Sixth ILRS AWG Meeting (Nice 2002)

Minutes ILRS/AWG Workshop
29 April 2002
Nice, France

9:05 AM  1. Opening

Welcome by Noomen. Thanks to ILRS/CB for paying expenses of room. Approval of agenda (Appendix 1). The names and e-mail addresses of the attendants are mentioned in Appendix 2.

9:15 AM  2. Minutes from Toulouse meeting

More of a status report due to "poor" attendance at Toulouse. In Toulouse, it was decided (and later approved by the MWG and GB) to extend the Etalon campaign for another 6 months. Little accomplished on other pilot projects.

9:20 AM  3. Actions since AWG Toulouse

Review of ILRS/AWG membership list (cf. ILRS web page).

9:25 AM  Review of Toulouse action items.

a. There was a discussion with Nurutdinov about his "auto" checker for SINEX submittals. Most "error" messages are clear, some are not; Nurutdinov will supply help, if needed. "Help" files will be expanded by 15 May 2002 (action Nurutdinov). Compressed (using the "compress" utility) files are accepted at this moment, but in the future "gzip" files only will be accepted. The restriction of 3 megabytes on file size is applied. The receipe for new users: use any browser to visit web site: http://ucscgi2.ncl.ac.uk/~nkn3/. A user must "register" on-line first. Later, the user will receive an e-mail from Nurutdinov with username and password to login for creating a personal account. After creating the account, the user will be able to submit his SINEX file using forms in the browser. The file will be checked immediately for SINEX format requirements and the results could be seen in the browser's window. BKG and Geosciences Australia have used the site successfully. All analysts are encouraged to use this tool. Husson is also looking into SINEX "checkers". Angermann is doing similar work for ITRF submittals and that system will be expanded to other parameters in the future (action Angermann).

b. There was a discussion about the Mendes mapping function, provided to the ILRS community by Pavlis. This new mapping function cannot be used in connection with the Marini-Murray model, since zenith delay and mapping are interwoven in the latter. Pavlis and colleagues have worked with the Saastamoinen model for zenith delay instead, and have found a similar accuracy for the two models (Saastamoinen and Marini-Murray) at zenith. At lower elevations, the Mendes/Saastamoinen combination is reported to perform better than the Marini-Murray model. The tropospheric delay proves to be relatively insensitive for wavelength (which is not taken into account in the Mendes model). The (zenith) correction itself is a function of wavelength, of course. The "conventions" are still not "frozen", and it appears too early to settle this. Some of the ILRS analysts are already using the new mapping function for the pilot projects (and elsewhere), with documentation.

c. Shelus' action item on the "new" differences between Analysis Centers and Associate Analysis Centers is frozen, pending redefinition of the products and procedures of the AWG.

10:15 AM  4. General Information

Noomen showed a diagram of the new IERS structure. He emphasized that only one unique, coordinated product is expected of the ILRS by the IERS for the operational products (such as Bulletin A, ITRF), rather than the present multi-submittals by the various AWG Analysis Groups. For other IERS activities, individual contributions are welcome.

Noomen presented a report on the new IAG structure. According to the Beutler presentation at Paris (IERS GB meeting April 20), there will be 4 commissions and 9 services, of which ILRS is one. The new IAG organization had been approved by the IERS GB in Paris.

IAG is organizing various projects, especially IGGOS. Noomen provided a summary of this project, which had been described by Beutler at the IERS GB meeting in Paris. The services (such as ILRS) are invited to participate in the organization of IGGOS; the pilot phase is to begin in 2003. Drewes had also described this at the ILRS Governing Board meeting during the week prior to this AWG meeting (where it was decided that Pearlman would participate in the organization phase of IGGOS).

Noomen described the SINEX Project that is soon to begin under Rothacher, to foster the interchange of solutions, for the production of combined solutions. A Call For Participation (CFP) will be distributed in mid-May. This is an implementation project, not a product unto itself (although that might change in the future). Rothacher will accept any submission, as opposed to a "blessed" ILRS solution. No constraints on the period covered by the individual analyses. A report will be presented at an IERS "Combination Research Workshop" in November (probably in Munich).

Noomen presented on another IERS activity: the "EOP Alignment Campaign". This aims at a harmonization of the EOP solutions, with ITRF2000 as the terrestrial reference. This also is based on a request from Rothacher.

