[Issues] Astrometry implementation and inspect plots

[Philippe Heraudeau]

Hi,

The inspect plots of the Astrometry such as the plots of the overlap 
residuals in GAstrom are misleading because
they are based on source lists from images which do not really exist yet 
with this astrometry!
They are made with source lists extracted from the 'reduced' with 'bad' 
astrometry to which the astrometric solution is
mathematically applied before making these plots but no new image is 
created as this point.
Therefore, the plots show what accuracy we _should_  get on the final 
images but not what we really get on the images.

Checking the residuals of the source lists extracted from the regridded 
frames  (which then have the new astrometric solution applied)
  with respect to the sources detected on the coadd made from these 
regridded frames shows than the accuracy is much worse than what the 
inspect plot shows.

Using these files as an example:
task = GAstromTask( instrument='MDM8K' , C=0, 
red_filenames=['Sci-TSCHNEIDER-MDM8K-------R-4-Red---Sci-54083.5873038-41d3f4854fea35a8a255e98cd42eca2532321705.fits', 
'Sci-TSCHNEIDER-MDM8K-------R-4-Red---Sci-54083.5876102-3e251b2ea5e7fd18321b5c3c8bd92a2dc99852c4.fits', 
'Sci-TSCHNEIDER-MDM8K-------R-4-Red---Sci-54083.5879274-49b5a6a875c70ed35b69593450829c4274effd16.fits', 
'Sci-TSCHNEIDER-MDM8K-------R-4-Red---Sci-54083.5882407-8db59e329bb9c51503028803bb1348cc42515147.fits', 
'Sci-TSCHNEIDER-MDM8K-------R-4-Red---Sci-54083.5885588-b56663af73bd45d8102a2095eb5368c8eadae7d7.fits'],chip='4',inspect=1)
task.execute()

The inspect plot of overlap residuals gives a nice accuracy of 
<DRA> =0.018 +- 0.052 arcsec
<DDEC> = 0.017 +-0.065 arcsec

But using the source lists extracted from the regriddedframes, the 
different exposures show a systematic offset of about 0.05 arcsec with 
respect to the coaddedframes
in one direction or another and the dispersion is 0.2arcsec for DRA and 
0.1 for DDEC which  is much worse than what gives the inspect plot.
This seems  to indicate a problem in the implementation of the proper 
astrometric solution to the images.

1) Could you please have a look at this very soon? it's urgent to solve 
this problem, can someone reproduce this?
2) We need an inspect plot which is based on source list extracted from 
images which have the _right_ astrometry (reduced or regridded frames) 
i.e. separate the computation of
the astrometric solution and the plotting part as it is done now.

Cheers,
Philippe

P.S: for people in Groningen, the plots are available at :
for the inspect plot:  
/Users/users/heraud/mdm_inspect_plot_field02r_chip4.ps.gz
for the source lists extracted from the regridded versus the ones from 
the coaddedframes: /Users/users/heraud/gastrom_pointings_vs_coadd.ps.gz
Pages 1 to 5 shows the DRA, DDEC of one of the 5 observations for chip 4 
versus the coadd, while page 6 shows all 5 pointings  together versus 
the coadd.
Another example with WFI  is at  
/Users/users/heraud/WFI_5pointings_vs_coadd.ps.gz


-- 

==========================================================================
Dr. Philippe Heraudeau
Argelander-Institut für Astronomie Email: heraudeau at astro.uni-bonn.de
Auf dem Hügel 71                   Tel.: +49 (0)228 733673
D-53121 Bonn                       Fax.: +49 (0)228 731775
Germany
=========================================================================