Subject: 
        FREXAS/PIXPHOT/PICMAN
   Date: 
        Fri, 31 Oct 1997 08:46:06 -0800
   From: 
        jmc@astro.caltech.edu (John Carpenter)
     To: 
        tchester@ipac.caltech.edu
    CC: 
        jmc@astro.caltech.edu




Hi Tom,

Here is background information for the PIXPHOT/PICMAN/FREXAS review at
the meeting next week.

John


********************************************************************************
FREXAS: 
   Functions: 
       (1) READ1 and READ1-READ2 frame source extractions.
       (2) READ1 photometry
       (3) Frame statistics

   Outstanding issues/tasks:
       The outstanding issues in FREXAS all concern using the PSF grid to
       extract source positions. Most if not all of these concerns likely
       deal with small errors (relative to the Level 1 specs), but nonetheless
       need to be quantified.

          A. The PSF grid vs. seeing is currently incomplete, and is
             time-consuming to fill out the grid using data only. As
             an alternative, the PSF grid can be filled out by smoothing
             PSF's made under good seeing with a symmetric Gaussian 
             function.

          B. Since PROPHOT determines positions using data-generated
             PSF's (which is also not a complete grid at the moment),
             there is potential for systematic errors between FREXAS 
             and PROPHOT positions. Preliminary tests indicate the bias 
             between FREXAS PSF positions (using PSF's generated from 
             protocamera data) and PROPHOT PSF's is about 0.05-0.1".
             Since neither PSF grid is optimal at the moment, this
             error is likely to smaller once the PSF grids are complete.

          C. Eventually, FREXAS will be tested with the PROPHOT PSF's
             once that grid is complete. 
  
          D. FREXAS uses a single PSF over a 6 degree strip to 
             determine source positions, while PROPHOT allows the
             seeing/PSF to vary over the scan length. Again this error
             is expected to be small, but remains needs to be tested
             and quantified.

          E. The focus will not be perfect in all three bands. This 
             increases the dimensionality of the problem since 
             (i)  the symmetry of the PSF will depend on seeing and focus, and
             (ii) the PSF may vary spatially across the chip and
                  may lead to focal-plane dependent position errors.
             Again these errors are likely to be small for FREXAS, but 
             need to be quantified.

PIXPHOT/PICMAN: 
   Functions: 
       (1) Assembles and registers Atlas images from calibrated frames
       (2) Deep source detection

   Outstanding issues/tasks:
          A. Need to determine how the FWHM of the gaussian filter used for
             source detection depends upon the seeing parameter. (FREXAS 
             returns a PFRAC parameter, which is an indirect measure of the 
             current seeing.)

          B. The cosmic ray rejection algorithm needs to be refined. Concerns
             have been raised on how well the current algorithm works in 
             source confused regions such as the galactic plane.

          C. The SIGLO/SIGHI/DETHR parameters are used to select the thresholds
             for detecting stars in the coadded images. Are a single set of
             parameters suitable both for the galactic plane and high galactic
             latitudes? Ultimately, these parameters must be fine-tuned at high 
             galactic latitudes where completion/reliability are defined by 
             level one specs.

********************************************************************************