Shelus reviewed the issue of ILRS Station Qualification. Main points currently are: (Asscociate) Analysis Centers involved in semi real-time quality control computations should deal with all LAGEOS data in order to assess the quality/quantity of observations coming from each possible station. This information will be used by the AWG to categorize the global network of stations into Core/Contributing/Associate, probably on a 6-month basis. The results will be made available through official ILRS channells, and ILRS (Associate) Analysis Centers can use this as a pre-screening filter for their computations. To avoid duplication of activities by the analysts, the task of this initial screening is best given to the data centers: they are to put data from Core or Contributing stations in one file (or directory), whereas the observations from Associate Stations should go somewhere else (action Data Centers); The regular judgement of the AWG on all stations and data will follow the recommendations from the International Workshop on Laser Ranging (IWLR) in Shanghai, which were judged by the AWG (repeatedly) as "reasonable"; The "schedule" for station qualification revision will be according to the AWG meeting schedule. The wording is to be finalized, and then the report itself should be circulated for comment to the AWG one more time before final acceptance and recommendation to the GB.

Appleby reported on the IVS/IGS/ILRS Working Group. A draft report had been written. It appeared unlikely that IVS data would contribute to the question of the location of the L1 and L2 phase centres of GPS satellites. The ILRS is capable to calibrate the radial accuracy of the microwave-derived orbits, but a full assessment of the center-of-mass to solve the problem appears not to be possible.

Noomen reported on several published papers in "Surveys in Geophysics" for November, 2001, and an invited paper at the AGU Spring meeting on the role of ILRS within IGGOS.

The next ILRS AWG Meeting will be in conjunction with the next International Workshop on Laser Ranging (Washington DC, October 7-11): October 3-4 at Raytheon or HTSI facilities.

Announcement: We have been informed of EU project FP6. ILRS can possibly contribute here (i.e. obtain funding); Ivan Prochaska will distrribute an announcement shortly. Deadline is June 7. EU requires a "big" project: nearly 50 participants are needed, and multi-countries (12-15). What can the AWG do? Pavlis fears that we are too small. Pearlman says that the ILRS can contribute 10 people from the USA. Chances of success are considered slim.

5. SINEX format

The ILRS AWG has been working with SINEX 1.0 for a while now. In doing so, we have had to add new options and make clarifications to properly understand the format in an unambiguous way. A second release SINEX 2.0 is under development now and we (ILRS/AWG) have provided input. Almost all of our recommendations have been accepted. The Statistics Block is subject to a different (general) viewpoint: it will probably be "highly recommended" instead of "required" (for arbitrary users); for the ILRS community (as agreed in earlier AWG meetings) it is mandatory. Also, the new format will include the option to contain full normal equations (i.e. including matrix and right-hand side).

6. Pilot project on "harmonization"

Husson presented the status and development of this pilot project (appendix 3). He gave an overview of the various quality control analysis (QC) activities. He noted that most analysts have switched to ITRF2000 coordinates. However, some Analysis Centers (ACs) still use very old coordinate solutions. Long term range bias trends from different ACs for the higher performing sites tend to track nicely, but absolute range biases may differ substantially. Range bias stabilities vary substantially between MCC and CSR. For most sites, MCC range bias stabilties are at the 10 mm level, while CSR stability results are typically between 10-20 mm. MCC appears to be using a different convention for the sign (i.e. +/-) of a range bias when a range bias is applied apriori (action: Husson: contact MCC). The use of different site coordinate appears to be the dominant reason for the differences in absolute range biases and range bias stabilities. This was agreed by all.

Husson pleas for the "quick-look" ACs to update their coordinate solutions to ITRF2000 as soon as possible. Certain site positions in ITRF2000 will require updating (e.g. Arequipa post-quake, Hartebeesthoek, Riyadh). Currrently, only CSR LAGEOS range bias (RB) results are used in the performance report card. Husson will start to incorporate other LAGEOS RB results from other ACs into the global performance report card, while at the same time investigate adding QC results of other satellites which have precise orbits (i.e. 1-3 cm RMS).  This will help ensure a more complete and realistic station characterization. In other words, Husson will develop a strategy for a unique bias report ("scorecard"), so stations get an unambiguous message on their performance (action Husson). All the above mentioned issues are important, since report card metrics will be used in the official ILRS station classification.

7. Pilot project on "benchmarking"

This project is led by Husson, Mareyen and Pavlis.

The goal of this pilot project is to benchmark the software that is in use by the various analysis groups, in particular for orbit determination and estimation of (geodetic) parameters (Appendix 4). The future requirement is very simple: anyone who submits a product to be combined must pass the software benchmark requirement. The benchmark test requires agreement on numerical results (within realistic limits), which can be achieved by minimizing the sources of variations (i.e. prescribe the input for the computations).

Husson showed a list of elements of the computation model. This was reviewed by the AWG members present (Appendix 5) and will serve as a draft for the standard computation model to be used for this benchmark test. It will be circulated within the community for general approval (action Husson).

