From: Pascal Willis Date: Tue, 30 Dec 2003 13:45:42 -0800 To: jean-michel lemoine -grgs-61332 From: Pascal Willis Subject: correcting DORIS/JASON data using LEGOS/SAA model Cc: Willy Bertiger , Bruce J Haines , Yoaz Bar-Sever Dear Jean-Michel, Happy New Year for 2004!!! Here are some very good news for you. Congratulations for your clever way to handle the JASON/SAA problem. following your latest email, I corrected the DORIS/JASON data using your correction file (JASON satellite drift tabulated per epoch). For test purposes, I only processed 8 days in 2003 (94 to 101 = April 4 to 11, 2003) in free network mode, using Gipsy/Oasis on a daily basis. 1)JASON-only positioning solution The estimated stations positions agree quite well with my own reference (using 10 years of all DORIS data) combined over the 8 day period : N E V everything in mm WRMS 38.6 57.8 51.9 JASON-only solution In my opinion, this is very encouraging with only 1 satellite. 2)JASON + other satellites positioning solutions I then combined the above 8 day solution with the regular weekly solution that I had previously computed using all satellites but JASON. This is only a combination at the normal matrix level. Here are the results (comparison with my position/velocity reference) : N E V everything in mm WRMS 17.0 16.7 19.3 all satellites but JASON (previous processing) WRMS 13.7 14.6 12.8 all satellites including JASON This is the first time that I can see a real improvement in the positioning results using the DORIS/JASON data. In my opinion, it totally validates your approach of the SAA effect. It could be interesting to continue these tests on other periods of time as some weekly solution give better results than others (sometimes closer to 10 mm) without using JASON data. 3) combining all satellites at the data level In this second approach, I do not combine the normal matrix of stations positions coming from JASON with the normal matrix of position coming from all satellites but JASON, but I really process the data together, putting tight constraints on clocks and tropospheric parameters. The first good news is that the DORIS residuals are now much smaller and closer to what we expect (3.8 mm/s for JASON) Here are the new results over 8 days : N E V everything in mm WRMS 16.1 14.5 15.7 all satellites (at DORIS data level) The results are definitively better with JASON but not as good as with the normal matrix approach. This might mean that the way I presently apply the correction does not account for all of the SAA effect. In my opinion, your corrections should also be used by the SOD to preprocess the DORIS measurements as there are presently more and more DORIS/JASON data rejected. This is an important and logistically difficult issue that you could discuss internally in Toulouse in order to help the JASON/SWT team get sufficient DORIS/JASON data in the future. I would appreciate if we could conduct some more extensive tests on a much longer period of time to have a more definitive opinion on these results. I could also send you a complete TOPEX/JASON clock drifts estimations series that could help you refine your model. If you have different versions of your model, it could be easy for me to tests several of them to verify how different they are. See also my technical question in my previous message about the time scale used in your file. Best regards Pascal NB: Please feel free to distribute this message. As you did all the work you should take the initiative on that. -- Pascal Willis Jet Propulsion Laboratory, MS 238-646 Ph : 1-818-393-4748 4800 Oak Grove Drive FAX: 1-818-393-4965 Pasadena CA 91109 Em : Pascal.R.Willis@jpl.nasa.gov