[daip] UVMOD/UVFIX question

[Eric Greisen]

Hanno Spreeuw wrote:
> Thank you for clearing this issue.
> So I guess it is a known fact among the experts that the accuracy of 
> astrometry on the brightest sources is limited by the angular distance 
> from the tangent point in the sense that beamsize/(s/n) accuracy cannot 
> be attained. To get the optimum accuracy one has to move the tangent 
> point close to the bright source.
> I was not aware of that.
> I thought that any source measurement in a facet with a size recommended 
> by SETFC would be ok.
> If I understand correctly, distortions grow quadratically with distance 
> from the tangent point, so with FPOS=0 250, the position error will be 
> at least 3.36"/16=0.21".
> 
> Perhaps it is simplest for the purpose of testing our source extraction 
> code to run UVMOD on data from a linear array.
> 
> Hanno.
> PS: Is it not recommended to do astrometry on images from FLATN? Does 
> FLATN cause an extra error?
>
It is a known fact that one has to do faceted imaging when the field 
exceeds some size.  It is also known that sources will appear smeared 
and at the wrong place in images made with the assumption that a 2-D
FFT is legal.  SETFC has a parameter which controls how much error you 
are willing to accept.  That parameter by default is too large I think 
and esp at very low declinations where the positional smearing and error 
are much larger for the same value of W.  The point I made about how 
many pixels had Clean components is also telling - 21 or so rather than 
3.  That is a big difference.  You should probably be making your images 
with 256x256 facets of about 6 arcsec cell size.

If you do the imaging correctly with small facets, then the Cleaning and 
CCs will be at the correct position.  FLATN will retain that position in 
the flattened output so astrometry should be fine at that point.  But 
the "If" at the start of this paragraph is very important.

Eric Greisen