Minutes of the EUREF AC Workshop (Brussels, Apr. 10-11, 1997)
by Carine Bruyninx, Royal Observatory of Belgium
Thursday, April 10, 1997
1. The EUREF Permanent GPS Network : Current Status and Future Plans
C. Bruyninx repeats the main goal of the EUREF subcommission which is the establishment, maintenance and enhancement of a three dimensional European reference frame. The anchor sites used to realize the European reference system were in the early years the European VLBI and SLR sites, later the IGS sites and in will in the future be the EUREF permanent stations. Responding the IGS call for participation for regional densification (Jan. 95), the EUREF combined solution is weekly forwarded to the IGS since April 1996. This solution is based on the contributions from 10 European analysis centres : the Italian Space Agency, the Institute for Applied Geodesy, the Bayerische Akademy der Wissenschaften, the Observatory Lustbuhel Graz, the Bundesamt für Landestopographie, the Warsaw University of Technology, the Astronomical Institute of the University of Berne, the Geodetic Observatory Pecny, the Nordic Geodetic Commission GPS data analysis centre and the Royal Observatory of Belgium. In May of 1997, EUREF will submit the combined solution to the IERS, so that in the summer of 1997 the EUREF permanent stations will be included in the latest ITRS realization. As usual C. Boucher and Z. Altamimi will compute the transformation formula's between this ITRS realization and the ETRS. This submission of the EUREF solution to the IERS will require a more rigorous follow up of all receiver/antenna changes in the EUREF sites in order to avoid misinterpretation of the computed station velocities.
2. EUREF Local Analysis Centre Reports
ASI/CGS Contribution to EUREF (M. Fermi, L. Ferraro, A. Nardi, C. Sciarretta, F. Vespe)
L. Ferraro presents the Italian permanent GPS network operated by the Italian Space Agency and the Nuova Telespazio Roma and Matera. The stations are located in Genova, Medicina, Venezia, Cagliari, Noto and Matera ; the data quality of some of the stations is demonstrated. It was commonly known that the station in Matera experienced a height jump in .. 1996 of several cm. L. Ferraro explained that at that time a fledge was mounted around the antenna in Matera, which caused probably the jump in the computed height component. Later the fledge was removed and the height of Matera returned to is normal value. The EUREF processing at the ASI is done using the MicroCosm software with the use of some GIPSY modules to combine the daily solutions into a weekly solution. The ASI LAC is the only EUREF analysis centre which is not using the Bernese software, therefore some of its processing parameters differ slightly from the other analysis centres : no ambiguity resolution is performed and the IGS phase eccentricity variation tables are not applied. A tropospheric zenith path delay is estimated each 2 hours using an a priori absolute weight of 1 cm and no relative weighting. Some of the other participants mended the ASI processing centre to apply less strong absolute weights on the troposphere. Future software improvements will include the introduction of the complete antenna phase centre corrections (elevation dependent) and ocean loading and ocean tides models. The earth tides model will be upgraded. L. Ferraro concludes by stating the open problems in the ASI/CGS solution :
Current Status of Permanent Data Analysis at the Institute for Applied Geodesy (G. Weber)
G. Weber describes first the hardware environment used for the EUREF processing which uses the Bernese 4.0 including its Processing Engine. Thanks to some additional scripts the EUREF processing is performed without any user interaction (15 stations in 30 min). The parameters estimated are station coordinates, tropospheric zenith path delays and ambiguities. Coordinates of the stations Kootwijk, Wettzell and Madrid are fixed during processing. Future plans in Germany include the operation of about 20 additional sites by IfAG. Presently the Wettzell, Osnabruck, Aurich, Karlsruhe and Leipzig sites are already installed and operating. Purpose of this network is to access to a global reference and DGPS needs. Presently time series of station coordinates are available for more than two years with a daily repeatability better than +/- 1cm horizontal and +/- 3cm vertical. Typical is a 13.6 day period in the height component, with an amplitude of 5 mm, probably caused by the fact that the Bernese does not correct the ocean loading effect. G. Weber mentions that the fixing of 3 stations is not optimal. This strategy was choosing at the start of the processing and should be reviewed. However to guarantee the continuity of the results a reprocessing from January 95 to May 97 is one of the plans for the future. Since each analysis centre is fully responsible for each own weekly SINEX contributions, existing SINEX files can be replaced by new ones if documentation about the changed processing strategies/subnetwork are made available.
Basic Information about the EUREF Analysis Centre at the Geodetic Observatory Pecny, Czech Republic (J. Dousa, J. Simek)
J. Simek explains that the GOP processing centre is a collaboration between the Geodetic Observatory of Pecny and the Czech University of Technology. Presently the GOP LAC processes fully automatically 11 stations using the Bernese 4.0 processing engine (1 day/1 hour) J. Dousa mentions that due to the decreasing quality of the Internet connection with the IfAG, data transmission is now also done from CDDIS and GRAZ.
