Evidence for Middle to Late Pleistocene faulting within the northern Upper Rhine Graben (Alsace Plain, France). Comptes Rendus de l'Academie des Sciences Series IIA Earth and Planetary Science 328 (12), 839–846. Lopes Cardozo, G.G.O., Behrmann, J.H., 2006. Kinematic analysis of the Upper Rhine Graben boundary fault system. Journal of Structural Geology 28, 1028–1039. Luo, X., Mayer, M., Heck, B., 2011. Verification of ARMA identification for modelling temporal correlations of GNSS observations using the ARMASA toolbox. Studia Geophysica et Geodaetica 55 (3), 537–556. Lutz, S., Steigenberger, P., Dach, R., Schaer, S., Meindl, M., Ostini, L., Sosnica, K., 2011. Reprocessing Activities at CODE. Swiss National Report on the Geodetic Activities in the years 2007–2011, Section 1 (Reference Frames), Swiss Geodetic Commission. In: Wiget, A. (Ed.), XXV General Assembly of the IUGG, Melbourne, Australia, 28 June–7 July 2011, 5–6. online. http://www.sgc.ethz.ch/download/IUGG_Landesbericht_20072011_Gesamt_Vollversion.pdf. Mälzer, H., 1967. Untersuchungen von Präzisionsnivellements im Oberrheingraben von Rastatt bis Basel im Hinblick auf relative Erdkrustenbewegungen. DGKL Rh. B, Heft Nr. 138. Mälzer, H., Hein, G., Zippelt, K., 1983. Height changes in the Rhenish Massif: determination and analysis. In: Fuchs, K., von Gehlen, K., Mälzer, H., Murawski, H., Semmel, A. (Eds.), Plateau Uplift: The Rhenish Shield — A Case History. Springer, Berlin Heidelberg, pp. 164–176. Mälzer, H., Rösch, H., Misselwitz, I., Ebert, M., Moosmann, D., 1988. Höhenänderungen in der Nordschweiz und im Südschwarzwald bis zum Bodensee. Nagra Techn. Ber. NTB 88–05. Nagra, Wettingen. Meghraoui, M., Delouis, B., Ferry, M., Giardini, D., Huggenberger, P., Spotke, I., Granet, M., 2001. Active normal faulting in the Upper Rine Graben and paleoseismic identification of the 1356 Basel earthquake. Science 293, 2070–2073. Nocquet, J.M., Calais, E., 2003. Crustal velocity field of western Europe from permanent GPS array solutions, 1996–2001. Journal of Geophysical Research 154, 72–88. OneGeology, 2011. Web page hosted by British Geological Survey. www.onegeology. org last checked: 29.12.2011. Röhr, C., 2012. Web page on the Upper Rhine Graben hosted by Dr. Röhr, Friedberg. http:// www.oberrheingraben.de/Grabenfuellung/Buggingen.htm last checked, 11.1.2012. Rozsa, S., Mayer, M., Westerhaus, M., Seitz, K., Heck, B., 2005a. Towards the determination of displacements in the Upper Rhine Graben area using GPS measurements and precise antenna modelling. Quaternary Science Reviews 24, 425–438. Rozsa, S., Heck, B., Mayer, M., Seitz, K., Westerhaus, M., Zippelt, K., 2005b. Determination of displacements in the Upper Rhine Graben Area from GPS and leveling data. International Journal of Earth Sciences 94, 538–549. Tesauro, M., Hollenstein, C., Egli, R., Geiger, A., Kahle, H.-G., 2005. Continuous GPS and broad-scale deformation across the Rhine Graben and the Alps. International Journal of Earth Sciences 94 (4), 525–537. Tesauro, M., Hollenstein, C., Egli, R., Geiger, A., Kahle, H.-G., 2006. Analysis of central western Europe deformation using GPS and seismic data. Journal of Geodynamics 42 (4–5), 194–209. Villemin, T., Coletta, B., 1990. Subsidence in the Rhine Graben: a new compilation of borehole data. Symposium on Rhine-Rhone Rift system; ICL-WG3 Symposium, 23th-24th March 1990, Abstracts. Geol. Inst. Univ, Basel, p. 31. Ziegler, P.A., 1992. European Cenozoic rift system. Tectonophysics 208, 91–111. Zippelt, K., 1988. Modellbildung, Berechnungsstrategie und Beurteilung von Vertikalbewegungen unter Verwendung von Präzisionsnivellements, PhD Thesis, University of Karlsruhe, Bavarian Academy of Sciences and Humanities Munich, 153 pp. Zippelt, K., Dierks, O., 2007. Auswertung von wiederholten Präzisionsnivellements im südlichen Schwarzwald, Bodenseeraum sowie in angrenzenden schweizerischen Landesteilen. NAGRA, NAB 07–27, Wettingen.