BIFROST Project: Measuring 3-D crustal deformation in Fennoscandia with continuous GPS observations

Continuous GPS in permanent networks - the cases of Sweden and Finland

Historical Perspective - the Beginnings

Finding the tools.
The first plans of GPS activities envisaged campaign type observations. However, at the beginning of the 1990-ies the National Land Survey of Sweden (NLS) had sketched in their program "Geodesy 90" two networks of satellite geodetic points in Sweden, fundamental points and densification points. These plans envisaged permanent monuments and infrastructure  (electricity, telephone) at the SWEPOS stations. In 1991 the mutual benefit for SWEPOS and the DOSE project was realized if all of these stations could be equipped permanently with receivers  and data downloading equipment (modem, PC). In joint planning meetings in 1991/92 station locations became finalised and a research equipment proposal was set up by OSO for the purchase of 16 TurboRogue receivers and antennas.

BIFROST.
In 1994 Jim Davis coined the acronym BIFROST (Baseline Inferences for Fennoscandian Rebound Observations, Sea-level and Tectonics) for the science investigations utilising the permanent GPS networks for the post-glacial rebound work. Bifrost in nordic mythology is the name for the rainbow, and the rainbow spans over the entrance port into Valhalla, the place of the immortal heroes. At the end of the rainbow you might dig for gold.
     Today the project group consists of researchers at  Chalmers, at the Harvard-Smithsonian Center for Astrophysics (USA), at the University of Toronto (Canada), at the University of Durham (UK), at the Finnish Geodetic Institute, and at the National Land Survey of Sweden.

EUREF.
Efforts to create EUREF, the European densification of the International Terrestrial Reference Frame where answered by the nordic countries in defining a set of stations at about five times the IGS station density, making the GPS data available to the EUREF groups, and setting up an NKG (Nordic Geodetic Commission) analysis centre. This centre operates at Onsala Space Observatory.
 

Creating Permanent Satellite Geodetic Networks

Finland.
Planning of the Finnish network FinnRef for continuous GPS observation capabilities commenced in 1991/92 at the Finnish Geodetic Institute (FGI). The first station to become operational was Metsähovi near Helsinki beginning of 1992, an IGS site. The first new stations became operational in 1994, and by autumn 1996 the system was completed, covering the whole country quite uniformly.  Three stations of this network were erected in cooperation with the company Posiva OY  (specialised on nuclear waste disposal) in order to monitor crustal stability in Finland. These antennas are mounted on steel-inforced concrete pillars. The other stations have steel grid masts of varying  height (up to 8 m), except at the Metsähovi station, which has a 25 m high mast that for this reason furnishes thermal expansion compensation using an invar steel rod. Receivers used today are Ashtech Z12 except Metsähovi (TurboRogue). (Summarised from Koivula et al., 1997)

Norway.
Between 1989 and 1993 the Norwegian Mapping Agency  built up their SATREF system specialised for navigation at sea and DGPS. The stations were not equipped with choke-ring antennas before October 1997. Also see the EUREF tracking network page for more information.
 

Permanent stations in Sweden supporting the BIFROST project.

• Utilisation of access to crystalline bedrock, and careful coupling of monument to bedrock;
• Length control and temperature compensation of pillars;
• Preventing of snow accumulation on antennas during winter;
• Securing power supply during grid outages;
• Securing data in the case of telephone line inaccessibility.

• concrete steel-inforced pillars with a heating coil thermostatically controlled at 20^o C;
• steel rods driven 1 m into bedrock;
• Dorne-Margoline B choke ring antenna, Ashtech receivers (other receivers exist for applications outside BIFROST);
• hemispherical acrylic glass radome with geometrical centre in the antenna element;
• local survey control points in a 20 m diameter area;
• remote telephone-dial-up station reboot;
• leased telephone lines;
• connection to electrical power grid;
• battery backup power supplies.

Data flow

FinnRef.
Download by FGI via dialed-up telephone line. Transfer to OSO.
 