The Nordic Permanent GPS Network and the Nordic Geodetic Commission Analysis Centre (J. Johansson)
J. Johansson presents the EUREF activities of the Nordic Geodetic Commission Working Group on Permanent GPS stations. The members of this working group are from the National Survey and Cadastre from Denmark, the Finish Geodetic Institute, the Norwegian Mapping Authority, the National Land Survey and the Onsala Space Observatory. The main purposes of the Scandinavian permanent GPS networks are the study of the post-glacial uplift and the EUREF and ITRF densification in northern Europe in support of geophysical, geodetic and navigational products. The Finnish, Swedish and Norwegian permanent networks are respectively known as the FinnNet, SATREF and SWEPOS networks. The Danish permanent network is presently set up. All pillars (mounted on solid bedrock) are of the same type in each national network with common features and stability monitoring. In the future all stations will be equipped with the same type of antenna (Dorne Margolin). The EUREF processing is done using a subnetwork of the 3 Scandinavian permanent networks. Although in the past all processing was done using the GIPSY software, the EUREF processing is performed using the Bernese 4.0 Processing Engine. Products from the daily and weekly processing carried out by the NKG and EUREF are available at the Onsala Space Observatory. So far used by the Mapping Agencies also using the Bernese Software.
EUREF Analysis Centre OLG : Strategies, Products and Future Activities (N. Fachbach, G. Stangl) presented by G. Stangl
G. Stangl explains that the OLG LAC is a collaboration between the Austrian Academy of Sciences and the Federal Office of Metrology and Surveying. The analysis software combines the Bernese 4.0 with AUTOBERN which replaces the BPE and runs during weekends. The part of EUREF stations to be analyzed at OLG was selected for investigating geodynamic and meteorological patterns in Europe. Presently 17 stations are part of the OLG subnetwork. The number of Austrian stations will increase and will form an Austrian subnet for Geodesy and Navigation computed with the shortest time delay possible. The OLG analysis centre uses the mean weather conditions in Graz as input parameters for the Saastamoinen model, these input parameters are extrapolated for the other stations. Zenith path delays are estimated each 2 hours using a priori absolute constraints of 10 cm and relative constraints of 2 cm. Special about the OLG approach is the use of control baselines for the localization of errors and net deformations.
Report an Data Processing performed by the WUT EUREF Local Analysis Centre (M. Figurski, M. Piraszewski, J. Rogowski)
M. Figurski explains that in January 1995 a close collaboration between the Warsaw University of Technology and the CODE processing centre was set up with as principal goal the testing of 4 different processing methods for the Polish network. Since January 1996 the WUT analysis centre uses the Bernese 4.0 software for the data analysis of their EUREF subnetwork, which consists of 15 stations (Borowiec, Jozefoslaw, Lamkowko, Borowa Gora, Wroclaw, Penc, Innsbruck,Pecny, Mendeleievo, Modra Pesok, Wettzell, Kootwijk, Metsahovi, Onsala and Matera). The computation strategy used by the WUT analysis centre is similar to the strategy used by the other EUREF LAC's. The minimal elevation is set to .. degrees and mean numbers of resolved ambiguities is equal to 90-93 %. An ionospheric model is computed as a priori model for the QIF ambiguity resolution strategy. In general, satisfactory results are obtained, however there are some jumps in the height component.
The ROB EUREF Analysis Centre (C. Bruyninx)
C. Bruyninx details the processing strategy used by the ROB analysis centre, which is presently including 11 stations in its network. The ROB analysis centre computes a tropospheric zenith delay parameter each 2 hours. A priori values obtained from CODE's troposphere estimation are introduced for the IGS stations and the troposphere is constrained to these values for the IGS stations. The other stations have a priori absolute weights of 10 cm for the troposphere and 5 m relative weights. Minimal elevation used by this processing centre is 10 degrees.
EUREF Activities at the CODE Processing Centre (T. Springer)
T. Springer mentions that when combining all weekly EUREF SINEX files (in the frame of the EUREF contribution to the IERS) a lot of problems were seen : antenna phase centre variation changes, antenna type/height changes, bad data quality for stations equipped with "old" rogues (WETT, MADR), snow on antenna's (KIRU), bad periods for some stations (MEDI, PENC, GRAZ, ...). Most of these problems can be correct by recreating the EUREF combined solution, although some problems remain unsolved and necessite the elimination of some stations for some periods. The internal consistency of the EUREF solutions shows that all solutions are of good quality. The NKG solution had some problems, but this was mainly caused by snow on the antenna's which degraded the data quality. The repeatability of the EUREF combination is around 2 mm in the east and north components for most of the stations and around 5 mm for the height component. Some sites behave worse, but this is mainly due to the above mentioned problems. Comparison of the EUREF combined solution with the ITRF94 coordinates shows very well agreement between coordinate sets with exception of POTS (error in ITRF94 coord), LAMA, VILL (new ITRF station) and UPAD. From several tests with different analysis strategies, the following conclusions are drawn :
Copies of the slides or papers related to all the reports can be obtained from C. Bruyninx.