The following cases are defined for the benchmarking process:

Orbit and station coordinates/EOPs fixed (-> orbit description + residuals)

Orbit solved for, station coordinates/EOPs fixed (-> orbit description + residuals)

Orbit and station coordinates/EOPs solved for (-> orbit description + geodetic parameters + residuals)

Output products:

  EOP and Station positions (in SINEX format)

  Normal point residuals (in text file)

  Orbital parameters (in text file); true-of-date satellite x,y,z,xdot,ydot,zdot at 2-min

intervals

Criteria to pass the benchmark may include the rms-of-fit, the recovered solutions for the station coordinates and the EOPs, the orbital parameters and the site residuals.

The documentation of the models will have to be finished on May 15. The invitation to participate in this project will be released on May 17. The first iteration will have to be performed by July 1. The initial results will be reported during the October AWG meeting in Washington (action Husson/Pavlis/Mareyen).

Because of these developments, the "banchmarking" and "orbits" pilot projects partly overlap.

8. Pilot project "orbits"

No report from Eanes.

9. Pilot project "positioning + earth orientation"

There is a need for a unique, combined EOP (and coordinates) product from the services, i.e., from ILRS, requested by IERS. To this aim, the ILRS AWG has initiated this pilot project a few years ago. At this moment, it has developed into a series of test solutions: a series A (LAGEOS-1 only, entire 1999) and a series B (various combinations of LAGEOS and Etalon satellites, various sets of EOP solutions; April 2001 - April 2002; the latter date has been extended to October 2002 after the request of the AWG coming out of this meeting). The options for the series B solutions are summarized below:

: LAGEOS: LAGEOS + Etalon

EOP: B1: B2

EOP + EOPDOT  B3: B4

In addition to improvements in the EOP products, the Etalon observations are also expected to improve the determination of GM and the characterization of SLR data.

9.1. Individual contributions

Individual presentations were given on the analyses as performed by the various institutes:

ASI

(Appendix 6) The series A solutions were affected by initial errors in the analysis (state-vector values). The solutions yielded clear correlations between the formal uncertainty of the UT component and the cross-track accelerations; this is not visible in the LOD product. As for the series B solutions: ASI contributed to all four possible combinations. In the initial combination, equal weights for all solutions were used. Afterwards, an optimal weighting scheme was developed.

BKG

(Appendix 7) BKG produced B1 and B2 solutions only, because of time and other constraints (crash workstation, UNIX to LINUX reconfiguration).

Mareyen reported on singularities in the normal equations (typically, the 3 smallest eigenvalues are near zero, whereas the 4^th value may have a significant non-zero value already). A variety of reasons was mentioned. The UT1 elements can cause a considerable bias in the solution, so a way out can be to fix one UT parameter at a good a priori value. Also, can LAGEOS-3 break the singularity?

The data distribution of Etalon appears inhomogeneous: there are many days without Etalon data. This may lead to EOP problems. Also, Etalon data is noisier, which should have effects on the weighting factor, further reducing the importance of the Etalon data. Overall conclusion: there are only problems and no improvements. Etalon is not adding more geometrical information. Devoti remarked that one may gain in gravity field information, in the sense that it may help in separating the low degree part from the high degree part (further analysis needed).

Geosciences Australia

(Appendix 8) Review of software, computation model and (treatment of) data sets. The A solutions yield a fit w.r.t. ITRF2000 of about 15 mm in each coordinate, and about 0.7 mas and 1 ms for x/y-pole and UT, respectively.

The series B solutions showed similar statistics: about 15 mm coordinates scatter for the B1 and B2 solutions. The addition of the Etalon data resulted in an improvement of the EOPs of less than 10%. The solutions B3 and B4 will also be generated in the (near) future.

Overall conclusion: LAGEOS dominates the solution, because of the difference in data volume. The low data volume of Etalon is suspected to be the main reason for the lack of significant EOP improvement.

CRL

(Appendix 9) CRL contributed all A and B solutions. Review of procedures, models and configurations. CRL applied a 1 to 9 weighting for Etalon (to compensate for the low data volume; completely opposite to what was done by BKG). The addition of Etalon data does not change the rms-of-fit; adding the EOPdots does. Procedures and results: show little difference after adding Etalon data. Otsubo makes a few final remarks on the (quality of) results expected from the IVS community in the near future. Can ILRS still play a role in EOPs?

DGFI

(Appendix 10) A remark was made about an error in the covariance matrix in the SINEX files (wrong scaling factor); needs to be re-submitted. Report of current activity on SLR data processing, covering time-span 1981-2002: time evolution of SLR reference frame (scale, origin), weekly station coordinates, low-degree gravity field coefficients, EOPs. Scale coming from SLR is more stable than scale coming from GPS; the same holds for geocenter. The repeatability of the weekly station coordinates is 1-2 cm (after Helmert transformation). The results have been obtained with a free network solution; a no-net-rotation and 1 EOP constraint was applied.

