STELLA Data Reduction Pipeline Files

Version 3.5
- implemented box-car filtering and thin-plate spline fitting for MEstimateScatter

Version 3.4a
- updated README files

Version 3.4
- implemented a much faster algorithm for fitting the scattered light (MSubScatterThinPlateSpline)
- fixed MMakeNormFlat to use the linear fit of the interpolated profile from 2 swaths
- added parameter NPIXCUTPROFILE to MMakeMormFlat to cut off the outermost pixels of each aperture
  because sometimes they can be bad
- thin-plate-spline fitting requires 'boost', please refer to cpp/README on how to install boost on 
  your machine
- updated MCreateMakeFiles to include boost

Version 3.3a
- fixed stella/cpp/mcreatemakefiles/src/MCreateMakeFiles.cpp

Move the stella directory to your home path.
edit loginuser.cl:
  set scripts to ~/stella/ and imdir to <image_directory>
mv <iraf-dir>/loginuser.cl <iraf-dir>/loginuser.cl.bak
ln -s ~/stella/loginuser.cl <iraf-dir>/loginuser.cl
add ~/stella/bin to your $PATH variable:
in ~/.bashrc add the following line to the end of the file:
export PATH=~/stella/bin:$PATH

To install the external libraries like cfitsio, blitz++, gsl, boost, and possibly mathgl, 
and compile the C++ files please refere to the README file in the cpp directory.

Please note that most parameter files are not up to date and some parameters might be missing. 
However, each IRAF task will tell you if a parameter could not be found in the parameter file.
The master task is stall.cl, which encapsulates all tasks needed to reduce a standard spectrum.
Note that the STELLA pipeline was developed for spectra going up/down, not left/right. You can 
use the C++ programs MImRotate or MImTranspose to rotate/transpose spectra if needed.
Please also note that, before using the IRAF programs, you must create the reference files 
manually. Taking one (probably slightly outdated) parameter file from the parameterfiles directory
and changing it is a good start.
Regarding APERTURE WIDTHS and SLIT SPECTRA please note that for the flat-field normalization
the profile NEEDS to include the full slit image including the wings. If you see steps in the 
normalized flat your aperture width is not wide enough. For object extraction the aperture width 
should be set to only include the wings of the object spectrum and not the wings of the slit/sky. 
You can use MSetApWidth to change the width of the apertures defined in an aperture-definition file.
Setting the profile as wide as you can, before the slit's wings start to go down, is the best for 
sky subtraction.

                _
               / \  object
      ________/   \____________sky
     /                         \
    /                           \  slit   
___/                             \___scattered light subtracted background (should be zero)

|                                  | aperture width for flat-fielding

      |                      | aperture width for object extraction


      
* I should probably mention here that I ignore the IRAF database aperture-definition parameters 
x_min and x_max, which normally describe the pixels that should be used for the sky subtraction 
             _
            / \
           /   \
__________/     \___________
|        |   |   |        |
1        2   3   4        5

3: x_center
1: x_min < 0 relative to 3
2: x_low < 0 relative to 3
4: x_high > 0 relative to 3
5: x_max > 0 relative to 3

1-2 sky
2-4 object
4-5 sky

I only use positions 

1 - x_low as defined in the database file
3 
5 - x_high as defined in the database file

The best sky subtraction you should get if you use TELLURIC=3 when running optextract. This 
algorithm first determines the profile from CCD rows which are not affected by sky and then does 
a linear fit of the profile for the object plus a constant for the sky, so positions 2 and 4 are 
obsolete. However, as there are multiple iterations (MAX_ITER_SKY), this will take some time 
longer than the other algorithms which only look at the outer most pixels to determine the sky 
and subtract it before the profile fitting.
      

If you want quality-check plots you need to uncomment the line
//#define __WITH_PLOTS__
in the file stella/cpp/cfits/src/CFits.h and install the MathGL library.