Second ILRS AWG Meeting (Delft 2000)
Minutes of ILRS Analysis Working Group Workshop #2
Delft, The Netherlands
May 22-23, 2000
Agenda (Appendix 1)
Attendees (detailed list Appendix 2):
Monday, May 22, 2000
1. Opening
Introductory remarks were given by Noomen. Pearlman and Noomen stressed the importance of the current ILRS AWG activities (the pilot project "positioning + earth orientation" in particular) for the future of the SLR/LLR community. The agenda was approved. Everyone present briefly introduced
2. Actions since Frankfurt
Updates on these actions are interwoven within the following sections.
3. Minutes from Frankfurt
Noomen briefly reported on the history of the ILRS, and the AWG in particular (Appendix 3). He mentioned the science goals and products of the ILRS, the customers, and the history of the AWG: membership, meetings, and pilot projects. The most important pilot project is the one on "positioning and earth orientation". The main theme of this project and the focus of the series of workshops to date is the development of an official ILRS product for submission to the IERS. This project is being conducted in a stepped approach, which allows for problems to be encountered and resolved at an early stage and to be solved with a minimum of effort and a minimum need for recomputations. It also keeps the project open to new participants to join along the way.
4. ILRS pilot project "POS+EOP"
4.1. Individual contributions
Noomen gave an overview of the contributions that have been received for the second phase of this project. Eleven organizations provided solutions, most including a time-series of Earth Orientation Parameters (EOPs) for both 2-day and 3-day intervals as agreed in January 2000 in Frankfurt. Solutions were submitted within 2 weeks of the March 1 deadline, although some had to be resubmitted with corrections for errors in solution content and/or format (DOMES numbers, numerical problems, etc.).
Next, the analysts reported briefly on their results.
Devoti showed two loosely constrained solutions (Appendix 4). In his SINEX representation, all sigmas (a priori and a posteriori) were scaled with a variance factor, which caused confusion among the analysts. The description of the SINEX format is not 100% clear in this respect. Action item Noomen.
Mareyen presented the BKG solutions (Appendix 5). She introduced the issue of how to handle "bad" stations (i.e. poor data quality or a very small data set). Should the data be thrown away altogether, or should we apply a low weight to it? The latter proved to be no solution, but only numerically degraded the covariance matrix further. After some debate, it was agreed that the individual analyst should reach his/her own conclusions, albeit that the SINEX file and the matrices within must be numerically stable.
Otsubo presented the CRL results, giving fits of about 2.1 cm for both series of parameters (Appendix 6). Otsubo questioned the equivalence of COVA and INFO representations in the SINEX files, and showed that the 3-day EOP solutions show a more stable pattern with respect to the IERS 1-day parameters than the 2-day EOP solutions.
Noomen presented four solutions provided by DEOS (2-day and 3-day EOPs, with and without range biases) (Appendix 7). He reported on the computation model and the constraints, and gave details on the iterations that had to be done to satisfy the SINEX/ILRS requirements.
Biancale presented the GRGS contributions (Appendix 8). He discussed the computation model that was applied, and talked about the GRGS strategy for eliminating poor stations. Adding additional satellites, LAGEOS-2 as a minimum, and preferably also the Etalons, should improve the solutions, particularly the UT component.
The results of the NERC group were presented by Otsubo (Appendix 9). NERC rejected 4 stations altogether, and solved for range biases for a few other stations. Otsubo recommended that the SINEX format be extended to include range biases as an additional parameter.
4.2. Comparisons
Devoti reported on transformations and comparisons of network and EOP solutions performed at ASI (Appendix 10). When the Helmert parameters to transform from individual network solutions to ITRF97 were applied to the EOP solutions, the agreement with the IERS Bulletin B series generally improved, i.e. systematic differences disappear. An exception (unexplained so far) was the UT component for the 3-day time-series. The (dis)agreement may be used as a tool to detect self-inconsistencies, and thus possible problems in the individual analysis. Another option is to solve for the Helmert parameters using both network and EOP information simultaneously. Self-consistent solutions will provide good results; inconsistent ones will not. Some problems with self-consistency appear in most of the current contributions.
Husson gave a presentation on network comparison, focussing on format and data integrity issues (Appendix 11). He identified a large number of problems in the submissions and will provide feedback to the individual data analysts. Action item: Husson
Nurutdinov presented preliminary results of a network comparison/combination (Appendix 12). He addressed the issues of deconstraining, editing, combination, constraining the combined solution and statistics in detail, using the TANYA software. The combined solution proves sensitive to the number of input solutions.