Experience

Radomes.
Other kinds than hemispherical radomes appear to cause deviation from spherical symmetry and hence comparatively large and mostly vertical offsets of the effective antenna centre and an elevation-dependent antenna sensitivity pattern (Emardson et al., 1998a,  Ågren, 1997). Snow accumulation is a general problem affecting the effective antenna position (Jaldehag et al., 1996),  and none of the four tested radome designs seems to prevent the problem. We also found that radome changes carried out at a number a stations in the course of a few days affect the network solutions of remaining stations, both in the geodetic and in the atmospheric parameters.

Local Surveys.
The problem attacked here is that tilting of the pillar or other kinds of local changes of the antenna position should be monitored with a local independent method. The Finnish design lets the antenna stay in place while the survey instrument is carried to the local control points. In the Swedish design the antenna is exchanged against a theodolite and markers at the control points are observed instead. The latter solution has the disadvantage that dismounting and re-installation of the antenna is prone to create mishaps, apart from the loss of data during the measurement.
 

Applications

Atmosphere.
Our standard processing of the continuous GPS data results in estimates of atmospheric propagation delay parameters. A network which is spatially denser than existing radiosonde networks has a profound and useful capability for meteorology, weather forecasting and climate studies. Furthermore, GPS data are acquired with a temporal resolution superior to any reasonable launching frequency of weather balloons. We have shown that the SWEPOS network can be used to follow regional scale weather systems such as air masses of different water vapour contents (Elgered et al. 1997, Davis and Elgered, 1998) The part of the propagation delay caused by atmospheric water vapour could be estimated with sufficient precision from the GPS data and be recomputed as time-series of the local, vertically integrated estimate for precipitable water at each station. If it is possible to derive the estimates with the required accuracy of this parameter almost in real time.
     GPS data from northern Europe have been used in the climate research project BALTEX which is supported by the WMO (Emardson et al. 1998a; Yang et al. 1999).  The goal of the project is to study the water and energy balance of the catchment area of the Baltic Sea. Our contribution is presently a data analysis producing time series of the atmospheric water vapour obtained form GPS data acquired at 50 sites in the area.
     The application of climate monitoring obviously require both long time series and small systematic errors.  GPS data have yet not resulted in a reasonably long time series useful for climate monitoring. The systematic effects seen in the atmospheric estimates form the Swedish network has been traced to the electro-magnetic environment at the GPS antennas (Emardson et al., 2000).
     Using data from a small part of the SWEPOS network we have assessed the usefulness of different interpolation models over spatial scales of tens of kilometres (Emardson and Johansson 1998). Such model would be required for real time positioning based on the GPS phase measurements by a moving receiver supported by fixed reference stations.

Climate: BALTEX project.
See a figure showing the catchment area of the Baltic Sea and the continuous GPS stations participating in BALTEX project. You can see time series of  GPS estimated vertically integrated content of precipitable water at two stations, Onsala and Visby, along with the corresponding analysis obtained from a high-resolution regional weather prediction model.

NewRTK.
Navigation satellite systems are in wide-spread use in a large number of applications. A general trend is the demand for better accuracy in real-time applications. Therefore, modelling of major error sources in real time is required.
 A Network Real-Time Kinematic service based on the SWEPOS network is conceived in the future such that GPS phase corrections for neutral atmosphere and ionosphere are derived from the network solutions and distributed nation-wide to users roving in the field. A typical NewRTK user would carry a single-frequency GPS receiver and a radio receiver with a built-in DARC decoder (Data Radio Channel). The goal of the project is to establish a service able to provide users with centimetre-level accuracy in real-time. Similar projects are run in the other nordic countries and efforts are undertaken to work towards a common nordic and international standards.

Other real-time services.
Developments for the nearer future worth mentioning are services in support of real-time positioning and navigation that distribute satellite integrity messages derived from the IGS network or from Wide Area Augmentation Systems (WAAS) like EGNOS (which has "visibility" problems at high latitudes).

