Mime-Version: 1.0
Message-Id: <p06020407c367b658d4ca@[129.116.190.133]>
In-Reply-To: <C3676A04.BC70%flemoine@puuoo.gsfc.nasa.gov>
References: <C3676A04.BC70%flemoine@puuoo.gsfc.nasa.gov>
Date: Mon, 19 Nov 2007 17:06:42 -0600
To: "Frank G. Lemoine" <flemoine@puuoo.gsfc.nasa.gov>,
        <ids.analysis.forum@cls.fr>
From: "John C. Ries" <ries@csr.utexas.edu>
Subject: Re: [ids.analysis.forum] DORIS on-site meteorological data
 (response)
Cc: "H.Fagard" <herve.fagard@ign.fr>
Content-Type: text/plain; charset="us-ascii" ; format="flowed"

Hi Frank,

Do you apply any apriori to the zenith delay estimates? Do you 
estimate the full zenith delay or only the wet part? I believe these 
will influence the effect of missing met info.

We have two results that are related to your question.

With PRARE, GFZ did not always have real met info, and so some 
climatological values were substituted in the data we were given. We 
found it important to replace these with 'real' met info (from ECMWF, 
I think). However, we were estimating only the wet part, so it would 
not be surprising that we needed good a priori pressure values so 
that the dry part could be computed with some confidence. We also 
noted that even when we had in situ measurements, they were not 
necessarily unbiased. We compared the met data with ECMWF and removed 
biases where we thought it necessary.

With DORIS, we found it beneficial to constrain the zenith parameter 
estimates at the few percent level, at least in terms of recovering 
good station coordinate estimates. In this case, we were estimating 
the full zenith delay (wet and dry). The models are certainly good to 
a few percent, and it is not unreasonable to use this to constrain 
the solution. However, this again assumes reasonably good values for 
the met info. I think this is described in one of our early DORIS 
papers...perhaps "Absolute Positioning Using DORIS Tracking of the 
SPOT-2 Satellite, Geophysical Research Letters, 19(20), 2039-2042, 
October 23, 1992 (M. M. Watkins, J. C. Ries, and G. W. Davis)".

I would expect that estimating the full (wet+dry) zenith delay, 
without any apriori constraint, should be relatively insensitive to 
not having the true met info. But if one is applying some apriori, or 
assuming that the dry part is known so that one needs estimate only 
the wet part (both are good strategies to improve the results), then 
I would think that one needs at least a reliable value for the 
pressure.

Best regards,
John