Special Study Group 1.181:

„Regional Permanent Arrays“

 

R. Weber

 

 

Introduction

 

The SSG 1.181 has been established by Section I of the IAG at the 22nd General Assembly (Birmingham) in August 1999. The basic idea was to pay attention and take advantage of the increasing number of GPS reference stations which were set up on both global and regional scales in recent years. Ideally, the latter should represent local densifications of the ITRF polyhedron. While, at the outset, these stations were built up in most cases to monitor active tectonic regions, recently the augmentation of real time surveying and probing the atmosphere have become more important tasks.

 

 

Objectives

 

The work of this study group aims at the tie of regional GPS networks to the ITRF as well as to study ambiguity resolution within a network of multiple reference stations at baselines with a length of up to several tens of kilometres. Especially the appropriate modelling of ionosphere and troposphere path delays as the limiting factors for ambiguity resolution and the influence of antenna phase centre variations should be discussed. Concepts and realisations of virtual reference stations will be compared. RTK solutions within active reference station networks, the benefits of using combined GPS/GLONASS receivers as well as the use of predicted IGS orbits are also subject of the investigations.

 

 

Members & Corresponding Members

 

The membership-list comprises 15 regular and 7 corresponding members (including chair-persons of the remaining Special Study Groups within Section I)

 

Members:

R. Weber          (Austria)  Chair

R. Bingley         (UK)

H. Bock           (Switzerland)

C. Bruyninx       (Belgium)

P. Clarke                      (UK)

H. Dragert        (Canada)

H. Hartinger     (Austria)

T. Herring          (USA)

J. Johansson    (Sweden)

P. de Jonge                  (USA)

T. Kato              (Japan)

A. Kenyeres      (Hungary)

J.F. Galera Monico (Brasil)

E. Ostrovsky     (Israel)

L. Wanninger    (Germany)

 

 

Corresponding Members

S. Han              (Australia)

M. Hernandez Pajares   (Spain)

H. van der Marel(Netherlands)

L. Mervart          (Czech Republic)

S. Skone          (Canada)

M. Stewart        (Australia)

H. Titz              (Austria)

 

 

SSG 1.181 Research

 

Primarily for communication  and information exchange between the Special Study Group Members a Web-Site has been established which is accessible via http://luna.tuwien.ac.at/ssg1181/ssg1181.htm . This Web-site summarizes the Working Programme which has been decomposed in a couple of work-frames (WF) discussed below. Basic assumption for all work-frames was the existence of a regional reference station network, equipped with dual frequency GPS (or combined GPS/GLONASS) receivers. The distance between the network stations should not be less than 20km and should , which was an outcome of our work, not exceed 100km. In order to determine rover coordinates we take advantage from the apriori knowledge of reference station coordinates at the +/-2mm level.

 

 

WF1: Reference Frame

Regional reference station networks are usually tied to realizations of the ITRS (ITRFxx). The tie may be established by regularly (weekly) processing station coordinates with respect to the closest ITRF stations using precise IGS orbits. Mean coordinates calculated over a 3 weeks span are of sufficient accuracy to be fixed for a subsequent modelling of the remaining error sources within the network area. Satellite orbits are provided in the most recent ITRS realization (e.g. ITRF2000), valid at the epoch of issue ti. Regional networks, operated by local companies or national surveying authorities, are advantageously embedded in the ITRFxx as well, but at a stable reference epoch t0. In order to calculate rover coordinates with respect to the regional frame the user has to transform the ITRFxx coordinates by means of a valid velocity model and the conformal similitude transformation into ITRF2000 (parameters provided by IERS) at the current epoch ti. After baseline processing the new station coordinates have to be transformed back to ITRFxx at epoch t0. Another option is to transform the satellite coordinates to the appropriate frame and epoch before baseline processing.

 

In order to serve the user community which is usually not pleased to work with frequently updated frames and changing coordinates one might ignore horizontal velocities and freeze the station coordinates at a given epoch (e.g. ETRF2000, epoch 1989.0). But keep in mind that caused by plate motion, this regional frame can deviate from ITRF2000 by a couple of centimetres per year.

 

The link of ITRFxx coordinates to a local geodetic datum is performed again by means of a similitude transformation. Transformation parameters are provided by local surveying authorities.