Tuesday, May 23, 2000
4.3. Issues
4.3.1. Procedures
None of the analysts had comments on the procedures used for the current pilot project "positioning + earth orientation" (provision of the dataset in ILRS NP format, timeline, submission procedure).
4.3.2. SINEX Format
The various elements of the SINEX format were reviewed for inconsistencies and unclear definitions. The general issues (mandatory/optional/recommended blocks) were already reviewed in Frankfurt. In summary:
1. Header line mandatory
The constraint code should be "2" (at least for the pilot project). This is to be used as a check (i.e. not to be provided by "hard coding" in the software). The entries for the solution contents have to be separated by 1 blank (and not more). All numbers should be preceded by leading zeros (if relevant). The date parameters should specify the epochs of the first and last observations, respectively.
2. FILE/REFERENCE mandatory
3. FILE/COMMENT mandatory (was recommended previously)
Reason for change: specify the center-of-mass value and the range bias solutions and/or model in this section (the later parameters may be included in a future new block in the SINEX format).
4. INPUT/HISTORY recommended
5. INPUT/FILES optional
6. INPUT/ACKNOWLEDGEMENTS optional
7. SITE/ID mandatory
The point code is a parameter which is used for (SINEX internal) referencing only (i.e. between blocks). In principle, this should be "A", unless there are more than 1 solutions available for a station (-> "B" etc.) (to be checked with SINEX format coordinator; action item Noomen). The sign of the latitude and/or longitude (if negative) should be included in the "degrees" component only, and not repeated in the "minutes" and "seconds" components.
8. SITE/DATA optional
The times should refer to the observation epochs.
9. SITE/RECEIVER not relevant
10. SITE/ANTENNA not relevant
11. SITE/GPS_PHASE_CENTER not relevant
12. SITE/ECCENTRICITY mandatory
Eccentricities should be zero for 7824 and 7810 (i.e. the coordinates solutions must represent the optical centers of the new positions). Check the contents with the official ILRS values (cf. list of action items; the URL will be ftp://cddis.gsfc.nasa.gov/slrocc/slrecc.txt).
13. SOLUTION/EPOCHS mandatory
1st Epoch should be filled with epoch of 1st observation; 2nd epoch should be filled with epoch of last observation; 3rd epoch should be filled with centroid (i.e. "weighted mean") epoch of observations. Note: The header of this block is SOLUTION/EPOCHS, and not SOLUTION/EPOCH. The documentation in the Sinex file description is ambiguous.
14. SOLUTION/STATISTICS mandatory
For the pilot project, the following elements are mandatory in particular: - number of observations, - number of unknowns (total, i.e. including satellite and other parameters, because this affects the rms-of-fit), - sampling interval (for LAGEOS: 120 sec), - square sum of residuals, - number of degrees of freedom, and - variance factor.
15. SOLUTION/ESTIMATE mandatory
For the pilot project, all position solutions should refer to 1 single epoch (which is defined in the invitation of the project). The solutions should be loosely constrained, i.e. the constraint parameter should have a value "2" (again: let your software decide on this, and check and correct this if different from "2"). The standard deviations are as they were used in the computations, i.e. without scaling with a variance factor.
16. SOLUTION/APRIORI mandatory
See notes on SOLUTION/ESTIMATE
17. SOLUTION/MATRIX_ESTIMATE mandatory
Must be consistent with SOLUTION/ESTIMATE. So, do not scale the uncertainties with a variance factor. No preference for COVA, CORR or INFO representation.
18. SOLUTION/MATRIX_APRIORI mandatory
See notes on SOLUTION/MATRIX_ESTIMATE
19. Footer line mandatory
A general remark on epochs in the SINEX files: it is advised that epoch values are rounded off to the nearest integer seconds.
At this point the Pilot Program will take a hard nosed approach to the submissions. Solutions that do not follow the SINEX format or the ILRS guidelines will be rejected for comparisons and combinations. It is the responsibility of the individual data analysts to make sure that the contents of the file is fully compliant. The people doing the comparisons and the combinations cannot be burdened for (nor will they have the time to) correcting format and/or contents problems.
In principle, all of the analysts are capable of providing their pilot project solutions with the required loose constraints so that the original normal equation (based on the observational data only) can be completely reconstructed. The current set of pilot project parameters (epoch station coordinates plus time-series of x- and y-pole and UT1-UTC) should not present any problem.
