IMAGING DATA DESCRIPTION
========================

Thomas Erben, 21.03.2013
(e-mail: terben@astro.uni-bonn.de)

AVAILABLE DATA PRODUCTS
=======================

- subdirectory coadd_VXXX (XXX: version number of the data):

  For each 1 sq. deg. pointing we offer FITS image pixel data for
  the available broad-band filters. Our FITS files are
  named like POINTING_FILTER.VXXX.swarp.cut.(weight., flag., sum.)fits.
  The name contains the following information:
  
  POINTING:   name of the pointing
  
  FILTER:     filter of the observation (possible are the five MegaPrime
              broad-band filters. See the accompanying file
              README_MEGAPRIME.txt for more information.
  
  VXXX:       internal THELI version number; see the accompanying file
              README_MEGAPRIME.txt for more information.
  
  swarp:      the image was co-added with swarp
  
  cut:        all images of a given pointing are cut to the same size
              (21k x 21k) and all filters are projected to exactly 
              the same pixel grid!
  
  (weight.,flag.,sum.): 
              for each pointing/colour we offer the co-added science
              image, a corresponding weight map, a flag mask and a
              so-called sum image. See Erben et al. 2005, AN 326, 432)
              and Erben et al. 2009 for a more detailed description of
              the image weights.  The flag image has a '0' where the
              weight is unequal to zero and a '1' where the weight is
              zero, i.e. a '1' indicates a 'bad pixel' of the co-added
              science frame.  The sum image gives, for each pixel, how
              many individual images contributed to that pixel.
  
              !! IMPORTANT: !!
              The SExtractor WEIGHT_TYPE for all provided weight images 
              is MAP_WEIGHT
  
              The science images are offered in uncompressed form
              and have a size of about 1.7GB. Weight, Flag and Sum images take
              about 40MB in gzipped form. The weights uncompress to the same
              size as corresponding science frames, the flag and sum images to
              about a quarter of this (integer pixel values).
  

  IMAGE HEADERS
  =============
  
  Image headers contain all necessary astrometric and photometric
  information to extract object catalogues; see Erben et al. 2009 for a
  discussion on the quality of our astrometric and especially of our
  photometric calibration. Important modifications with respect to this
  work are described in Hildebrandt et al. 2011 (arXiv:1111.4434).
  Astrometric header keywords follow standard
  WCS descriptions (see e.g.  Greisen & Calabretta A&A 395, 1061).  We
  discuss here some important header keywords to obtain photometric
  quantities:
  
  - TEXPTIME: total exposure time in seconds
  - EXPTIME:  'effective' exposure time in seconds (always 1s for CARS; 
              the pixel unit of all CARS images is ADU/s)
  - MAGZP:    magnitude zeropoint; apparent object AB magnitudes need to be
              estimated via:
  
              mag = MAGZP - 2.5 * log10(object counts);
  
              If using SExtractor 'correct' apparent AB magnitudes are
              obtained by putting the MAGZP value into the
              MAG_ZEROPOINT configuration parameter.
  - GAIN:     The effective GAIN of the exposure. To obtain meaningful
              magnitude error estimates within SExtractor the GAIN
              configuration parameter needs to be set the the GAIN
              header value!
  - SEEING:   measures 'mean' image seeing for the exposure. Put this
              value into the SEEING_FWHM SExtractor parameter to
              obatin a meaningful SExtractor star/galaxy separation.
  
  The given estimates for magnitude ZP errors (keyword MAGZPERR) for the
  different filters are discussed in Erben et al. 2009. 
  
  IMPORTANT: Note that MAGZP, GAIN and SEEING are in general different for
             each pointing/filter!
     
- subdirectory masks_VXXX (XXX: version number of the data):

  This directory contains image masks:
  All masks are in ds9 region format. Masks with the ending '_wcs.reg'
  are in WCS-coordinates. All other masks are in pixel coordinates
  of corresponding images.
  In the case of multi-band data, usually only one band (the major
  science band) is masked.
  The end user typically only deals with the 'candidate' or
  'lensingcandidate' masks. The 'lensingcandidate' version covers
  areas around bright stars 'very' conservatively.  

  - POINTING_stars.reg:
    masks around bright stellar sources.

  - POINTING_FILTER_voids.reg
    Masks covering significant overdense are underdense regions of an
    image. Such regions typically denote artefects in teh image
    (bright stars etc.). But beware of rich galaxy clusters!

  - POINTING_FILTER_saturated.reg
    masks around saturated (ow weight zero) pixels that are not yet
    covered by POINTING_stars.reg or POINTING_FILTER_voids.reg

  - POINTING_FILTER_asteroid.reg
    masks around short asteroids that are not yet covered by one of
    the previous masks

  - POINTING_candidate_FILTER(_wcs).reg:
    A combination of all the above

  - POINTING_lensingcandidate_FILTER(_wcs).reg:
    A combination of all the above with a more conservative masking
    around bright stars

SUGGESTIONS, QUESTIONS, REQUESTS etc.
=====================================

For questions concerning the described products please contact:

Thomas Erben (terben@astro.uni-bonn.de)