 

 

WF 2: Impact of the Atmosphere (Ionosphere, Troposphere)

The Ionosphere is an inhomogeneous (it consists of a number of horizontal layers in an altitude between 60 – 1000km with varying density of charged particles), anisotrop (the refractive index depends on the propagation direction of the wave) and dispersive (the phase velocity of a wave is frequency dependent) medium. The impact of varying ionospheric conditions within the area of the permanent station network is still the most restricting factor in view of reliable and precise (near) real time point positioning. We may separate the ionospheric refraction in a large scale and a medium scale part. The large scale part (absolute electron content) may be sufficiently described by a single layer model with a model height of about 450km in which all free electrons are assumed to be concentrated. The medium scale part is dominated by medium scale ionospheric disturbances which can be recorded by regional permanent networks. Small scale ionospheric disturbances with wavelengths of a few hundred meters (ionospheric scintillations) cannot be captured by observations of such a network.

 

On basis of a single layer model the phase propagation effect at the single station P can be described by

                       

where VEC denotes the Vertical Electron content, z is the zenith distance of the satellite, f the frequency of the wave and A=constant= 80.6 m3s-2. If we convert this equation in order to describe the differential effect between stations P and Q we find

 

This formula separates the absolute (first term) from the relative ionospheric propagation error. Both terms are in first approximation proportional to the baseline length d which allows for a linear interpolation between the reference stations.

Vertical delays almost cancel out in single difference observations over baselines smaller than 100km but horizontal gradients, which might reach up to 10mm/km at solar maximum (VEC = 100* 1016 m-2), do not. Unfortunately double differencing the observations does not reduce first order ionospheric effects, but their single difference effects are added.

 

The hydrostatic part of the tropospheric delay can be modelled with sufficient accuracy by well-known standard models. In relative positioning these models usually account for the height difference of the stations. The delay in zenith direction can be mapped afterwards by means of more or less experienced mapping functions (e.g. Niell mapping function) to the satellites elevation. Residual tropospheric refraction may stem from the wet delay component and from small scale local disturbances. We distinguish between an error in modelling the tropospheric delay at the starting point P of our baseline (absolute) and the relative tropospheric delay between P and Q. We can state that the relative tropospheric delay is again proportional to the distance between the reference stations and can be interpolated properly. Satellite specific interpolation errors might reach up to 1cm (at 10 degree elevation angle). Besides this satellite specific interpolation scheme more general models describe the residual zenith delay for all satellites as a whole. This allows for a clear separation of tropospheric from orbital errors. In general large height differences (> 1000m) within the reference station network should be avoided.

 

 

WF 3: Satellite Orbits

Three kinds of freely available satellite orbits have been investigated: Broadcast ephemeris, IGS precise orbits and IGS Ultra Rapid orbits. Broadcast ephemeris are part of the navigation message and currently of a quality of about +/- 2 meters. Deviations up to 80 meters for specific satellites are possible. Real time applications are also served by the IGS Ultra Rapid Orbits (IGU) which provide 48 hours satellite ephemeris comprising a 24 hours observed and a 24 hours predicted part. The IGU orbits are of a high quality for most of the given satellites (< 25 cm) but usually 2-3 satellites are missing. IGS precise ephemeris are the most accurate choice. They provide satellite orbits with an accuracy of a few centimetres and corresponding clock corrections at the 0.1 ns level. Due to the delayed availability of 2 weeks their use is restricted to post processing applications.

 

To discuss the impact of errors in orbit representation we may decompose the whole error vector (vector between calculated and true satellite position) in 3 components. The first component  points from the satellite towards the first site of our baseline, the second component  is parallel to our baseline and the third component is perpendicular to  and . A good approximation for the difference in pseudorange measurements at the start and end point of our baseline  then reads: , where d denotes the baseline length, D is the rough distance between satellite and receiver and  is the height angle of the satellite. Thus, in relative point determination the effect of orbital errors is obviously proportional to the baseline length. In view of a given reference station network these errors are easy to interpolate between the reference sites at the mm –level if the distance between reference sites does not exceed 100km and the orbital errors do not exceed 80 m.

 

In summary orbital errors are satellite specific and usually of long-periodic character. A change between sets of broadcast ephemeris might introduce unpredictable jumps in the orbital representation of several tenth of meters. Currently standard rover software is not well suited for handling IGU orbit information, which restricts their use for precise positioning in real time.

