[daip] Re: IMMOD problem?

[Christopher Flatters]

-----BEGIN PGP SIGNED MESSAGE-----

Alan Roy <aroy at mpifr-bonn.mpg.de> writes:

> I've managed to break IMMOD, I think, with a somewhat strange
> image.  It's a ROSAT image of an Abell cluster from which I want 
> to generate and subtract a Gaussian source model.  IMMOD (99DEC31) 
> works fine on the cleaned VLA image of the cluster, but I suspect 
> that the lack of a restoring beam on the ROSAT image may be 
> causing the problem.  Symptoms are that IMMOD produces an
> image containing entirely zeros.

You have a number of problems.

Firstly, the original image is divided by FACTOR before subtracting
the model. Here FACTOR is 0.0 so the original image disappears.

Secondly, ZEROSP needs to be the total flux in the model rather
than the peak flux. Your model covers many square pixels so the
flux should be considerably higher.

Finally, IMMOD does not like pixel coordinates. It assumes that
the spatial coordinates of an image are in degrees and divides the
dimensions of the Gaussian distribution by 3600.0 to convert from
arc seconds to degrees. This makes the model much smaller than
a pixel.

The best way to get around this is to use AXDEFINE to relabel
the coordinates of the image. The best approach would be to
label them with celestial coordinates if you know the coordinate
system that has been used. If not, the following will allow you to
specify the component size in pixels.

GETN whatever
NAXIS = 1
AXTYPE = 'RA'
AXINC = 1.0 / 3600.0
AXVAL = 0.0
AXREF = 169
AXDEFINE
NAXIS = 2
AXTYPE = 'DEC'
AXREF = 189
AXDEFINE

    Chris Flatters
    cflatter at nrao.edu

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 6.5.1 for non-commercial use <http://www.pgp.com>

iQCVAwUBOPOebgrvHSGSTDWdAQGnOAP/Sng/496nHbRsdh0hTtAcYUTSWaaLzE7f
baGwEZpejBl9m1JME0JXCzPuzq4PPMI6FCiMThiJK0t9sQb5S5sKwwZ3E/TxtSNQ
KQuNBvoZM3sZMHpbU09Avxi28vw9AFmO/M8/cPxKMD70MGBl3fzZFu4+PNVzdag5
bL1pYb/EcOo=
=wScj
-----END PGP SIGNATURE-----