No Pilot Project B submittals since no experience with modeling Etalon yet. Also, the estimation of EOPDOT is not in the software yet.

JCET

JCET has computed new series A and series B3/B4 solutions (minor changes from last year's solutions, except for introduction of Etalon data). No "rate" solutions as yet. JCET filtered out stations with a 50 NP requirement for both LAGEOS targets on a month by month basis (reducing the network by 10-40%). The good stations in series A show a formal uncertainty of a few mm (no biases are estimated for these stations). W.r.t. ITRF2000, the mean height difference is 1.8 mm, whereas the scatter in height is 5.6 mm.

As for the series B solutions, JCET has done an extensive test with the relative data weights of Etalon vs. LAGEOS (gradually increasing the absolute importance of Etalon). The outcome is not satisfactory yet.

NERC

Appleby gives an update on his experiences with the B4 solutions. NERC used only the best stations and did not estimate a range bias (because of the correlation with station height). Noisy EOPDOT's maybe due to reduction of data amounts. NERC also tried to compensate for the low Etalon data volume: a single Etalon observation is worth 2 times as much as a single LAGEOS observation. Not yet happy with solutions.

CSR

On behalf of Eanes, Husson reported on the analyses done at CSR. These cover a time span of 1993-2002 (monthly solutions). Annual and semiannual signals are visible in x and y (plus/minus 5mm) and z (-5 - +15 mm).

9.2. Comparisons and combinations

ASI

(Appendix 11) Luceri showed that the total amount of Etalon data has about doubled, but that the effects are only marginal on EOPs (improvements are about 5% in any of the EOP components, irrespective of the analysis institute). Typical fit values are about 17 mm for LAGEOS and 13 mm for Etalon. The results (difference w.r.t. C04) get worse when EOPDOTs are a part of the solution. Use of a DOP factor (Dilution Of Precision) analysis suggests that the improvement in results is (almost completely) due to the increase of total number of observations. Simulating more Etalon data (70%; all possible observations) leads to a maximum improvement of the results by about 20%. The conclusion is that Etalon is not bringing in any "new" geometry (although the DOP factors are related to the statistical aspect of observations only, not the actual quality). A study for the optimal data weights of the various satellites resulted in values of 1.0, 0.2, 0.8 and 0.7 for the LAGEOS-1/2 and Etalon-1/2 satellites, respectively.

(Appendix 12) Devoti presented statistics of all individual solutions w.r.t. ITRF2000 (series A solutions). Depending on the solution, the scale and geocenter values may be at the level of 1-2 ppb and a few tens of mm, respectively. The coordinates residual scatter ranges from 8 mm (CSR) to up to 20 mm (Geosciences Australia). The NERC solution "spikes" in April, whereas the others are "smooth". Devoti uses a so-called "looseness" parameter to assess the level of constraints on the solution. This parameter should be around 1. Very low and very high numbers are suspect. Note the anomalous CSR number for this parameter (<<1 for the rotation parameters); there are some comments in the SINEX file for CSR (action Devoti/Eanes).

ASI has computed a combination solution without the estimation of Helmert parameters, since the solutions already contain a large uncertainty (after "loosening"). In this combination, a covariance scaling factor is applied (may vary from month to month). Combination solution is good (about 8 mm wrms difference w.r.t. ITRF2000, and 0.15 mas / 0.03 ms for pole and UT, respectively), except for CSR results, because it is "too close" to ITRF. The combined EOP solution "fits" C04 better than the individual ones (for all solutions, except CSR). The averaging/combination itself looks OK, but the derivation of proper statistical properties of the combination product is still quite uncertain (all solutions are to a high degree correlated).

One of the attendees raised the question on the best product: a combined product from many solutions, or the "best" individual one.

Minutes ILRS/AWG Workshop

30 April 2002, Nice, FRANCE

09:05 AM  Comparisons and combinations (continued)

DGFI

(Appendix 13) Kelm used the A solutions only. There was a short review of the solutions actually used.

The solutions are processed at 3 different levels. Level 1 encompasses a normal equation analysis (recover NEQs, subtract loose constraints and check eigenvalues of results). In theory, the 4 smallest eigenvalues should be zero because of a rotational rank defect of 3 for the coordinates and an offset rank defect of 1 for UT. This could be numerically verified in most cases. But large negative eigenvalues are observed for CSR; too many eigenvalues near zero for JCET (probably due to the additional EOPDOT content, which should not be included in the series A solutions); IAAK had (too many) negative eigenvalues as well.