Friday April 11, 1997
1. Related Reports
Four Years of Daily Analysis of the GPS Data from the Swedish Permanent GPS Network (J. Johansson)
The geophysical goal of the Swedish permanent GPS network is the study of the post-glacial uplift. Comparison of the from GPS obtained (using GIPSY) and the theoretically computed uplift shows a good agreement, except for the region of maximal uplift (1cm/y) probably due to the presence of a GPS station with bad data quality. A lot of studies have been done concerning site specific errors, which is of course an important topic if site velocities need to be obtained. The presence of a randome surrounding the GPS antenna (to avoid snow on the antenna) has a clear influence on the computed height component due to the change in the phase eccentricity variations. Also in the EUREF network sudden height changes were seen due to the use (or change) of randomes.
2. Discussion
Subnetwork redistribution
Some stations like Wettzell (WTZR) are processed by 9 LAC's, while other stations like (a lot of the Scandinavian ones) are only processed by one LAC. As a result the covariance matrix of the EUREF combined solution will be distorted. Moreover, the quality of weekly solution for stations which are only processed by one analysis centre is more difficult to assess.
To remedy to this problem a complete redistribution of the subnetwork processed by each analysis centre would be optimal. This rather theoretical proposal is in practice not possible since most of the EUREF LAC's wish to restrict their subnetwork to the geographical area of their interest.
A supplementary problem is that a lot of LAC's link their subnetwork to the IGS network by including some of the European IGS stations in their subnetwork. Taking into account that the EUREF permanent stations will be included in the ITRF96 list of stations published by the IERS, each of them can be used in the future as anchor station. The unreliable behavior of some IGS stations in Europe (for example Madrid) coming out of the time series should encourage the LAC's to replace such stations and introduce instead a neighboring EUREF station with excellent records into their processing (in the case of Madrid, Villafranca would make a perfect alternative).
In order to make a first step towards a more equal distribution of the analysis the following proposals were put forward :
It should be mentioned that this is only a proposal and no formal engagement has been made by the LAC's to include or exclude the proposed stations. A graph with information about the number of LAC's processing the data of each EUREF station will be included in the EUREF permanent network information system. All LAC's will consult it before changing their subnetwork and try as much as possible to include stations which are only processed by one analysis centre and exclude stations which are processed by 5 to 9 analysis centres. All representatives agreed to contact C. Bruyninx before changing their subnetwork and to send the associated station log file to the network coordinator before including a new EUREF station into their subnetwork.
Recommendations for data analysis
The following processing parameters were agreed upon to be recommended for the processing of EUREF subnetworks :
It is agreed upon that a 15 degrees cut off angle improves the repeatability of the height component significantly with respect to a minimal elevation angle of 20 degrees. There was some discussion about going to an elevation angle lower than 15 degrees, this should de-correlate the height component, but it is only efficient if elevation dependent weighting of the observations is introduced and if the mapping function of the troposphere is reliable below 15 degrees of elevation. It seems that in the future there could be some changes with this respect in the IGS community, but it is too soon now to give more details :
Each LAC is strongly recommended to keep Tim Springer informed about any changes in the LAC's processing centre so that he can assess the influence on the covariance matrix of the LAC's solution.
Exchange of information concerning problems and changes
All analysis centres are asked to keep the network coordinator informed about problems they encounter during processing. The network coordinator will create and update a "problem file" and make it available through the permanent network web site (http://www.oma.be//KSB-ORB/EUREF/eurefhome.html).
Tim Springer will make available the STACRUX file (regrouping station/antenna/receiver changes and problems).
3. Diverse
IGS annual report 1996
J. Zumberge contacted all LAC to contribute to the IGS annual report 1996. His preference goes to one contribution regrouping all the activities related to the EUREF permanent network.
The EUREF report will have the following structure :
C. Bruyninx will merge all contributions into one EUREF contribution. In order to be included in the EUREF report to IGS, the LAC contributions (text preferable in ascii format) will have to be sent to C. Bruyninx before April 23. The final EUREF contribution will then be forwarded by C. Bruyninx to the IGS. The IGS CB will reformat the final text, so that from our side no special attention has to go to the layout. Figures and graphics are accepted by the IGS CB in numerous of commonly used formats.
EUREF symposium, June 4-7, 1997 in Sofia
Before the symposium (on Monday June 2), an EUREF LAC meeting was originally planned (before planning the workshop in Brussels). LAC's who wish a second EUREF LAC meeting at Sofia can contact C. Bruyninx who will make the necessary arrangements for this meeting. The EUREF Technical Working Group meeting which will take place on Tuesday June 3 and a detailed report from the network coordinator about the present status of the EUREF permanent network is on the agenda.
A special session about the EUREF Permanent Network is planned on Wednesday June 4. The EUREF Technical Working Group encourages all EUREF LAC's to make a paper presentation during this session.
IAG symposium, September 3-9, 1997 in Rio de Janeiro
The EUREF TWG has asked the network coordinator to organize a contribution to the IAG Symposium detailing the development of the EUREF Permanent Network. All LAC's have been contacted in the past month and a abstract was submitted to the organizing committee. The title of the presentation is :
"The Use of the EUREF Permanent GPS Network for the Maintenance of the European Terrestrial Reference Frame".
J. Johansson has agreed to present the paper at the IAG symposium on behalf of the EUREF LAC's. The IAG presentation will mainly be based on the EUREF contribution submitted to the IGS for the annual report 1996.