2. Data Reduction Guide with IRAF Examples

We thank John Carr (O.S.U.) for contributing the lion's share of these notes on 
CSHELL data reduction.  John wrote nearly all of the following material on point 
source data reduction.

Point Source Data Reduction

Note:	Data should have been taken in nod mode (AB or ABBA, object is in 
both A and B beams and  nodding along the slit)

1) Flat Fields
a) combine the flat-field images: e.g. IRAF function 'imcombine' with avgsigclip
b) combine dark images: e.g. IRAF 'imcombine'
c) flat = a) - b) :   e.g. IRAF 'imarith'
d) fix badpixels using the badpixel mask: e.g. use fix.cl IRAF script
e) normalize the flat-field, producing a frame called 'normflat'

2) Object Data
a) form differences A1-B1, A2-B2, ...B1-A1, B2-A2, ...:  e.g. IRAF 'imarith'
b) combine the differences for each beam to give an A image and a B image:  	
	e.g. IRAF 'imcombine'
c) divide each by the normalized flat-field:   e.g. IRAF 'imarith'
d) correct for badpixels with the mask: e.g. use fix.cl IRAF script
e) Examine the images for any additional bad pixels or other problems
f) You may save disk space by stripping out NICMOS image areas where there is no 
data 
	e.g. use strip.cl IRAF script

3)  Do the same for the standard star data, if any.

4) Extract the Spectra
The IRAF noao.twodspec.apextract package is well-suited to this. The three 
extraction steps can be done interactively at once with the APALL task 
(dispaxis = 1) :
a) EDIT  the aperture and width (and any background regions subtract)
b) TRACE the aperture, defining its position as a function of position across the 
frame
c) EXTRACT  the one-D spectrum from the frame

5) Extract Calibration Spectra (based on spectral lamp data)
Assuming n>=1 images with n>=2 lamp lines (often in different orders):
a) combine the images to be used: e.g. IRAF 'imcombine' or 'imcopy'
b) extract a calibration spectrum for each beam (A,B) such that the calibration data 
are
 extracted from the same part of the image as the object data.  Do this using APALL, 
setting 'referen' and 'profile' to the object data image, and do not edit or trace 
the 
aperture.

We now have 1-dimensional spectra

6) Wavelength Scale Data (using  noao.onedspec IRAF tasks)
a) for each line in calibration spectra, convert the line wavelength to the 
equivalent 
wavelength in that order (done in find_lines program)
b) use 'IDENTIFY' to fit a wavelength scale to the spectrum
c) use 'REFSPEC' to assign that wavelength scale to object spectrum
(careful to apply the A calibration to the A spectrum, etc.)
d) use 'DISPCOR' to produce a dispersion correct spectrum for each beam

7) Reduce the standard star data in the same way

8) Divide each object spectrum beam by the appropriate standard star beam

9) Combine All Spectra 
Since the A, B beams will have, in general, different wavelength scales, 
use the function 'COMBINE' (noao.onedspec IRAF task), which will resample 
the spectra and average them.  This is the final spectrum.

10) Plot the final spectrum using SPLOT (noao.onedspec IRAF task), which 
can compute equivalent widths, display wavelenght and intensity units, etc.


Extended Emission

If your object is extended along the length of your slit, then the standard point 
source spectral extraction technique outlined above will probably not work 
optimally. First perform the  frame arithmetic outlined in steps 1) - 3) above. 
Divide the object and standard star data by the normalized flat. Telluric 
correction: you may divide the flat-fielded object data by a flat-fielded standard 
star frame which has been artificially extended along the slit direction, but it is 
best to rotate the images first so that the slit is exactly perpindicular to the 
dispersion axis.  Additional background subtraction can be done with BACKGROUND 
(IRAF noao.twodspec.longslit task), and then you should register all frames by 
moving them along the slit direction. IDENTIFY  (IRAF noao.twodspec.longslit task) 
will allow you to measure the centroid of flux peaks interactively, and IMSHIFT 
(IRAF images task) allows you to shift the images to a common center. Combine all 
shifted images with IMCOMBINE (IRAF images task). What you do from here depends on 
what you want to get out of the data as well as its morphology.