 

 

WF 4: Concept of Virtual Reference Stations

Calculating the position of a rover receiver which is located close to one or more reference stations in post-processing mode is a well-known task and widely documented in literature. We may distinguish between models restricted to baseline lengths up to 15 km which solely need short period observation data (Fast Static) and ambiguity resolution techniques valid over large baselines which expect 60 minutes or more observation data to achieve cm-accuracy.

 

More recent concepts take full advantage of observations obtained at permanent stations of a regional network to provide cm position accuracy in real-time within the covered area. These concepts are based on a two dimensional modelling of distance and azimuth dependent error sources as listed in the previous sections. The models in use allow to separate ionospheric refraction on the one hand from orbital deviations and tropospheric refraction by means of the geometry-free and ionospheric-free linear combinations. Parameters of the satellite specific two-dimensional models are fitted advantageously on basis of zero difference residuals. A further separation of orbital and tropospheric errors is also possible but demands a common modelling of residual zenith path delays. Basic assumption for calculating correction models is that all ambiguities within the reference network have to be solved correctly in advance. Model parameters should vary very slowly to reduce the effort of frequent updates. The gain of correction models obviously diminishes with increasing baseline lengths. A current suggestion concerning the distance between the reference sites is 50-70km.

 

Calculations are usually carried out at the computer center of the reference station provider. In existence of a two way data-link between rover and computer center the provider simulates on basis of the approximate position of the rover and the obtained error models observation data of a so-called Virtual Reference Station (VRS). Station dependent corrections like antenna phase center variations are also accounted for. The VRS observation data can be distributed via GSM, GPRS or radio link in RTCM format. The VRS concept is clearly superior to usual network adjustment at the rover due to the smaller of amount of data which has to be transferred between reference and rover (basically the VRS data plus correction models) and the quality check already performed at the computer center. Last, but not least, point positioning can be carried out with standard software at the rover.

 

The VRS concept is able to supply a huge number of users within the covered area with correction data. Networks supporting this concept are under construction or were already established e.g. in Switzerland, Norway, Austria and Germany. A slightly different concept which also models the main error sources within the area of interest and applies these models to a code and phase data based single point position of the rover is the well-known Precise Point Positioning (PPP). This concept of precise positioning is successfully realized e.g. in Canada and Sweden.

 

 

WF 5: GPS/GLONASS integration

Due to the renaissance of the GLONASS system the number of reference sites equipped with dual frequency-dual system receivers is steadily rising. The use of GLONASS satellites, in addition to GPS, improves in various cases the ability to fix the rover position within the reference station network. Moreover it allows for an improved monitoring of the troposphere and ionosphere. These advantages stem from the increased number of active satellites (currently 27 GPS + 10 GLONASS satellites), the slightly improved geometry and slightly different center frequencies. On the other hand, positioning with GLONASS suffers from a number of disadvantages which have to be tackled. The still low number of active GLONASS satellites leaves periods with less than 2 satellites above horizon of the permanent station network. This harms or even prevents ambiguity resolution within the network. Moreover inter-system ambiguity resolution is more complicated than ambiguity fixing within one system. The PZ90 broadcast ephemeris provide state vectors which have to be updated more frequently compared to osculating elements. Precise GLONASS orbit predictions are still not in view. The impact of the mentioned drawbacks will continuously decrease as the number of active GLONASS navigation satellites will, according to plans of the Russian Federation Ministry of Defense, increase to about 18 till 2006.

 

A more detailed discussion of topics covered by the SSG would go beyond the scope of this summary. Thus, the interested reader is referred to the large number of publications provided by the SSG members over the past 4 years.

 

Robert Weber

(Chair of SSG 1.181; rweber@luna.tuwien.ac.at )

 

Below is a reference list of recent publications of the SSG members related to the topics of this study group.

 

 

References:

 

R M Bingley, A H Dodson, N T Penna, F N Teferle, S J Booth and T F Baker. "Using a Combination of Continuous and Episodic GPS Data to Separate Crustal Movements and Sea Level Changes at Tide Gauges in the UK." Book of Extended Abstracts of the WEGENER 2000 Conference, San Fernando, Spain, September 2000.

 

A H Dodson, R M Bingley, N T Penna and M H O Aquino. "A National Network of Continuously Operating GPS Receivers for the UK." Geodesy Beyond 2000, The Challenges of the First Decade, Edited by Schwarz, International Association of Geodesy Symposia, Vol 121, Springer-Verlag, 2000, ISSN 0939-9585, ISBN 3-540-67002-5, pp 367-372.