Time Transfer.
The station Borås, located at the SP  Swedish National Testing and Research Institute,  is a member of SWEPOS and collocated with the national laboratory for time and frequency. Three cesium frequency standards are available at SP and one is used for the realization of UTC(SP). The station is a member of the international timing network from which UTC (Coordinated Universal Time) and International Atomic Time (TAI) is calculated. Two GPS receivers and one GLONASS receiver are hosted in a continuously temperature-controlled and monitored environment set at  24  ± 0.5 ^oC. The receivers all utilises external 5 MHz from the cesium frequency standard on which UTC(SP) is based. The GPS station in Borås is unique in the sense that, in addition to the GPS receivers, also the antenna cable is thermally controlled. It is embedded in a water pipe set at 7 ±1 ^oC in order to practically eliminate electrical delay variations. Furthermore, the entire pillar is temperature-controlled by means of electrical heating and cooling water circuits. A similar setup has recently been established at at the Onsala IGS station. The Onsala station with its two hydrogen maser clocks will also participate in the international timing network.

Geodetic Reference.
The SWEPOS array forms the reference network for precise geodesy, mapping and cadastre in Sweden. Selected stations in Sweden, Finland, Norway, Denmark and the Baltic states are processed at the NKG analysis centre (located at OSO) for the European reference frame (EUREF, Bruyninx et al., 1998).

BIFROST (Solid Earth).
Sixteen hundred days of continuous GPS observations have been providing a data base from which motions and related parameters can be derived. Apart from fitting straight lines to the position time series  in order to infer constant rates of motion throughout the temporal extent of the data, the batch of information provides important clues as to the noise character and perturbing effects that affect the GPS positions and the inferred motion.
     An example is given: Sundsvall, black curve are observed position time series with green background showing one-sigma standard deviation, red oscillatory curve contains annual and sub-annual variations.
     Thus, investigations are pursued exploring the data on a statistical basis, but also computing forward model predictions of for instance atmospherically and hydrologically induced loading effects. The maps of motion that we arrive at (horizontal, vertical) *)  (vertical rates isolines and colour map) shows clear relation to our post-glacial rebound model.
      Looking at baseline length rates the high correlation between observations and model becomes even more clear.  In the baseline figure the station SAAR apparently creates a group of outliers. The anomalous behaviour of this station is also seen in the map of horizontal motion. SAAR is not a standard SWEPOS station, rather it is monumented on a building. It is operated by ESA at Kiruna for the IGS network. Baseline length rates have the advantage of being invariant to rotations and translations (of the reference frames). In our region the baselines are almost horizontal.
     The observed motions in our maps are based on an empirical orthogonal function approach of time series analysis in order to attenuate regionally correlated noise (example shows Sundsvall vertical, horizontal, red wiggly curve the EOF predicted, regionally coherent motion). The vertical results from BIFROST-GPS have an important bearing on the determination of the component of regional sea level change that is unrelated to the crustal motion.
     Previous communications of  SWEPOS GPS operations and  results from BIFROST see Scherneck et al. (1996) and Scherneck et al. (1998) .

The remaining systematic errors which might influence the crustal motion results are summarised as follows

• scattering - microwave scattering in the antenna environment may couple with the slowly emerging satellite geometry and thus introduce long-term perturbations of effective antenna centre (Jaldehag et al., 1996,  1996a,Elósegui et al., 1995).
• temporal variation of scatterers and reflectivity - the reflectivity of the ground around the stations may introduce seasonal variations.
• hydrologically induced loading deformations - the hydrological records might not be accurate enough (hgs99).
• atmospheric perturbations - wet delay gradients and other higher degree models (keyword: atmospheric tomography) might have to be estimated, but there are limitations in the number of solvable parameters given the observation geometry, network density etc. For a summary see Emardson and Jarlemark, (1999).

Ocean loading tides.
Using a precise point positioning processing option in GIPSY/OASIS-II GPS observations are used to solve for residual station displacements, for instance ocean loading tides (Scherneck, 1991,  Scherneck et al., 2000).

Go to Bibliography and References.

*) Some systems complain "there is no windowed ancestor". If you have similar problems viewing these links, save information on your disk and use a less sophisticated viewer, for instance the raw ghostscript rather than ghostview. Or accept GIF graphics (horizontal, vertical) .  Back to text.