Mime-Version: 1.0
Message-Id: <a06110402bf96178e886d@[172.16.0.131]>
In-Reply-To: <200511071736.jA7Ha2B05914@cnes.fr>
References: <200511071736.jA7Ha2B05914@cnes.fr>
Date: Tue, 8 Nov 2005 11:40:20 +0100
To: ids.analysis.forum@cls.fr
From: "H.Fagard" <herve.fagard@ign.fr>
Subject: [ids.analysis.forum] Re: Dyonisos and other long-term observing
 DORIS stations
Content-Type: text/plain; charset="iso-8859-1" ; format="flowed"
Content-Transfer-Encoding: 8bit

Dear all,

At 18:35 +0100 7/11/05, Tavernier Gilles wrote:
>You can notice that the residuals have reduced with time, which is 
>normal according to the increasing number of contributing satellites.

Maybe we should compute a stability index, and compare sites using 
only periods with at least 3 satellites?

>Dyonisos might be penalized by its very long time series in the 
>rating given by Hervé.

This is correct, and is also true for all "old" sites. Does this mean 
that the length of service criteria is not that important, and we 
should only argue about  history as of early 1994?

Now let's take a look at how these old sites are pealized by their 
long time series. From the three files resulting from Karine and 
Martine's work, I have extracted subsets defined as follows:
- data start in 1993 (the earliest data used by the ACs),
- only Alcatel antennas to allow for a less biased comparison.

Results are as follows, looking at the stability index:
- From IGN analysis, DIOA ranks as number 16 out of 35,
- From LCA analysis, DIOA ranks as number 32 out of 36,
- From INA analysis, DIOA ranks as number 15 out of 36,

So there are at least 14 sites which are "as penalized as Dionysos by 
their very long time series", who behave better thant Dionysos 
stabilitywise.

>If you look at the residuals after 2004, they are much smaller

I would even say that results improve at the beginning of the 
5-satellite period, and stay so or keep on improving even after the 
loss of Topex, probably because the ACs are continuously improving 
the quality of their products.

>despite the "poor monumentation"

This is true also for other "poorly monumented" sites, although there 
are only a few sites on which we can perform such a comparison, as 
almost all stations had been renovated by then.
Nevertheless if you look at the graphs for the following antennas: 
ROTA - SAKA - COLA - GALA - PURA - RIDA - TLHA you will see that 
results are much smaller at all these sites after the start of the 
5-sat period, despite their "poor monumentation". It is even true for 
COLA which was undoubtedly the worst antenna stabilitywise (a 3-meter 
tower with no guy wires on top of a many storey building) before it 
stopped transmitting.

>and it might be worth keeping this station in the network, provided 
>that we improve its monumentation.

Among the 7 sites above, 3 have been removed, and replaced with 
completely new sites at new locations, whith high quality 
monumentation and responsive maintenance. Does anybody complain about 
this evolution?

Looking back further, there have been quite a few more examples of 
old sites removed, and replaced with new ones. And we are very happy 
with these new sites (I'm pretty sure some of you would be reluctant 
to our moving them ;-).
My proposal is not to move stations at every opportunity, but the 
network has been slowly evolving and improving over time and it 
should keep on doing so if needs be.

Re. maintenance quality: if IGN had been so suspicious about new host 
agencies' responsiveness when deploying the network over more than 10 
years, we would not have today such an homogeneous 56 station 
network. I'm not saying everything is perfect, we do have a couple of 
sites where maintenance efficiency is an issue, but it is only a very 
small percentage.

If we install a new station at Riyadh or Zelenchuskaya, there is no 
risk as long as we keep Dionysos as a backup during the new station's 
validation stage.

Best regards,

Hervé