Dear Pascal, Thank you for your detailed comments and for recalling the proposal we (Jim and me) presented in Nice. But the point which is may be not clear for everyone is that we have FIRST to establish the long-term multi-technique solution using all the old and current data to be used as a reference. Therefore until we did not establish this long-term solution, (e.g. ITRF2004), we could not confidently start the weekly analysis. Regards Zuheir Pascal Willis wrote : > Dear colleagues, > > I do not totally agree with Jim and Zuheir and I would like to share > here some personal thoughts concerning this important issue. > > I hope that Markus and other people will express their opinion on this > important subject. Otherwise, we should stop this campaign as soon as > possible. > > In my opinion, there is presently some confusion in the definition of > the products of the campaign and what we (or other people) want to use > them for. > > First of all, I agree with the comments from Jim concerning the local > ties that should not be added several times in the least squares > adjustements because they are not measured every week. This would > totally mix up the a posteriori standard deviations on all parameters. > > In my opinion, we need to produce 2 things in parallel using a specific > strategy: > > - a time series of cumulative solutions (position/velocities using all > available data over several years) > > - a time series of weekly stations positions > > As mentioned by Jim and Zuheir, the cumulative solutions can not be > derived directly from the combined time series of the same weekly > positions (problem of the ties). > > One way to solve the problem is to obtain first a cumulative solution > (just to start with), using all present weekly solutions from all > techniques and all available local ties. This is very similar to the > proposed ITRF2004 as if I can understand it. The goal of this initial > product is to provide the best long-term positions and velocities as > well as a realization of the frame. > > For weekly solutions, the trick is first to combine all techniques > solutions + the relevant local ties and only use at this point the > available cumulative solution to provide the results in a coherent frame > (using 7-parameter transformation). The goal of this product is to > provide the best "weekly coordinates" from all geodetic techniques (as a > goal of the GGOS project). > > Creating a new cumulative solution can be done by adding the previous > cumulative solution (years) with the new weekly solutions (1 week) > without any additionnal ties (except for the very few ones for new > stations for which ties were not used before). > > When the first cumulative solution is good enough, the use of 2 > different cumulative solutions to provide the weekly coordinates does > not change the weekly coordinates themselves too much. However, the > cumulative solution must be changed from time to time as there is small > improvement in the velocity determination. This can be done every week > or less often (see after). You only see jumps in weekly coordinates if > the cumulative solution is not good enough. So even for the first week, > I strongly recommend to use a cumulative solution estimated over several > years to avoid any discontinuity in the series (basically everything we > have at present). > > There is nothing new in this technique and you can see that we have use > it for years at JPL for the DORIS IGN/JPL solution. We currently provide > 2 different weekly solution for 2 different purposes : > > - 1 in free-network that is used by the IERS campaign and that can > directly be used in the combination) > - 1 being projected and transformed for which positions are > expressed in ITRF2000. This series is continuous (using a cumulative > solution that now changes about only once a year) and can be used by > geophysics who are just interested to see differences in coordinates. > > This allow at each step to have weekly coordinates in a well-defined > reference frame that can be used for geophysical purposes and also > cumulative solutions that are the current way to maintain the > terrestrial reference frame for geodesists. > > So there is no need to create "weekly solutions in isolation" and this > should be avoided as already discussed in Nice. The problem would be > very different if we had no data at all and would start with just 1 week > but this is not the case, we have years of results for all techniques. > > I do not understand why we have this discussion now and not 1 year ago > as these issues were the same by then, as well the possible solutions > and the technical tools to solve these problems. > > Jim and Zuheir were correct in trying to wake up this IERS campaign > which had every ambitious goals but in which nothing is presently done > (except the technique-dependant part). > > Finally, concerning the comments on the VLBI and Laser solutions, I also > agree that the call for participation was quite clear on that. The goal > was not to show the best possible results (fixing parameters is always a > good way) but being able to do a real multi-technique combination using > strength and weaknesses of each technique. The fact that nobody noticed > that the series were not fulfilling the requirement means that nobody > really uses the present time series. This is quite worrying for the > future of the campaign. It is more than time that the groups who > proposed to to inter-techniques combinations express their opinion and > start producing something. > > I totally agree with Jim. It is then more than time to get long-term > time series of solutions for all techniques to allow these absolutely > necessary cumulative solutions as well as to verify that all technique > solutions fulfill the requirements of the campaign. This was discussed > in Nice and it was said that several series already existed and could be > downloaded without waiting. > > Concerning the meeting itself, it seems unfortunately that it is the > only way to push people to get results in a reasonable time frame. Do > you think that we would have started this discussion if no meeting was > announced before the end of 2005? I guess I may sound rather pessimist > these days on the human nature. > > If we postpone this meeting, we will lose even more time as nothing will > be done before another meeting will be announced and finalized. So I > would strongly recommend to have this meeting but to focus the > discussions on what could not be done, why it could not be done, and how > we could solve the problem (or not) in the future to be more > constructive. We need to be able to define a time table for the next > steps to come. > > Best regards > Pascal > > NB: there are also ways in least squares adjustement to substract or > correct old local ties by adding or substracting the relevant normal > matrices. -- ---------------------------------------------------------------------- Zuheir Altamimi Email : altamimi@ensg.ign.fr Institut Geographique National Phone : 33 1 64 15 32 55 ENSG/LAREG FAX : 33 1 64 15 32 53 6-8 Avenue Blaise Pascal 77455 Champs-sur-Marne, FRANCE ----------------------------------------------------------------------