4.4. 2-day EOPs vs. 3-day EOPs
This question of "what can the ILRS realistically provide?" proved difficult to answer because little information on the exact quality of either series of parameter solutions (including the station coordinates) was available: Biancale, Devoti and Otsubo had comparison results, but not in sufficient detail. It was recognized that a 3-day time-series will lead to more stable solutions for EOPs, but that a 2-day EOP series may better follow the physical behavior of the (sometimes very irregular) rotation of the Earth. The question was put in a different perspective during the discussion on the next phase of the project "pos+eop" (see notes on agenda item 5).
5. Future of ILRS pilot project "POS+EOP" 5.1. Multi-sat solutions? 5.2. Time-series? 5.1. Combinations? 5.2. Operational analyses? 5.3. Official ILRS product(s)?
Considerable progress has been made on the pilot project during the past several months. To take the project one step further, we agreed to augment the current observational data set by adding LAGEOS-2 data. It is expected that the time-series of EOPs will benefit in particular from this, since it will allow a decorrelation of the behavior of the ascending node of the satellite and the UT/LOD component of the earth rotation. The alternative for the next step (extending the time interval for LAGEOS-1 to perhaps 6 12 months) would probably improve the means to assess the internal quality or consistency of individual solutions, but it would prevent a judgement on the absolute quality of the laser solutions.
It was agreed to continue the pilot project and distribute an invitation to participate in the next phase with both LAGEOS-1 and LAGEOS-2 for the same September 5 - October 4, 1999 period. Noomen will provide a data set with LAGEOS-2 observations. The minimum contribution of analysis centers would be a combined LAGEOS-1 and -2 solution. Separate LAGEOS-1 and/or LAGEOS-2 solutions are also welcome. Since the data volume is more or less doubled, new EOP solutions are requested for only 1-day intervals. If deemed necessary, these can be reduced to 2-day or 3-day solutions.
Noomen presented his findings on interviews with analysts involved in making combination solutions: Altamimi (IGN), Ferland (NRCan), Herring (MIT), Ray (USNO) and Springer (AIUB) (Appendix 13). Their solution parameters include station coordinates and velocities, EOPs, orbits, troposphere and ionosphere parameters, etc. In general, they responded very positively to combined parameter products, since this will generally improve the reliability of the solutions. Specific questions, answers and suggestions on a number of elements of such combinations (editing, weighting, frequency, constraints and systematics) are summarized in Appendix 13.
As a result, we agreed to invite contributions of comparisons and/or combination of the individual solutions to the "POS+EOP" pilot project. A number of ILRS (Associated) Analysis Centers have put forward suggestions for such activities and it is expected that a number of candidates will participate. Nurutdinov mentioned that one IGS analysis center is tasked with the computation of the official combination product, whereas two other institutes serve as back-up.
6. ILRS pilot project "harmonization of QC results" 6.1. Status report 6.2. Future In his report on the "harmonization of QC results" (Appendix 14), Husson showed the discrepancies between the range and time biases as reported by the various real-time analysis centers (CRL, CSR, DEOS, MCC) which may be quite significant. Actions are under way to better understand the reasons for these differences, which may include models for station coordinates, velocities, eccentricities and center-of-mass offsets applied. The project will continue on a steady basis.
7. ILRS pilot project "orbits" 7.1. Definition 7.2. Future
This project was proposed in Frankfurt in January 2000, but no formal invitation has been released yet. Eanes agreed to be the coordinator.
8. ILRS pilot project "software benchmarking" 8.1. Definition 8.2. Future
Like the previous project, this one was identified in Frankfurt as very useful to improving the quality and reliability of the analysis results. Husson agreed to be the coordinator, but no official start has been designated. He did show some first ideas concerning this project (Appendix 15). It was agreed to put this one "on hold" for a few months to give the analysts time to work on the "POS+EOP" project.
9. Miscellaneous
9.1. Data screening
During the EGS General Assembly in Nice, Andersen (NDRE) raised the question whether the ILRS could provide data sets with screened observations on some satellites. The general reaction at the AWG meeting in Delft was that this is not very realistic, since all analysts will apply different screening procedures, and blunt errors (at the level of meters) are easily detectable.
9.2. Data flow
Noomen has been in contact with Moore (NCL) about a tool for checking the contents and integrity of SINEX files in the beginning of 2000. This action will be continued.
