High SNR R&D VLBI experiments in 2002
Statistics of 7 high SNR R&D experiments. Postfit residuals and formal uncertainties of polar motion are obtained from global 2003b_pre solution.

RD experiments 

code    db-name     sked  db    5-9   used  fit eps Y   UT1

rd0202  $02MAR05XA  2090  1889  1704  1663  12  55  51  2.3
rd0204  $02APR03XA  2032  1952  1776  1733  11  50  50  2.2
rd0205  $02MAY15XA  1915  1035   905   877  12  70  86  4.5
rd0206  $02JUN12XA  1962  1874  1712  1671  17  83  79  3.6
rd0207  $02JUL10XA  1953  1416  1300  1275  14  80  70  2.9
rd0209  $02NOV06XA  2031  1859  1229  1180   8  68  55  2.3
rd0210  $02DEC04XA  2049  1947  1836  1787   9  52  51  2.2

R1 experiments 

code    db-name     sked  db    5-9   used  fit eps   Y     UT1

r1009   $02MAR04XA  4596  4240  3866  3853  21  48.6  60.2  1.8
r1013   $02APR02XA  3277  3121  2890  2856  23  62.3  67.7  2.0
r1019   $02MAY13XA  2468  2324  1928  1857  25  50.5  55.3  2.3
r1023   $02JUN10XA  2565  2425  2124  2106  29  49.0  57.7  2.4
r1027   $02JUL08XA  3596  2731  2403  2295  24  76.0  83.5  2.1
r1044   $02NOV04XA  3411  3103  2809  2793  21  65.0  69.4  2.0
r1048   $02DEC02XA  3319  3083  2916  2875  21  60.8  63.2  2.0

R4 experiments 

code    db-name     sked   db    5-9  used  fit eps   Y     UT1

r4010   $02MAR07XE  1325   1298  1231 1206  23  57.6  60.8  2.1
r4014   $02APR04XE  1757   1718  1575 1554  21  55.7  52.4  2.2
r4020   $02MAY16XE   744    702   648  644  21  73.5  71.0  2.9
r4024   $02JUN13XE  1432   1390  1149 1121  30  82.4  87.3  4.1
r4028   $02JUL11XE  1587   1543  1396 1370  22  55.0  62.4  3.0
r4045   $02NOV07XE  1393?  1745  1566 1554  18  57.3  60.9  2.6
r4049   $02DEC05XE  1618   1245  1141 1128  18  77.7  91.3  3.0


sked -- number of scheduled observations
db   -- number of correlated observations
5-9  -- number of pairs of observations with quality codes 5-9
used -- number of used observations
fit  -- wrms of postfit residuals in psec
eps  -- formal uncertainties of the estimates of nutation in obliquity (muas)
Y    -- formal uncertainties of Y polar motion (muas)
UT1  -- formal uncertainties of UT1 estimates  (musec)

Differences of the estimates of pole coordinates against IGS: 

--------------------|---------|---------|---------|
                    |    R&D  |   R1    |    R4   |
--------------------|---------|---------|---------|
X-pole (uas)        |   167.2 |   131.1 |    81.9 |
Y-pole (uas)        |    72.6 |   107.9 |    99.9 |
--------------------|---------|---------|---------|
(hint: standard deviation of the rms over 7 points is about 50% of the rms itself)

Why it is desirable to move to the high SNR in R1 sessions.

Today (2003.02.27) We have 48 R1 sessions recorded in 256 Mbit/sec mode for the period of about one year. It is clear that R1 sessions do not produce better EOP than R4. It is fact. So, unless we change observing strategy and strategy of data analysis we will get results similar to the 10 years old Mark-3 style of R4. We learned the lesson: increasing data rate per se does not improve VLBI results.

These tables show that high SNR R&D experiments give EOP with formal uncertainties not worse than R1, R4 and even better.

Investigation of IMF shows that troposphere is not the major source of errors in VLBI results. Imperfection of the geophysical models is, certainly, the factor neither. The major factor which limits precision and accuracy of VLBI is INSTRUMENTAL ERRORS. So, the main reserve in improvement of VLBI results is development of a technology which would allow us to investigate, understand and eventually reduce effect of instrumental errors. High SNR sessions are the most advantageous for this strategy: when the level of noise is reduced, instrumental effects became more visible.

High SNR experiments are inevitable in the future. How it was already shown, antennas' slewing sets the limit of the number of observations per sessions. Current R1 sessions are rather close to this limit. So observing in 1 Gb/sec mode we will not in a position to set significantly more scans than in R1 (in fact, no more than 30% than in R1). But we can increase SNR limit. The challenge is to transform high SNR in high accuracy. We tried high number of scans approach. We see it does not work. Let's try high SNR strategy. I propose to change the SNR limit in R1 sessions to 80/60 -- the same as it was in the R&D sessions.

Back to Leonid Petrov's discussion page.

Last update: 27-FEB-2003 18:32:48