[daip] Smoothing option in IMAGR
Kazi Rygl
kazi at mpifr-bonn.mpg.de
Fri May 7 12:02:34 EDT 2010
Dear Eric Greisen,
On Fri, 7 May 2010, Eric Greisen wrote:
> Kazi Rygl wrote:
> > Dear Sir/Madam,
> >
> > I want to report a problem concerning the SMOOTH option in IMAGR for AIPS
> > versions DEC09 and DEC10. When imaging a spectral line data with the option
> > SMOOTH 1 0 (Hanning smoothing) and inspecting the image cube with ISPEC, I
> > noticed that the smoothing was not correctly applied: the spectral shape
> > remains as if unsmoothed.
> > I tried the same operation in the DEC08 version, where the spectral shape
> > was indeed altered by the smoothing.
> >
> > Could you help me to overcome this problem?
>
> SMOOTH is a basic parameter of the UV data reading and has not deliberately
> changed in many years. It should be no different in recent versions than
> 31DEC08. Look at the history file of the output image and see if SMOOTH
> appears there and what its values are.
In the history files of the images made with DEC10 and DEC08 I find the
same line:
1266 IMAGR SMOOTH( 1) = 1.00000E+00
> Try POSSM with SMOOTH in the various versions to see if SMOOTH fails there.
In POSSM (DEC10) SMOOTH works fine.
> Try running SPLIT or SPLAT with SMOOTH and feeding the result into IMAGR with
> no SMOOTH. IMAGR does so much to the data after the data are read that it is
> sometimes hard to see what is going on but without more to go on I cannot see
> how it could be failing. I suppose I could try it again.
I did, as you suggested, a SPLIT with SMOOTH 1 0 and afterwards imaged
the data using IMAGR with SMOOTH 0. In SPLIT SMOOTH worked properly!
I don't know why the SMOOTH option in IMAGR it did not work for me.
>From now on I will SPLIT the data before imaging them, and then my
problem is solved.
Thank you very much for your fast reply & help!!
Kazi
>
> Note I used SMOOTH=1,0 for a 24000 channel cube last December and it worked as
> intended (we had ringing in the spectra due to channels that were too wide).
>
> Eric Greisen
>
More information about the Daip
mailing list