Level 2 represents a minimal constraint solution analysis: 3 rotational sum constraints are applied for good sites, and a no-net-constraint is applied for good sites, and one UT value in the middle of the interval is fixed. The covariance matrices are scaled with their relative weights (yielding the same accuracy level for the best sites). Checks are the EOPs w.r.t. each other, and coordinates w.r.t. each other. Best fit, using the 7 "best" observing sites, obtained by the following analysis groups: CRL, DGFI (for pole); DGFI, NERC (for UT); CRL, DGFI (for coordinates).

Level 3 represents a combined solution analysis. Here, the NEQs are combined pairwise to check the eigenvalues again. The eigenvalues for a few combinations proved OK only; best by NERC/DGFI. This needs further study.

Conclusion: A combination of solutions is not yet justified because of "best fit deficiencies" (for the "A" solutions, anyway).

Finally, Kelm presented a number of recommendations to study/solve the problem further.

A discussion was beginning to develop on a requirement for a minimum number of observations for inclusion of a station in the solutions. Also, a link between drift in UT and 1cpr cross-track accelerations was raised, but this is not very interesting since UT is by definition not "solvable" by satellite techniques.

09:55 AM  JCET

(Appendix 14) Pavlis provided two sets of solutions: one including all possible stations, and one including only stations that measure "continuously" (i.e. 14 stations). Improvements of the statistics are seen in all aspects, when using the "continuous" subset instead of the total set: geocenter, scale, rms-of-fit w.r.t. ITRF2000.

 

JCET also investigated the relative weighting between LAGEOS and Etalon: can we get a proper relative weighting using a "Lerch-like" analysis? It does not seem to work. Are we averaging over too short an interval?

10:05 AM  HTSI

(Appendix 15). Husson basically showed his San Francisco presentation complimented by analysis of new solutions that were submitted by JCET and ASI. His presentation focused on the time series analysis of biases, heights, and baselines from the best 8 sites in the "A" solutions (i.e. 1999).

Husson showed the effect of modeling errors on vertical site positions and scale. He showed the relationship between bias estimation (errors) and its effect on site heights and baseline lengths. He asked the question "can range bias be separated from height and if yes what are the limiting factors?".

Husson reported on a detailed performace analysis (data quantity and data quality) of the thirteen 28-day periods in the 1999 dataset. He recommends data with known but irrecoverable bias problems (i.e. scale errors, barometer errors, epoch errors, unstable biases, etc.) be deleted. Husson will provide a bias correction table in SINEX format, for any fixable (i.e. based on engineering information, not on orbital analysis) bias problems in 1999 (action: Husson). After the bias problems in 1999 are fixed, Husson will incorporate more recent bias correction models and then work his way back into the past.

There are still a few lingering SINEX issues. For example, a few solutions have some round-off errors in apriori ITRF2000 site positions when propagated forward.

The data analysts have signficant differences in their treatment of data (e.g. biases, data weighting, minimum number of normal points, arc lengths), but these differences are slowly narrowing. He showed one example from the DGFI solution, the consequences of erroneously estimating an epoch and scale bias.

Husson suggested that each center perform some "reality" checks on their own solution prior to submission and provided a recommended set of criteria. The reality checks and their criteria would be > 5 mm range bias differences between LAGEOS-1 and LAGEOS-2 for a given site or set of sites, > 20 mm height peak-to-peak variations for a given site, and > 20 mm baseline lenghts peak-to-peak variations. Clearly, if either one of these values exceeds the limitations specified here, it indicates the presence of (modeling) errors.

Most centers who estimated biases every 28 days had considerable difficulty separating range bias from height. CSR was the most successful in separating these 2 quantities. In general, it appears that 28 days may not contain enough data to adequately solve for sub-centimeter level range biases. Husson showed that if you average the site-by-site range bias estimates over an entire year, then mm level range bias aggreement per site per analyst center is possible. The results from CSR, CRL, and ASI were given as supporting evidence.

In conclusion, Husson's final recommendations were to be cautious in the estimation of biases and to use any "known" apriori biases. Each AC should perform its own data integrity checks prior to submission. If a solution does not pass certain pre-defined "reality" checks, then it may not warrant to be used in the combination solution.

A few discussions were triggered from his presentation. First, the advantages and disadvantages of using just the core sites versus the full network were debated. Second, do the SLR sites really have range biases; how stable are they; and to what level do we know the answers. If there are "real" biases, how should the analysts treat them.

Dunn's opinion is that solving for biases extracts "signal" from the data. If you think there is a bias in a station, contact the station to see if it can be verified. Please note the correlation between range bias estimates and height in Husson's presentation.

