[daip] IMERG

[Eric Greisen]

Lynn D. Matthews wrote:
>  	Eric,
> 
> I have placed four files in ftp at 
> ftp://cfa-ftp.cfa.harvard.edu/outgoing/lmatthew/
> 
> These are:
> 
> VLA_orignal.fits (a deconvolved VLA channel image)
> 
> GBT_original.fits (GBT map channel image from cube made with SDGRD)
> 
> GBT_hgeom.fits (GBT channel map regridded to match VLA data using HGEOM)
> 
> GBT_hgeom_addbeam.fits (regridded GBT map with beam parameters added to 
> header using "ADDBEAM").
> 
> As you'll see, there isn't much in the VLA image, although the little 
> HI cloud at the center is astrophysically quite interesting.
> 
> Simply using COMB I can get a combined map that looks pretty good, but 
> trying to do things "right" with IMERG, I can't get anything sensible.
> 
> Appended below is an example of inputs I tried (although I've tried many 
> variations). The scaling factor the program computes essentially 
> downweights th GBT data to nothing. If I force this to something more 
> significant, then I see nothing but a series of concentric squares equal 
> in size to the GBT map.
> 
> Perhaps my maps aren't of the type the program is expecting, but I've 
> tried altering the GBT map units to Jy/beam, forcing the two images to 
> have the same FOV, etc., all to no avail.
> 
> I have seen where CASA does indeed have tools to deal with this type of 
> combination, although I don't have it installed and have no prior 
> experience with even the rudiments of it, so that would be a major 
> undertaking.
> 
> Lynn
> 
> AIPS 1: IMERG :  Task which merges high and low res. images
> AIPS 1: Adverbs     Values                 Comments
> AIPS 1: ----------------------------------------------------------------
> AIPS 1:                                    High res. image
> AIPS 1: INNAME     'VLA 70'                   Image name (name)
> AIPS 1: INCLASS    'SUBIM'                    Image name (class)
> AIPS 1: INSEQ         1                       Image name (seq. #)
> AIPS 1: INDISK        4                       Image disk drive #
> AIPS 1:                                    Low res. image
> AIPS 1: IN2NAME    'GBT V=70 B'               Image name (name)
> AIPS 1: IN2CLASS   'SUBIM'                    Image name (class)
> AIPS 1: IN2SEQ        2                       Image name (seq. #)
> AIPS 1: IN2DISK       4                       Image disk drive #
> AIPS 1:                                    Output  image
> AIPS 1: OUTNAME    'TEST'                     Image name (name)
> AIPS 1: OUTCLASS   ' '                        Image name (class)
> AIPS 1: OUTSEQ        0                       Image name (seq. #)
> AIPS 1: OUTDISK       4                       Image disk drive #
> AIPS 1:
> AIPS 1: BLC         172         173        Bottom left corner of input
> AIPS 1:            *rest 0
> AIPS 1: TRC         340         341        Top right corner of input
> AIPS 1:            *rest 0
> AIPS 1: UVRANGE       0.17        0.46     Min and max baseline (klam.)
> AIPS 1:                                    defining annulus of overlap
> AIPS 1: FACTOR        0                    If > 0.0 , use as norm.
> AIPS 1:                                    factor for low res. image
> AIPS 1: AVOPTION   ' '                     Averaging option '  ','MEAN'
> AIPS 1: DOWEIGHT     -1                    <= 0 => do not correct for
> AIPS 1:                                    beams (BMAJ, BMIN, BPA)
> AIPS 1: BADDISK    *all 0                  Disks to avoid for scratch.
> 
> 
> 
> 
> On Sat, 20 Jun 2009, Eric Greisen wrote:
> 
>> Lynn D. Matthews wrote:
>>> I have a GBT map that I've created using SDGRD and would now like to 
>>> combine this with some VLA data. Is IMERG the appropriate task for this 
>>> purpose?
>>>
>>> So far, I've been unable to obtain anything useful from IMERG (e.g., I get 
>>> something that looks similar to the original VLA map with some extra 
>>> noise/negative flux in the region of the GBT map). This after regridding 
>>> the GBT map to match the VLA pixel scale, adding the GBT beam size to the 
>>> header, and converting the units to Jy/beam. I have done no corrections to 
>>> the input (deconvolved) VLA map. Am I misunderstanding what the program 
>>> wants in terms of the input maps?
>>>

I have played around a bit.  One thing you forgot to do is to multiply 
the GBT image by the VLA single-dish beam.  I did that and then padded 
the GBT image with 0's so that it was also 512x512.  Using BLC 0, TRC 0
in IMERG it actually did something reasonable.  The factor it used was 
.13 which is about correct.  Unfortunately, the GBT image really does 
not die out reasonably at the edges even with PBCOR so the comkbination 
still does not look physical.

Other than the typo which I will now fix, I don't think the difficulties 
lie in the Fortran.

Cheers,

Eric Greisen