Date: Mon, 29 Mar 2004 13:47:28 -0600 From: "John C. Ries" To: jean-paul.berthias@cnes.fr Cc: Jean-Michel Lemoine , Pascal Willis , Laurent Soudarin , Gilles Tavernier , Jean-Jacques Valette , Jean-Pierre Granier , Martine Feissel-Vernier Hi everyone, In preparation for the upcoming meeting, and to stimulate discussion, I had a thought or two about any attempt to model or repair the Jason-1 DORIS data for the SAA effect. First, the frequency acceleration in the SAA might be modelable but this model must presuppose that the mean frequency drift that would be present without the SAA is somehow known. This isn't possible, and can only be inferred by assuming the DORIS stations heights are already correct. Thus SAA data cannot be used for station positioning. Second, the non-SAA data have been corrupted due to the mechanism Pascal pointed out...all the SAA data have an erroneous drift model applied. The drift model is correct as an average across the network but it is incorrect (and growing) for the non-SAA stations. This is easily seen as the non-SAA stations appear to be moving in height in the opposite direction of the SAA stations. There is, of course, the issue of the data escaping the CNES editing. In short, I don't see how any Jason-1 data can be used for positioning, neither SAA with a model or non-SAA data. The 'correct' frequency drift model can never be known. I can estimate the 'correct' frequency drift for the non-SAA stations in the POD process, but it can be recovered only because I fix the stations. It may be possible to parameterize the data well enough to keep it in the POD process, and recover at least the along-track time-bias information that is Doppler's main strength. See ya, JR