STScI Logo

CALCOS v2.8.2 Release Notes

---------------- CALCOS v2.8.2 ------------------------------

relevant PRs:
62433, 62440, 62590

relevant Trac tickets:
394, 401, 406

There are four new command-line options, to compress and/or bin the
csum image and to allow the user to provide a text file to override
wavecal shifts in the dispersion direction.

Wavecal processing has been modified.  Chi square minimization is now
used instead of cross correlation to find the shift in the dispersion
direction.  Two values of this shift are now printed (they're usually
the same); one is the value originally determined (labeled "orig."),
and the other is the one that will be applied.  The latter could be
based on an average of the shifts of the other stripes, or it could
have been overridden by the user (see below).

The new command-line options are:

  --compress parameters (compress csum image)
  --binx X_bin_factor (csum binning in X)
  --biny Y_bin_factor (csum binning in Y)
  --shift filename (file to specify shift values)

The first three are for the csum file.  The fourth gives you the
option to specify the shift in the dispersion direction, overriding
the value that calcos finds.

The csum image (either FUV or NUV) may be binned in either the
X (dispersion) or Y axis (but not in the PHA axis).  If 'binx' or
'biny' is not specified, the default binning for that axis will be
used; currently the default binning is 1, i.e. no binning.
Independently of binning, the csum image may be compressed (this uses
the pyfits on-the-fly compression option).  The compression parameters
are given as a string consisting of two components separated by a
comma; the default value is "gzip,-0.1".  The first part is the
compression type, which may be "gzip", "rice", or "hcompress".  The
second part is the quantization level, which specifies how the
floating-point image values are to be converted to integer before
compression.  If the quantization level is positive, it is interpreted
as a value relative to the RMS noise level in the image background;
a value of 16, for example, means that the quantization level will be
1/16 of the noise level.  If the quantization level is negative, it's
the actual floating-point increment that corresponds to a difference
of one in the scaled integer image.  The compression in pyfits
supports other parameters such as tile shape, and parameters for
hcompress, but these are not currently supported in calcos.

The shift file (option --shift filename) may be specified in order to
override values of the wavecal shift (shift1a, shift1b, shift1c) in
the dispersion direction.  This is a text file with five columns:

  rootname   fpoffset   flash number   segment/stripe   shift1

The first four columns are used for identifying which particular lamp
exposure is to be overridden; however, any or all of those columns may
be given as "any" (without quotes), which matches any value.  All
strings are case insensitive.  Blank lines and lines beginning with
"#" will be ignored.  A single shift file can be used for an entire
association; that's why rootname is included as a selection criterion.
The rootname in this context is the portion of the file name that
precedes the suffix.  The rootname is taken from the actual name of
the raw file, rather than from the ROOTNAME keyword, so that if the
raw file has been renamed without changing the keyword, the name in
the shift file will be the name that most people would expect.  It is
redundant to specify both rootname and fpoffset.  Both are included
to make it easy to set shift1 to the same value for all rootnames for
a given fpoffset.  If rootname is specified, fpoffset can be given as
"any", and vice versa.