Dear Colleagues, unfortunately I am in Chile these days and did not have access to e-mail last week. Therefore, my comments on various issues raised are late. Let me try to comment on various e-mails now. IERS CPP Workshop: ----------------- Zuheir Altamimi proposes to reduce the workshop to a 1/2 day meeting in San Francisco. In my opinion this would have several disadvantages: - I already know of several people that intend to present their results at the workshop. I am sure that we easily get some 12-15 presentations for the workshop. Assuming about 20 minutes for each presentation and some discussion, we easily fill one day, even without the very necessary in depth discussions about various issues. - If there is no real workshop with presentations, people will not work on the combination topic and we will loose additional time. A meeting is always stimulating work. - I do not agree with Zuheir and Jim that the solutions presently available within the CPP cannot be used. There are already several months of weekly solutions available and we need to know whether they can be used or not, whether there are any problems to be corrected, etc. With several months of weekly solutions a combined solution can be generated. It would ebe very interesting to see how this solution performs ... Therefore, before reducing or even cancelling the workshop, I would like to propose the following: I send together with this e-mail a list of the topics/sessions (see below) that I think important to cover at the workshop and I ask who will contribute presentations. Depending on the contributions we get, we can decide on the time interval for the workshop. I think that there is a lot of interesting material available to discuss ... Proposal for Sessions: --------------------- 1) Combination Pilot Project and CONT’02 ("Weekly" SINEX) 2) Long Time Series and their combination (ITRF, IERS2005, ...) 3) Handling of Systematic Biases 4) Conventions, Modelling, Parameterization 5) Local Ties PLEASE LET ME KNOW AS SOON AS POSSIBLE WITH A SHORT E-MAIL JUST WHETHER AND WHAT TOPIC YOU WILL CONTRIBUTE TO THE WORKSHOP IN SAN FRANCISCO SO THAT WE CAN GET AN ESTIMATE OF HOW MUCH TIME WE NEED FOR THE MEETING. Comments to e-mail of Jim Ray (30.09.2004): ------------------------------------------ > Without the EOP rates, it makes combination with other > SINEX files rather difficult. (As I understand it, the ILRS solutions do > not include EOP rates either, but their reference epoch is always noon, > which does allow them to be combined with other noon solutions.) The IVS > combinations are a promising first step, but I do not see how they can > be used by the CPP at this point. Assuming a piece-wise linear model for the EOPs (offsets and rates, continuous; or offsets with linear interpolation inbetween) the IVS combined solutions can be combined with e.g. the GPS solutions even if no rates are given. The epoch of the EOP is taken into account by correctly weighting its contribution to the offsets (and rates) at the beginning and end of the day. This is not really optimal (I would prefer to include the rates, too), but available information (offsets) can be correctly combined. If the offset is given in the middle of the actual observation interval, offset and rate will be almost uncorrelated, so that the offset alone can be used. It is clear, however, and I fully agree, that -- as soon as possible -- offsets and rates should be included in the SINEX files. > In our opinion, a meaningful multi-technique combination must start with > a long-term combination, preferably at least 5 years but even longer would > be better. Therefore, we asked that the IERS explicitly request the longest > possible span of weekly solutions from all the techniques. This was already > envisioned in the original CfP: "Submissions for all earlier weeks are also > sought, where feasible." However, as far as I know, this request was never > fulfilled. I agree that the more information we have (the long the time series) the better the combination will be. And it is clear that this will be the final goal. This does not mean, however, that the data of several months we now have cannot be used. As pointed out above, the intra-technique combinations are the official series of the Technique Services to start with and we have to know their quality etc. There is a reason why I didn't push for long time series so far: I think that it is extremely important to have the Technique Services involved (this was a general agreement with the Technique Services to my understandding) and to have them perform their intra-combination. They have a lot of expertise with Analysis Center solutions and that's why I would like (to the extent possible) to have the official intra-technique solutions for the IERS activities (even if it requires more time; we need the knowledge and experience of all the groups). The Technique Services put a lot of effort recently into the combination work and now succeeded to get combined solutions. It would not be a wise step to bypass or ignore these official solutions. It will take some more time to get all techniques to produce long time series of combined solutions, but the preparations are on the way and I think that the Technique Services should be responsible for their official products (as in the case of the IGS). To my knowledge, the DORIS series is available over the entire time interval, ILRS and IVS are working on extending their solutions to the past ... but it takes time, manpower, ... > Long-term solutions are available outside the CPP domain for > DORIS (thanks to Pascal) and for the IGS (back to 1999, thanks to Remi), > but we do not have homogeneous combined solutions for SLR or VLBI for > sufficiently long periods. So, we cannot "start generating and delivering > the weekly inter-technique solutions", as requested. I do not believe > anybody can either. Even if a combination on a weekly basis might not be the final way to go, such solutions can be produced in order to: - check the correctness of the combined intra-technique solutions - to study the combination procedures - to propose modifications, if the solutions are having deficiencies If we agree (do we ???) that the intra-technique combined solutions should finally be the basis for an IERS combination, then we should focus on them, test