Response to the SNAP calibration review panels's report
                    of 08 Sep 2006

                      Michael Richmond
                        13 Sep 2006
                        14 Sep 2006

  My task: reply to the report in the areas of flatfielding and filters.


>  One unknown that they should think of examining is how small-scale
>  variations in  filter properties (such as micro-bubbles) will add to the 
>  error budget. The simulation effort will be important, as  the  observed
>  supernova spectra will be different than either the calibration stars or
>  internal LED calibration sources. 

     I know nothing about the properties of micro-bubbles in filters.
     So, first, I would have to find out

             a) how large are they?
             b) how common are they?
             c) what is their effect on light as a function
                      of wavelength?

     Once I know that information, it is possible that I could
     devise a simulation to investigate the effects of micro-bubbles
     on photometry of stars, and on photometry of supernovae.
     It is crucial to know how common the bubbles may be.

     It is also possible that a proper answer requires detailed 
     simulations using the end-to-end SNAP simulation software,
     which I do not understand at the moment.


>  Intra-pixel  quantum  efficiency  variations will most likely be significant
>  and need to be characterized both in the lab and on-orbit. A detailed  plan
>  for measuring intrapixel variations needs to be developed.

     The review panel has ignored the "Dithering and Mowing" 
     section (Section 4.5, pages 55-56) of the Calibration document, 
     which discusses exactly this point.  


>  With both the detector and filter testing, it is important to scope how much
>  testing needs to be done in the laboratory and on-orbit. In particular, the
>  applicability of spot calibrations (such as stellar flats) to the entirety
>  of  a  focal plane  needs to be justified.  

     Section 4.6 the Calibration document discusses at length
     the evidence that "stellar flats" improve the quality of
     photometry across a large field.  It includes references 
     to papers in the refereed technical literature.   

         page 60: Manfroid (1995) shows
                   that ground-based telescopes need to use
                   stellar flats to achieve photometric accuracy
                   below the one percent level.

         page 60: vanderMarel (2003) describes how the HST 
                   team has been using stellar flats (called
                   "L-flats" in their description) for years
                   to improve the photometry across the field
                   of ACS.  A more recent version of this document
                   is "ACS Flat Field Update and New SBC L-flats",
                   by Mack, Bohlin, Gilliland and van der Marel,
                   in "The 2005 HST Calibration Workshop," eds.
                   Koekemoer, Goudfrooij, and Dessel.
                      

>  The baseline strategy is good, but the error introduced by color  
>  variations  from  the observed flat to the observed supernova 
>  needs to estimated.  

     This can be addressed either by a standalone set of simulations,
     or by the SNAP simulation pipeline.   It requires as input
   
        a) detailed knowledge of how the pixel-to-pixel sensitivity
           of CCD and IR detectors varies with wavelength