9.3. Next meeting
The next meeting of the AWG will be scheduled in conjunction with the Twelfth International Workshop on Laser Ranging, which will take place in Matera on October 16-20, 2000. Since it is unlikely that the analysts will want to be diverted from the Workshop, the AWG will meet on the Thursday and Friday prior to the Workshop.
10. Closure
Noomen summarized the action items coming out of this workshop (Appendix 16). He thanked the attendees for their participation, their input to the discussions and the contributions that they have made to the ILRS pilot projects.
Appendices:
1. Agenda 2.List of participants 3.Introduction, status report pilot project "POS+EOP" (Noomen) 4.EOP+network solution ASI (Devoti) 5.EOP+network solution BKG (Mareyen) 6.EOP+network solution CRL (Otsubo) 7.EOP+network solution DEOS (Noomen) 8.EOP+network solution GRGS (Biancale) 9.EOP+network solution NERC (Otsubo) 10.EOP+network comparison ASI (Devoti) 11.SINEX Format and Data Integrity (slide show, .ppt file) (Husson) 12.EOP+network comparison NCL (Nurutdinov) 13.Comparison/combination of parameters: result of inventory (Noomen) 14.QC harmonization: status report (Husson) 15.Pilot project "software benchmarking" (Husson) 16.ILRS AWG action items
Editor's Note: If you would like to receive a hardcopy of any of the appendices, please contact R. Noomen at ron.noomen@lr.tudelft.nl .
1. Opening
2.Actions since Frankfurt
3.Minutes Frankfurt
4.ILRS pilot project "pos+eop"
4.1.Individual contributions
4.2.Comparisons
4.3.Issues (procedures, format, constraints, parameters, .....)
4.4.2d EOPs vs. 3d EOPs
5.Future of ILRS pilot project "POS+EOP"
5.1.Multi-sat solutions
5.2.Time-series
5.3.Combinations
5.4.Operational analyses
5.5.Official ILRS product(s)
6.ILRS pilot project "harmonization of QC results"
6.1.Status report
6.2.Future
7.ILRS pilot project "orbits"
7.1.Definition
7.2.Future
8.ILRS pilot project "software benchmarking"
8.1.Definition
8.2.Future
9.Miscellaneous
9.1.Data screening
9.2.Data flow
9.3.Next meeting
10.Closure
Appendix 2: List of participants
Richard Biancale, CNES/GRGS (richard.biancale@cnes.fr)
Roberto Devoti, Telespazio (roberto@caronte.mt.asi.it)
Van Husson, HTSI (van.husson@honeywell-tsi.com)
Maria Mareyen, BKG (mamy2@ifag.de)
Horst Mueller, DGFI (horst.mueller@dgfi.badw.de)
Ron Noomen, DEOS (ron.noomen@deos.tudelft.nl)
Konstantyn Nurutdinov, Newcastle University
(konstantin.nurutdinov@ncl.ac.uk)
Toshimichi Otsubo, CRL (otsubo@crl.go.jp)
Mike Pearlman, CfA (mpearlman@cfa.harvard.edu)
Remko Scharroo, TU Delft (remko@deos.tudelft.nl)
Appendix 16: ILRS AWG action items
- Noomengenerate a dataset with LAGEOS-2 observations for the period September 5 - October 4, 1999.
- Noomen generate and distribute an invitation to participate in the next analysis step of the pilot project "pos+eop".
- Noomen generate and distribute an invitation to participate in the comparison/combination step of the pilot project "pos+eop".
- Noomen clear up uncertainties in the SINEX format description and interpretation (coordinate with Herring, SINEX format coordinator). Update of official description?
- Husson report the analysts on SINEX format integrity problems.
- Central Bureau provide an internet file with eccentricities (NEU and XYZ), DOMES numbers, SODs and occupation intervals. The URL will be ftp://cddis.gsfc.nasa.gov/slrocc/slrecc.txt.
- Central Bureau provide internet access to station site logs.
Noomen provide minutes of the Delft AWG meeting. - Husson put minutes of Florence AWG meeting on the internet.
- Husson put minutes of Frankfurt and Delft AWG meetings onn the internet. Noomen+Husson put a description of the ILRS pilot project(s) on the internet.
- Noomen mail eccentricity, DOMES, SOD and period information to Husson.
- Husson obtain information on station coordinates and related information from "real time" analysis centers.
- Husson develop prototype plan for pilot project "software benchmarking".