Pearlman's opinion is that analysts may be spending too much time trying to extract information from the poorer stations.

Noomen's opinion is that ignoring biases (with whatever background) is unrealistic in a real world environment where modeling is never "perfect".

Pavlis's opinion is that for real-time data products, use the better stations only. First step is to have some minimum data requirements and define the consistent systems.

Otsubo's opinion is you need to estimate biases, because the LAGEOS center-of-mass correction is site dependent at the few to several mm level.

11:10 AM  NCL

Nurutdinov tried to analyze the B3 solutions and make a combination solution. He noticed a significant number of (sometimes marginal) problems that have to be corrected. A file of combined solutions is available, but not yet in a presentable form. Most results were reported orally or with hand-written overhead sheets.

11:20 AM  Break

11:45 AM  9.3. Issues to be considered

Action item on Shelus to review IAU Resolutions and the IERS reactions to same. He will circulate an e-mail to identify the appropriate web-site for AWG members to review.

The 2^nd half of the morning (the last part of the workshop) continued with a discussion on a number of fundamental issues for the ILRS AWG, the pilot project "positioning and earth orientation" in particular: Etalon data, EOPs, date volume requirements and biases.

11:50 AM  Continuation of an enhanced Etalon campaign?

This issue was introduced with a number of questions. Is a 5% gain sufficient to continue? Or should we continue to work with this one year of data that exists now for future study? What should we do in terms of target definition? What are the trade-offs? What about "continuous" Etalon observations? What effects does Etalon have on "gravity"? Can the Etalons help to separate J₂ and J₄? Is there an alternative to Etalon?

In the discussion, is was brought up (again; the same thing was reported a year ago in Nice) that there are other products that can be generated with increased Etalon data, not just EOPs (but also GM, station characterization, maybe gravity field terms or temporal variations therein).

Recommendation: Ask the IERS MWG and GB to extend the Etalon campaign to October 2002, and then (at the AWG workshop in Washington) say what we have with EOP and beyond (action Noomen).

12:30 PM  EOP and EOPDOT, or EOP only?

Similar to the above, have we really looked at this? Can we make this decision at this time? UT is generally recognized as a nuisance product. LOD, and x/y-pole are the prime EOP products from ILRS. What about x-dot and y-dot?

The consensus seems to be that, for the present, we should precede to UT (as a nuisance parameter only), x, y, LOD, xdot and ydot. Please do not interpret the UT solutions. Do not constrain UT if you solve for it but do not include it in the SINEX files as a product.

12:50 PM  "CORE" Network Guidelines?

If stations have poor data quality or quantity, singularities will be introduced into solutions. Keeping the geometry of the stations fixed, the consistency of the products will be improved. How can we apply this for specific ILRS products?

One proposal from Pavlis was to require a minimum of 50 normal points per LAGEOS target through a month as a minimum to include that station in a solution. Where did the number "50" come from? Is this too simplistic? What about number of normal points and passes? This speaks to quantity. What about quality?

After discussion, the idea behind such a pragmatic approach was honored, and it was concluded that a 30 normal point minimum for both LAGEOS targets per 28 day period is required for the station to be included in a solution; if this is not achieved, the data of this station should be deleted altogether (for this analysis).

In addition, it was agreed that for the current project the ILRS analysts should only work with observations taken by Core or Contributing Stations. This categorization is based on station performance over the past year.

01:10 PM  Biases?

The issue of "biases" pervaded through many of the presentations and discussions of the past 2 days (e.g. in the presentation on "comparison/combination" by Husson/HTSI). A suggestion is to set up a set of known biases to be used (action Husson). Pavlis made the suggestion to include the partials for range biases in the (internal) normal equation, even if the bias is zero (to save processing time in case new insights develop, or new material becomes available).

CSR/SLR range bias solutions coming from Eanes' quick-look analysis have to be examined since they may be affected by an old set of station coordinates significantly.

A comment was made by Dunn that by definition SLR observations do not have any biases. That is to say: every element of the observation can be modeled (or calibrated) perfectly, and what is called "biases" is nothing more than "apparent biases" because of modeling deficiencies. Systems have to be examined (and then calibrated) to the mm level, including the corresponding satellite signatures. Most stations have never been examined to that detail before.

It was decided to set up a situation such that we will assume that there are no biases for either core or contributing stations when generating our ILRS standard products. There was a big discussion with respect to satellite center of mass corrections that are station dependent at the level of about 5 mm. We will all work with the same a priori biases from the engineering group, as an initial condition.

01:35 PM  Combination of EOP Solutions?

