12/15/93:
I have modified CCDPROC to more fully support scan table observations.  In
combination with the ability to have the number of scan rows encoded in the
image header automatically, this allows such data to be processed in a
fairly foolproof and documented way.

First if ccdproc.scancor=no then the NSCANROW keyword and nscan parameter
are ignored.  For actual scanned data this may be useful to override
things.  Otherwise the following steps are taken.  The logic is slightly
complex so that everything is done in the right order and only as needed.

The task wants to apply dark count and flat field calibration images which
have been scanned by the same number of rows.  [Zero calibration images are
assumed not to be scanned.  This made sense to me but if desired the zero
images can also be treated like the darks and flats.]  This is similar to
the way flat fields are checked for subset (filter/grating).  If the
appropriate dark or flat has not been scanned then it is scanned in
software; i.e. a moving average is taken over the unscanned image.

The number of scan rows is determined for each object being processed from
the NSCANROW keyword or appropriate translation in the header translation
file.  If this keyword is not found the nscan parameter value is used;
i.e.  it is assumed the object image has been scanned by the specified
amount.  This allows using the software in cases where the number of scan
rows is not encoded in the header.  In the case of dark and flat images if
NSCANROW is not found a value of 1 (unscanned) is assumed.

The set of possible calibration images (from the zero and flat parameters
or the list of input images) is searched for one which has been scanned
with the same number of lines as the object being processed.  If one is
found it is processed as needed before applying to the object.  If one is
not found then an unscanned one is sought.  It is an error if neither can
be found.  An unscanned image is first processed as necessary (overscan,
trim, etc.) and then scanned in software to create a new image.  The new
image has the name of the unscanned image with the number of scan lines
appended, for example Flat1.32.  It also has the NSCANROW keyword added as
well as a comment indicating the image from which it was created.  This
approach allows the calibration image to be created only once for each
different scan format and the number of scan lines may be changed for
different observations and the appropriate calibration made from the
unscanned image.

The following example shows how this all works.  There are four object
images using two filters and two scan row values and unscanned
zero, dark, and flats.

cc> dir
Dark.imh     FlatV.imh    obs019.imh   obs021.imh   pixels
FlatB.imh    Zero.imh     obs020.imh   obs022.imh
cc> hselect obs* $I,filter,nscanrow yes
obs019.imh      V       24
obs020.imh      V       32
obs021.imh      B       24
obs022.imh      B       32
cc> ccdproc obs* overscan+ trim+ zerocor+ darkcor+ flatcor+ scancor+
obs019.imh: Dec 15 17:53 Zero level correction image is Zero
Dark.imh: Dec 15 17:53 Zero level correction image is Zero
Dark.24.imh: Dec 15 17:53 Converted to shortscan from Dark.imh with nscan=24
obs019.imh: Dec 15 17:53 Dark count correction image is Dark.24.imh
FlatV.imh: Dec 15 17:53 Zero level correction image is Zero
FlatV.imh: Dec 15 17:53 Dark count correction image is Dark.imh
FlatV.24.imh: Dec 15 17:53 Converted to shortscan from FlatV.imh with nscan=24
obs019.imh: Dec 15 17:53 Flat field image is FlatV.24.imh
obs020.imh: Dec 15 17:53 Zero level correction image is Zero
Dark.32.imh: Dec 15 17:53 Converted to shortscan from Dark.imh with nscan=32
obs020.imh: Dec 15 17:53 Dark count correction image is Dark.32.imh
FlatV.32.imh: Dec 15 17:53 Converted to shortscan from FlatV.imh with nscan=32
obs020.imh: Dec 15 17:53 Flat field image is FlatV.32.imh
obs021.imh: Dec 15 17:53 Zero level correction image is Zero
obs021.imh: Dec 15 17:53 Dark count correction image is Dark.24.imh
FlatB.imh: Dec 15 17:53 Zero level correction image is Zero
FlatB.imh: Dec 15 17:53 Dark count correction image is Dark.imh
FlatB.24.imh: Dec 15 17:53 Converted to shortscan from FlatB.imh with nscan=24
obs021.imh: Dec 15 17:53 Flat field image is FlatB.24.imh
obs022.imh: Dec 15 17:53 Zero level correction image is Zero
obs022.imh: Dec 15 17:53 Dark count correction image is Dark.32.imh
FlatB.32.imh: Dec 15 17:53 Converted to shortscan from FlatB.imh with nscan=32
obs022.imh: Dec 15 17:53 Flat field image is FlatB.32.imh
cc> ccdlist *.imh
cc> ccdlist *.imh
Dark.24.imh[96,96][real][dark][][OTZ]:
Dark.32.imh[96,96][real][dark][][OTZ]:
Dark.imh[96,96][real][dark][][OTZ]:
FlatB.24.imh[96,96][real][flat][B][OTZD]:
FlatB.32.imh[96,96][real][flat][B][OTZD]:
FlatB.imh[96,96][real][flat][B][OTZD]:
FlatV.24.imh[96,96][real][flat][V][OTZD]:
FlatV.32.imh[96,96][real][flat][V][OTZD]:
FlatV.imh[96,96][real][flat][V][OTZD]:
Zero.imh[96,96][real][zero][][OT]:
obs019.imh[96,96][real][object][V][OTZDF]:
obs020.imh[96,96][real][object][V][OTZDF]:
obs021.imh[96,96][real][object][B][OTZDF]:
obs022.imh[96,96][real][object][B][OTZDF]:

Frank