Date: Tue, 17 May 2005 15:20:21 +0200 (Paris, Madrid)
From: Martine Feissel-Vernier <feissel@ensg.ign.fr>
To: ids.analysis.forum@cls.fr
Subject: Re: station breaks 
Message-ID: <Pine.WNT.3.96.1050517143931.274D-100000@verdi.ensg.ign.fr>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

Dear colleagues

You will find hereafter some comments following P. Willis' and 
N. Zelenski's messages about the core network of stations proposed 
by P. Willis and J. Ries for T/P and Jason1 POD. 

Our comments concentrate on potential "problem stations". They are based
on stability analyses of stations coordinates derived by three IDS
Analysis centers: IGN-JPL (ign), INASAN (ina), and LEGOS-CLS (lca) from
all DORIS satellites except Jason1. 

1. Nikita's note

   Although we are not sure we understand exactly how to interpret 
   figure 6 (TOPEX POD improvement (c349-363) and magnitude of DORIS
   station correction), we notice that the stations AMTB, CACB, FUTB,
   MAHB, METB that seem to be best improved have a large rate of
   missing weeks in the time series:
   ign: 28%-38%    ina: 8%-30%     lca: 7%-30% 
   CACB and FUTB have data gaps up to 200 days.

   We have a question about the significance of the improvement: its order
   of magnitude is 0.001-0.006 mm/s for the largest effects.  Is this to
   be compared to the value of the rsidual summary in table 3 
   (0.4-0.5 mm/s)?

2. Pascal's tables

   Taking advantage of the availability of the three solutions, we cheked
   the stability of the four stations in table 3 (proposed stations 
   with revised coordinates). We also selected the stations we find less
   stable in the ign or the lca analysis and checked their stability index
   in the other two solutions. 

   Our stability index is based on two characteristics of the non linear
   station motion, namely the stability for a one-year sampling time
   (measured by the Allan standard deviation) and the type of noise. White
   noise is considered good for stability, flicker noise is considered
   bad. The general noise status in DORIS non linear station motion is
   white noise. Stations with stability indices < 4 are considered stable. 
  

       Stability indices in the ign, lca and ina solutions      
   ---------------------------------------------------------------
   Domes No  Sta.  Time series    Stability index     Stability
                   Start   End    ign   lca   ina   ign  lca  ina
   ---------------------------------------------------------------
   A. Stations with revised coordinates
   91401S001 AMSA 1993.1 1997.3    (1)  4.7  5.6          no   no
   41609S001 CACB 1993.0 2003.2    5.1  6.2  5.9     no   no   no
   21604S003 PURA 1993.1   -       3.3  3.7  4.1    yes  yes   no
   92403S001 RAQB 1996.3   -       3.9  5.5  4.5    yes   no   no
   
   B. Stations that are found unstable in the ign or lca solution
   30602S004 ASDB 1999.3   -       3.3  4.6  2.1    yes   no  yes
   22006S001 MANA 1993.2 2003.0    3.7  5.1  3.3    yes   no  yes
   30313S001 MARA 1993.0 1998.5    5.7  3.5  7.3     no  yes   no
   10503S013 META 1993.0 2000.8    4.7  2.5  4.8     no  yes   no
   41705S007 SANA 1993.1 1996.8    5.1  2.3  2.9     no  yes  yes
   41705S008 SAOB 1997.1 2000.9    5.1  3.6  7.2     no  yes   no
   50107S006 YARA 1993.0 1999.8    4.4  4.0  5.0     no   no   no
   ---------------------------------------------------------------
   Note 1. The AMSA series in the ign solution ends at 1996.0. 
           The stability estimation shown takes into account 
           only series longer than 4 years

   
   All stations except those in part B are consistently found stable in
   both the ign and lca solutions. About a half dozen of those are found
   unstable in the ina solution. It is seen from this table that the
   judgement on the reliability of some stations may depend on the
   Analysis Center.  

   We are not arguing for any change in the Willis & Ries list of station
   coordinates and velocities, but we suggest that the performance of
   the stations in the above list be considered with special care.


Concerning the preparation of the IDS contribution to ITRF2004, we'll make
specific proposals at a later stage. 

Best regards
Martine Feissel-Vernier and Karine Le Bail
IDS Analysis Coordinator team