User-Agent: Microsoft-Entourage/10.1.4.030702.0
Date: Thu, 05 Jul 2007 18:18:39 +0000
Subject: Jason DORIS version 2 data (from Jean-Paul Berthias)
From: "Frank G. Lemoine" <flemoine@puuoo.gsfc.nasa.gov>
To: <ids.analysis.forum@cls.fr>
Message-ID: <C2B2E67F.9FC4%flemoine@puuoo.gsfc.nasa.gov>
In-Reply-To: <C839170D12C1FA488B22101CF46F3D3C512881@sgt-xchange1.sgt-dom.com>
Mime-version: 1.0
Content-type: text/plain; charset="US-ASCII"
Content-transfer-encoding: 7bit


----Original Message-----
From: Jean-Paul Berthias [mailto:jean-paul.berthias@cnes.fr]
Sent: Thursday, July 05, 2007 5:15 AM
To: ries@csr.utexas.edu
Cc: Cerri Luca; Nikita Zelensky; Frank G. Lemoine; Pascal Willis;
sluthcke@puuoo.gsfc.nasa.gov; Perrachon Pascal; Houry Sabine; Mercier
Flavien
Subject: Re: RE : Jason DORIS version 2 data

Dear all,

What we are looking at is the difference between 2 orbits computed using
different datasets. The new dataset has been produced using the same
strategy 
over the whole interval, however this is not true for the old set. As the
SAA 
problem worsened more and more of the contaminated data were removed. These
actions were not continuous in time and they probably introduced
discontinuities 
in the number and quality of the distributed data. I think the
discontinuities 
that we see in Nikita's plot are just the consequence of the steps that were
taken on the CNES side to limit the impact of the SAA.

What I find more disturbing is the plot of the Z shift between the DORIS/SLR
solution and the SLR/Xover solution. I am not sure that we can trust the
trend 
analysis given the dispersion in the data itself, but there is still a
clearly 
visible discontinuity.

Now I assume Nikita is using all the data in the new dataset without any SAA
specific preprocessing. Therefore the orbit quality is directly related to
the 
performance of the SAA model. In a sense it is not surprising that the model
cannot fully correct all of the complexity of the actual problem. At CNES a
downweighing of the SAA stations has been maintained even when the SAA model
is 
used. Maybe such a strategy should be recommended when using the Jason data.

Best regards,
Jean-Paul

------ End of Forwarded Message