them, improve them etc. Concerning the introduction of local tie information: producing weekly inter-technique combined solutions with local tie information included, does not mean, that the multi-year inter-technique combinations have to start from these weekly inter-technique solutions. Multi-year solutions can always easily start from the original intra-technique solutions to avoid the problem of multiple local tie constraints. > Under the circumstances, it is not clear to me that there is any > justification for such a workshop. Personally, I am not interested in > spending two days to see a lot of ppt presentations of the nice things > that could be done ... someday. Can anybody say that they will be > prepared to report anything new or useful? I do not think that Zuheir > and I will have anything more than the papers will we anyway present for > the AGU, which are tests using solutions from outside the CPP database. > Zuheir, please correct me if I am incorrect about this. I think that there are quite some results to be shown at the workshop. Within the FGS (Forschungsgruppe Satellitengeodäsie) we have a project (Geotechnologien) with several positions to carry out research in this field. So there are new results from DGFI, from GIUB, from FESG, ... and I know that other groups are working hard as well. The workshop is an opportunity to see each other, to exchange results and discuss them in pleno. The workshop is also a motivation to generate results and to show the progress achieved. Comments to e-mail of Zuhier Altamimi (05.10.2004): --------------------------------------------------- > I of course support all your comments, especially for the > work we did together and our conclusions we stressed > during the Nice meeting and in particular the danger of > using local ties at a weekly basis. The danger would be > more aggravated if the IERS tends to make available as > official products to non-specialist users, weekly combined > solutions which incorporate, each week, "not-weekly-measured" > and may be "out-of-control" local ties. I agree with this. If other groups are combining weekly solutions that already contain the local tie information, this would be very bad. If we keep the solutions internal, however, there is no such issue and weekly solutions may be produced and validated. The CPP is a pilot project at the moment to perform studies. > The only way I see to have some "control" on the local ties > and by the way to avoid their over-weighting is to include them > once in a long-term solution as we suggested during the Nice > meeting. Again, it is much more realistic and safe to generate > long-term, well controlled per technique solutions (free of any > local tie) and then to combine these long-term per-technique > solutions including the local ties only once. It is certainly correct, that the intra-technique combined solutions should be used for the accumulation (to avoid the local tie problem). This does not necessarily mean, however, that weekly inter-technique combinations cannot be produced. One type of solutions does not exclude the other, at least at the moment, when we may not have the full time range of decades of intra-technique combinations. Comments to e-mail of Pascal Willis (11.10.2004): ------------------------------------------------ > For other solutions (some are currently being recomputed) it is not > sure that they will be available before san Francisco. So we should > start with what we have now and most probably recompute new > combinations when we have more available solutions. I agree. > In my opinion, the goal of the campaign is not to provide a > "bullet-proof" product for weekly coordinates since the very first > weekly combination but to do it and by doing it to assess the > technical difficulties and the problems in combination and > validation. The solutions are not going to be made available > worldwide at present. My understanding is that it is only for our own > investigations. One day, when everything will have converged we may > distribute them broadly but we are far from it. That's my point of view as well. > Maybe the conclusion is that these inter-technique combination will > not provide any useful feed-back for the techniques as stated by Jim. > For the time being, I don't know the answer to this question. Let's > try it and see if it is useful or not. Because the combined intra-technique solutions are official solutions we have to make use of them. Progress should be made by improving step by step these official solutions. I still think that it is a very important achievement that the ILRS and the IVS are now computing official combined solutions as the IGS is doing since quite a long time. Each of the techniques should reach the same level of product generation, expertise, and reliability. We should keep in mind that the individual techniques are generating more than just the parameters we are now trying to combine. An official intra-technique combination can in principle handle additional parameters that are not (or not yet) part of the IERS CPP. > Following Jim's remark, I agree that it would seem very strange to go > to the meeting in December if no combination has effectively been > done at all before the meeting. It would seem to me also as a waste > of time. Combinations groups need to start something, that will not > be perfect, that will not be compatible and by assessing the quality > of those we will learn something and we will be able to go another > step further and improve. I think that from the list of presentations to be expected for the workshop in San Francisco we will see, whether it is worthwhile to meet there and for how long. We will thus get an overview of the status and can decide thereafter. Before this is clear, I do not see a reason to change the present schedule. To develop software to perform the combination tasks takes time. I think that it is important that several groups can do the combination work and thus validate each other. Even if it takes more time to get a few groups involved (and not just one or two) in the combination of solutions, it will finally be an advantage, because of the mutual quality checks and the progress to be expected from it. We know that this is exactly what is happening with the IGS Analysis Centers and products. Best regards, Markus