This item on the "issues" list triggered questions only: How should we organize ourselves to produce an ILRS combination product? How can we move to this? Daily EOP solutions are a reasonable target, in spite of what GPS is doing. This is going to be a (semi) real-time product for IERS Bulletin A. In the current form of the pilot project, the solutions are generated on a 28 day basis. Clearly, this will be of no use to IERS Bulletin A. One option is to continue the 28 day solutions to keep this strength and then put it in a moving window, moving every 7 days. Another question is who will do the official combination?

01:45 PM  Call for Participation for EOP Combination Center

Noomen brought forward the question of making the next step towards an official ILRS combination product, i.e. releasing a Call for Participation, and opening this for institutes outside of the ILRS. This triggered a lot of discussion "against".

The main argument for going outside the ILRS is that experience can be "bought". A secondary argument is that the time available for most analysts within the ILRS is limited (although by no way ILRS groups would be excluded from such a function, of course).

Arguments against were that "outsiders" are not familiar with the specific elements of the SLR technique, which might possibly harm the (quality of the) ILRS products.

Since the discussion was not developing towards a clear concensus, it was decided to start by making a draft version for this call for participation, circulate it for tuning, and discuss and finalize it at the AWG meeting in Washington (action Noomen). The decision to go outside the ILRS for a combination center is defered for now.

02:05 PM  Other ILRS Products?

Although the emphasis has been on station coordinates and EOPs so far, we should not forget possible other products which are unique or "best" for ILRS.

After the development of an official EOP product, station coordinates is the next most needed product, namely for ITRF submittal. A discussion concerns why this task should be split from the EOP product?

As a conclusion, this aspect shall be included in the Call for Participation and be iterated within the Analysis Working Group.

02:20 PM  Pilot Project

Update the pilot project work with what has gone on at this workshop. Rename to AA and BB.

Noomen thanked the attendents for their contributions.

July 4, 2002

G. Appleby, R. Noomen, P.J. Shelus

Appendices:

1  Agenda

2  List of participants

3  Harmonization (Husson)

4  Benchmarking (Husson, Mareyen, Pavlis)

5  Benchmarking standards (Husson)

6  EOP+network solution ASI (Luceri)

7  EOP+network solution BKG (Mareyen)

8  EOP+network solution Geosciences Australia (Govind)

9  EOP+network solution CRL (Otsubo)

10  EOP+network solution DGFI (Angermann)

11  EOP+network comparison/combination ASI (Luceri)

12  EOP+network comparison/combination ASI (Devoti)

13  EOP+network comparison/combination DGFI (Kelm)

14  EOP+network comparison/combination JCET (Pavlis)

15  EOP+network comparison/combination HTSI (Husson)

16  ILRS AWG action items

Appendix 1: Agenda

ILRS Analysis Working Group workshop #6

Nice, France, April 29-30, 2002

Agenda

opening

minutes AWG Toulouse

actions since AWG Toulouse

reports, presentations, membership

general information

organizations: IERS, IAG

IERS/IAG projects/products: Bulletin A, SINEX, EOP alignment, IGGOS

Station qualification

IVS/IGS/ILRS working group

ILRS related presentations, publications

IERS Combination Research Workshop

Next meeting

SINEX format

parameters

pilot project "harmonization"

6.1. status report

6.2. future

pilot project "benchmarking"

7.1. status report

7.2. future

pilot project "orbits"

8.1 definition

8.2. products. format

8.3. future

pilot project "positioning + earth orientation"

9.1. individual contributions

= ASI

= BKG

= Geosciences Australia

= CRL

= DGFI

= JCET

= NERC

= CSR

9.2. comparisons and combinations

= ASI

= DGFI

= JCET

= HTSI

= NCL

= questions/issues:

- quality of individual station coordinates? EOPs?

- addition of Etalon data?

- weighting

- quality of combined station coordinates? EOPs?

= RFP for official ILRS combination center(s)

9.3. future of "positioning + earth orientation"

= multi-sat?

= time-series?

= combinations?

= operational analyses?

= official ILRS product(s)?

