Date: Thu, 18 Dec 2003 15:58:58 -0800 From: Pascal Willis To: Martine Feissel Cc: Gilles Tavernier , Jean-Francois Cretaux , Serguei Kuzin , Jean-Jacques Valette , Zuheir Altamimi , Herve Fagard Subject: Fwd: DORIS/GPS latest comparison Dear Martine, if we want to have one day a reliable sub-cm DORIS reference frame, we will have at some point to look into details at stations specific problems. following my previous contribution at the IDS workshop on this subject, here is a recent message concerning DORIS/GPS recent intercomparisons (see attached). To help finalizing some possible processing standards between groups, here is the "pre-processing" that I apply to the DORIS data : 1) changing station name (breaks) 23 6 2001 AREA ARE1 16 11 1994 COLA COL1 1 4 1995 DIOA DIO1 1 1 1999 KRAB KRA1 10 10 1994 SAKA SAK1 25 11 2001 SAK1 SAK2 3 10 2002 SODB SOD1 3 11 2002 FAIB FAI1 col 1, 2,3 = day, month, year col 4 = DORIS old acronym col 5 = DORIS new acronym after the break 2) The DORIS periods of time that I do not use are : AMSA 1 1 1996 1 1 2010 antenna fall AMSB 1 1 1990 1 1 2010 antenna fall, no data used for this station HELB 1 2 2000 31 7 2002 antenna fall DORISmail # 199 OTT1 1 1 1990 1 1 2010 antenna fall, no data used for this station OTT2 1 1 1990 1 1 2010 antenna fall, no data used for this station SODA 1 1 1993 1 1 1996 subsidence related to volcano col 1 = doris acronym col 2, 3, 4 = day, month, year for the start of the period to delete col 5, 6, 7 = day, month, year for the end of the period to delete col 9,... = explanation I know that I should also delete some EASA data, but I am hesitating to throw away all 1997 to 2001 without further confirmation from Herve Fagard (already contacted). Hope that it helps Best regards and Happy Holidays Pascal >Date: Thu, 18 Dec 2003 14:45:41 -0800 >To: Michael B Heflin >From: Pascal Willis >Subject: DORIS/GPS latest comparison >Cc: Zuheir Altamimi , Nocquet >, Claude Boucher >, Patrick SILLARD > > >Hi Mike, > >I have compared my recent DORIS solution (tf_031217a) with an >external solution obtained by combining your recent GPS solution >(031009) with GPS-DORIS local ties usual the full covariance >information for all of them. > >It gives a comparison for each GPS/DORIS collocation but >unfortunately only for these. > >The global WRMS that I get are now quite encouraging, but still not >breaking the 1 cm in position and 2 mm/yr in velocity : > > N E V > WRMS 11.4 10.3 15.6 position in mm > WRMS 3.1 2.3 3.2 velocity in mm/yr > >I also looked in detail on a station by station basis, and I think >that you could be interested to get the information given in annex. > >basically the conclusion is that there is nothing larger than 3 cm >in position and 6 mm/yr in velocity that we cannot explain. > >In my opinion, a change in convention could be adopted for GPS in >the case of the BAKO station. > >All other problems are related to DORIS analysis > >Best regards >Pascal > >----------------------------------------------------- > >I use the Gipsy convention for the results > >col 1 = name >col 2 = difference in North >col 3 = difference in East >col 4 = difference in Vertical >col 5 = sigma North >col 6 = sigma East >col 7 = sigma Vertical > >everything in mm and mm/yr > >1)results in velocities > >BAKO-CIBB/CICB > CIBB 3.6 -1.4 -7.5 1.5 1.6 1.0 velocity > CICB 3.6 -1.4 -7.5 1.5 1.6 1.0 velocity >The -7.5 mm in V is explained by the fact that you put a break in >your GPS series around 2000.8. The estimated GPS velocity is then >6.5 mm/yr >The DORIS series has no break and the estimated DORIS velocity is -2.2 mm/yr >If you don't put a break in your GPS time series, the V velocity is >almost 0 (from 1999.5 to 2003.5) as it is the case for the DORIS >series (1993-2003.8) > >PDEL-PDLB/PDMB > PDLB -3.9 1.4 -10.8 1.9 2.5 3.1 velocity > PDMB -3.9 1.4 -10.8 1.9 2.5 3.1 velocity >Most of the DORIS slope in V (-7.5 mm over 1999-2003.8) comes from >DORIS data in 1999-2001 that are outside the GPS observations >(2003.3-2003.7). If I look at the common GPS/DORIS period, the DORIS >slope is also compatible with GPS (almost 0 in V). We may have to >check if other DORIS groups see this non-linearity after 2003. We >may then have to rename this station or at least assume that the >velocity could be different. > >SANT-SANB/SAOB > SANB -7.3 -10.8 -5.7 1.6 2.1 1.3 velocity > SAOB -7.3 -10.8 -5.7 1.6 2.1 1.3 velocity >GPS has good observations both from 1993 to 2003. DORIS only from >1997 to 2003.8 (I exclude the SANA station that is 20 30 km away in >the early years). So my DORIS velocity is basically driven by my >1997 data that are not too good and gives very noisy results. >However, the GPS slope can fit the DORIS results but the DORIS data >are too bad to really estimate the velocity correctly. I may need 1 >or 2 more years of observations for this one. > >STJO-STJB > STJB 3.6 -0.6 -9.2 1.6 1.9 1.2 velocity >GPS observation from 1993 to 2003.0 >DORIS observations from 1999.7 to 2003.8 >My -7.6 mm/yr DORIS velocity comes mostly from data after 2003.0 >(that you do not have for GPS). I now wonder if I should not put a >break there in my time series. The possible DORIS break would be -2 >to -3 cm in V and it is quite difficult to make a decision at this >point. This has to be confirmed by analysis from other DORIS groups >to be 100% sure. However, during the GPS observation period, DORIS >and GPS have the same trend. > >THTI-PAPB/PAQB > PAPB 6.5 -2.1 2.0 1.6 1.8 1.0 velocity > PAQB 6.5 -2.1 2.0 1.6 1.8 1.0 velocity >Looking at the DORIS plot, I may have a small problem with the >DORIS-DORIS local tie (change of antenna around 1998.5) that could >affect my velocity estimation. I would not need to change it too >much to get closer to the GPS estimation and the local tie accuracy >is assumed to be 3 mm. I may need to downweight this local tie a >little bit more. I will check with Herve Fagard. > >2)results in positions >EISL-EASA/EASB >*EASA 29.3 -3.3 -11.7 5.5 6.0 4.7 position > EASB 7.2 1.2 -1.8 4.7 5.4 3.7 position >My series is very noisy in 1997-2001 (EASA observation) and could >easily lead to a 2-3 cm difference in position depending if I throw >away these DORIS data or not. I think there was some concern about >the stability of the EASA antenna, this is why it was replaced by >EASB wich is much stable. I will check with Herve Fagard too. > >FAIR-FAIB >*FAI1 2.8 19.0 -33.5 4.6 4.7 3.8 position >the position is estimated with very few data as it is after the >earthquake (FAI1 is my new name for FAIB after the earthquake). Here >the error is mapped in 2000.0 so I would not worry with that one >presently. > >PDEL-PDLB/PDMB >*PDLB 18.8 12.9 38.3 5.6 7.0 9.0 position > PDMB 19.8 14.1 36.5 5.9 7.4 9.3 position >this could be an artifact of the problem seen in velocity > >all other residuals are smaller then 3 cm > >Best regards >Pascal >--