[daip] aips imagr

[Eric Greisen]

Cao Hongmin wrote:
> Hi Eric,
> 
> In aips and for VLBI data, we first do amplitude calibration and then do 
> fringe-fitting for phase calibration,
> does amplitude calibration (eg. the amplitude calibration is quite bad) 
> will influence the fringe-fitting solution?
> 
> I am ambiguous with the concept -" image aliasing", what kinds of 
> reasons will result in image aliasing, and could we can find some 
> methods to alleviate it?
> 
> Does natural weight will have a larger field of view, comparing with 
> uniform weight?

In FRING, the examination is basically done over phase alone without 
weighting by amplitude.  Data weights may enter a little and that might 
be affected by amplitude calibration, but this should have little affect.

Image aliasing arises from the Fourier transform of the narrow "bed of 
nails" function used to put the image on a regular grid of pixels.  The 
actual image that you make goes to infinity mathematically and is 
replicated every 1/(pixel-spacing) in both the X and Y directions.  One 
can see this is say Westerbork images in which a strong circular 
sidelobe ring touches the edge of the image and is reflected back into 
the main image.  In general, aliasing matters if there are strong beam 
sidelobes (the very regular Westerbork array makes such) and there are 
sources outside the field which you have imaged.  In VLBI, sources 
outside are diminished enormously by smearing in time and bandwidth are 
are essentially never a problem.  VLBI arrays rarely have a lot of 
regularity and so are inclined to have smaller external sidelobes.  Note 
that Cleaning with the UV-plane based algorithm in AIPS should reduce or 
eliminate the problem, so long as aliased signals are not picked up in 
the Clean as real objects and even then those "objects" are often 
removed as the Clean progresses.

Useful field of view is set by the bandwidth and time averaging 
primarily although in VLA observations the curvature of the sky forces 
one to use small facets to counteract issues with phasing.  In VLBI, sky 
curvature is never relevant since the other two kill the visibilities 
after only very small angles.  Weighting has nothing to do with these 
issues.

Again let me recommend books on interferometry with long, much thought 
out explanations, rather than brief remarks in e-mail from me.

Eric Greisen