[evlatests] Referenced Pointing Stats

[Rick Perley]

    I have commented recently that 'referenced pointing' seems to be 
less effective these days than in the past.  Here are some statistics to 
back this claim up:

    In reviewing the long history of 'flux density' measurements, the 
data taken in January 2006 stand out as exceptionally good.  I reviewed 
the distribution of the gain solutions from the source 3C147 at 43 GHz 
taken then, and compared these to the data taken in late December 2010, 
of the same source is very similar weather (light winds, mostly clear 
skies). 

    In January 2006, we had 15 observations of 3C147, using 22 
antennas.  Of these ~330 solutions:

       Four had a amplitude low by 5%, and 19 others were low by 3 to 5%. 

    In December 2010, we had 9 observations of 3C147, using 25 
antennas.  Of these ~240 solutions:

       Three had an amplitude low by 20%, 12 others low by 10 to 20%, 26 
others low by 5 to 10%.  (It is meaningless to count the rest, as the 
antenna pointing is too uncertain). 
   
    For reference:  at 43 GHz:

    A 9 arcsecond error results in an amplitude loss of 3%
    A 12 arcsecond error ...........................................   5%
    A 16 arcsecond error .............................................10%
    A 22 arcsecond error .............................................20%

    Thus, in January 2006, ~1% of the antenna pointings were in error by 
more than 12 arcseconds.
    But in December 2010, 17% of the pointings were worse than this. 

    I need to emphasize that referenced pointing *is* making the large 
corrections needed to get the beam 'more or less' on source.  But the 
claimed accuracy -- which we have met in the past -- is not currently 
being met.