Adam J., W. Augath, C. Boucher, C. Bruyninx, P. Dunkley, E. Gubler, W. Gurtner, H. Hornik, H. van der Marel, W. Schlüter, H. Seeger, M. Vermeer, J.B. Zielinski, 2000,"The European Reference System Coming of Age", International Association of Geodesy Symposia, IAG Scientific Assembly, Springer, ed. K.-P. Schwarz, Vol. 121, pp. 47-54

 

Bruyninx C., 2001, "Overview of the EUREF Permanent Network and the Network Coordination Activities", EUREF Publication, EUREF Publication, eds. J. Torres, H.Hornik, Bayerische Akademie der Wissenschaften, München, Germany, No 9, pp. 24-30 Bruyninx C. and M. Yseboodt, 2001, "Frequency Analysis of GPS time series from the ROB EUREF analysis",
EUREF Publication, EUREF Publication, eds. J. Torres, H. Hornik, Bayerischen Akademie der  Wissenschaften, München, Germany, No. 9, pp. 37-42


Becker M., C. Bruyninx, D. Ineichen (2001) "The EUREF RNAAC: 1999 Bi-Annual Report", IGS Technical Reports, eds. I. Mueller, R. Neilan, K. Gowey, Pasadena, JPL, Pasadena (in press)

 

Bruyninx C., M. Becker and G. Stangl, (2001) "Regional Densification of the IGS in Europe Using the EUREF Permanent GPS Network (EPN)", submitted to special issue of Physics and Chemistry of the Earth, Proc. IGS Network Communications Workshop, Oslo, Norway, July 2000 (in press)

 

G. Blewitt, D. Lavallee, P.J. Clarke, K. Nurutdinov, W.E. Holt, C. Kreemer, C.M. Meertens, W.S. Shiver and S. Stein (2000). “GPSVEL project: towards a dense global GPS velocity field”. In 10th General Assembly of the WEGENER Project, extended abstracts book, Boletin RAO 3/2000.


A. Sehmisch, O. Boehme, S. Canan, P.J. Clarke, G. Taylor and S. Twynholm (2001). Integration of GPS and GIS in vehicle tracking systems. GISRUK 2001 abstracts volume.

 

G.A.Milne, J.L.Davis, J.X. Mitrovica, H.G. Scherneck, J.M. Johansson, M.Vermeer, H.Koivula (2001), “Space Geodetic Constraints on Glacial Isostatic Adjustment in Fennoscandia”. Science, Vol.291, pp.2381-2385.

 

Monico J. F. G. (2000),Precise Point Positioning: Using GPS: A solution for Geodynamics,Brazilian Journal of Geophysics Vol.17, 2000  ( In portuguese) S.M. Alves Costa; E.S. Fonseca Junior; J. A. Fazan, J.F.G. Monico, P. O. Camargo. (2001), “Preliminary Results of SIRGAS 2000 Campaign” - IBGE Analysis Center: IAG Workshop - Cartegena.

 

Ostrovsky E.  (2001), “The G1 GPS geodetic-geodynamic reference network in Israel”, Israel Journal of Earth Sciences, Issue on Geodetic Studies in Israel and the Eastern Mediterranean Region. Titz H. (2000), “Regionale GPS/GLONASS Echtzeitsysteme in Österreich”, Geowissenschaftliche Mitteilungen Vol.50, pp.11-20, TU-Vienna.

 

Wanninger, L. (1999): “The Performance of Virtual Reference Stations in Active Geodetic GPS-networks under Solar Maximum Conditions”, Proc. of ION GPS '99, Nashville, pp 1419-1427. Wanninger, L., May, M. (2000): “Carrier Phase Multipath Calibration of GPS Reference Stations”, roc. of ION GPS 2000, Salt Lake City, pp 132-144. 


Wanninger, L., (2000): “Präzise Positionierung in regionalen GPS-Referenzstationsnetzen” Deutsche Geodätische Kommission , Series C, Volume 508, Munich.

Weber R., Fragner E. (2001) „The Quality of Precise GLONASS Ephemerides”, Proceedings of the 33. COSPAR Assembly, Warsawa , July 2000

 

Weber R. (2001) The International GLONASS Service – Pilot Project GPS Solutions, Vol.4 No.4, pp.61-67,2001

 

Böhm J., Schuh H., Weber R. (2001) „Comparison of troposheric gradients determined by VLBI and GPS”, Physics and Chemistry of the Earth, Part A, Vol. 26/6-8, pp. 385-388.