[daip] IM2UV and UVMAP questions
Eric Greisen
egreisen at nrao.edu
Wed Nov 8 16:37:02 EST 2006
AIPS en Tuktoyuk writes:
> good afternoon,
>
> from the original interferometer UV data I created a dirty map with IMAGR
> (niter=0), and the map looks like it should dirty. Then using IM2UV I fourier
> transformed the dirty map back to the UV plan. The original UV data had a max
> amplitude around 12Jy while the IM2UV version has a max amplitude around 1200Jy,
> wow. Do you know why this happened?? Why 100X rather than another multiplication
> factor??
> I have done the same procedure for only a single uv-channel and the resulting
> amplitudes in the UVPLOT are the same 100x larger with less visabilities due to
> it only being 1 of the 128 uv-channels.
>
> As per the image brightness: The original dirty map has a max of ~0.5899Jy/bm
> and a total flux of ~0.069Jy. When I use UVMAP on the IM2UV set to transform it
> back to the image plane the image has a max of 0.969Jy/bm and a total flux of
> 2.00Jy. this equals to the total flux being ~29X larger in the new map from
> UVMAP. I assume this is directly related to the increase in the different UV
> data sets, true??
>
> If I export the dirty maps to IDL and fourier transform to UV and then back to
> the image plane I get the same image back with the same brightness' as the
> original. This is true for both the 6-channel cubes and a single channel image.
> Therefore this only seems to be a problem with AIPs.
>
> Is this due to a normalization factor of the FFT and inverse FFT of AIPs?? Or is
> it something else??
>
The IM2UV task does something quite wierd - it FFTs the image and then
pretends that each cell in that rectangle is a UV sample. Your
comment suggests that there is a scaling error - something that FFTs
are prone to. I should look for that as long as I am messing with
this code.
The IDL program you are describing would be like using the AIPS task
FFT. That makes a complex image in the UV plane but does not then
pretend that these are samples for a whole new gridding and
weighting. Thus the FFT and FFT back should produce something much
like the original output from UVMAP.
Eric Greisen
More information about the Daip
mailing list