= time line?

action items

closure

Appendix 2: Attendance

• Detlef Angermann: angerman@dgfi.badw.de
• Graham Appleby: graham.appleby@nerc.ac.uk
• Per Helge Andersen:   per-helge.andersen@ffi.no
• François Barlier:   francois.barlier@obs-azur.fr
• Roberto Devoti:   devoti@asi.it
• Peter Dunn:   peter.j.dunn@raytheon.com
• Ramesh Govind:   rameshgovind@auslig.gov.au
• Van Husson: van.husson@honeywell-tsi.com
• Reiner Kelm:   kelm@dgfi.badw.de
• Cinzia Luceri: luceri@asi.it
• Maria Mareyen   mamy@ifag.de
• Ron Noomen: ron.noomen@deos.tudelft.nl
• Konstantin Nurutdinov:  konstantin.nurutdinov@ncl.ac.uk
• Toshi Otsubo: otsubo@crl.go.jp
• Erricos C. Pavlis:   epavlis@umbc.edu
• Mike Pearlman:   mpearlman@cfa.harvard.edu
• Leonid Petrov:   pet@leo.gsfc.nasa.gov
• Peter J. Shelus: pjs@astro.as.utexas.edu

Appendix 5: Benchmarking standards (Husson)

• dynamic model:
  • - gravity field
    • + static component: EGM96
    • + maximum degree and order 20 x 20
    • + time dependent component C(2,0), C(2,1), S(2,1)
  • - GM [m³/s²]:   398600.4415D9
  • - solid earth tides: Wahr model
  • - ocean tides : Ray GOT99.2 or CSR 4.0
  • - atmospheric tides: not applied
  • - direct solar radiation : applied; C_R fixed at 1.13
  • - albedo:   not applied
  • - earth shine: not applied
  • - thermal forces: not applied
  • - dynamic polar motion:   applied (see below)
  • - empirical accelerations:   constant along-track (4-day interval)
  • - third body attraction : Sun, Earth's Moon, Mercury, Venus, Mars, Jupiter, Saturn, Uranus, Neptune from JPL's DE403
  • - relativistic effects:   applied (PPN @ GR)
• reference frame:
  • - station coordinates
    • + a priori values:   ITRF2000
    • + estimated: all
    • + reference point: IAR for stations with zero eccentricity; site marker for others
    • + permanent tide correction:   ITRF convention
  • - station velocities
    • + a priori values:   ITRF2000
    • + estimated: no
  • - earth orientation parameters
    • + a priori values:   IERS C04
    • + frequency [days]:   1
    • + estimated: yes: PM, UT1
  • - precession
    • + model: IAU 1976
    • + corrections: not applied
  • - nutation
    • + model: IAU 1980
    • + corrections:   1996 theory applied
  • - station deformations:
    • + tidal uplift:   Love model (h₂=0.609, l₂=0.0852)
    • + ocean loading :   Scherneck model (Ray GOT99.2 tide)
    • + atmospheric pressure loading  not applied
  • - geocenter
    • + motion : not adjusted
    • + tidal frequencies: implicitly modeled with ocean loading
• model for station deformation
  • - pole tide:   applied
  • - definition of origin:   realized by adjusted stations'coordinates
  • - definition of orientation: ITRF2000 a priori (constrained at 1 m)
• measurement model:
  • - satellite used:   LAGEOS-1
  • - wavelength of observations [nm]   as reported by stations in data records
  • - biases:
    • + range biases:   estimate one bias per site per 28-day (0 m a priori value; 100 m a priori std.dev.)
    • + time bias: not estimated
    • + tropospheric bias: not estimated
  • - data weighting:   all sites 1 meter
  • - center of mass [m]:   0.251
  • - model for tropospheric delay Marini-Murray model
  • - minimum elevation [deg]: 10
  • - relativity scale: LET
  • - speed of light [m/s]: 299792458
  • - relativistic effects:   applied (GR)
• integration:
  • - arc length: 28 days
  • - integrator:   Cowell 11th order or of equivalent quality
  • - step-size [s]:   120
  • - estimated parameters:   station coordinates at epoch; EOPs at 1-day intervals; satellite state-vector; satellite acceleration; 28-day range bias/station
  • - constraints:   1 m on station coordinates and equivalent for EOPs, etc.

Appendix 16: ILRS AWG action items

• Angermann:  extend SINEX format checker for ILRS purposes
• Devoti: get in contact with Eanes on "looseness" of series A solution
• Husson/Pavlis/Mareyen:  work out plans for "benchmarking"
• Husson: develop strategy for unique bias report
• Husson: make available table with LAGEOS data problems 1999
• Husson: contact MCC on sign of applied range biases
• Noomen:  ask IERS for specification ILRS products for IERS purposes
• Noomen:  urge "quick-look" analysis centers to use or introduce ITRF2000
• Noomen:  ask MWG and GB for extension of Etalon campaign
• Noomen:  develop Call for Participation for ILRS combination center
• Nurutdinov:  expand help functions for SINEX format checker at NCL (May 15)
• Nurutdinov:  accept gzip input files in SINEX format checker
• Shelus (new): distinguishment between ILRS ACs and AACs
• Shelus: review IAU Resolutions and the IERS reactions
• Data centers:  split incoming data in "Core+Contributing Stations" and "Associate Stations"