[daip] accuracy estimate eposw

[Eric Greisen]

Sara Beck wrote:
> Dear Eric,
> 
>      I can't find anywhere an estimate of the accuracy of the task
> eposw (only statements like: 'probably not accurate enough for you').
>  I am not working with high accuracy but just need some rough lower
> limit. Will the standard J2000-B1950 switch be as accurate as 0.1
> arcsecond?  Thanks for your advice, Sara Beck

It should be accurate to 0.1 asec except way away from the pointing 
position.  The B1950 coordinates were not well defined in some ways so 
there is a bit of a warp over really large fields.  Also the b1950 
conversion requires an observation reference date for the coordinates 
and that is not recorded anywhere.

If you really want accurate go back to the uncalibrated data and run 
UVFIX.  Then recalibrate, reflag etc.  The FG and SU tables should be 
all fine and probably you could run UVFIX on the multi-source file jsut 
before you ran SPLIT.  UVFIX does a very accurate coordinate shift and 
recomputation of u,v,w.  It is viewed as the "gold standard" in the 
business.

Eric Greisen