3.2.2 The Extractions

You are now ready to begin the extractions. If you can combine your object frame at this point, it will save you a fair amount of work; use imcombine. However, if you've tweaked the telescope between individual exposures you should first check using implot that the objects line up exactly. If not, you should treat your frames separately and combine after they are extracted. For the purposes of this outline we will assume that you have three object frames ( object1, object2, object3), one quartz-lamp frame ( flat1), and one comparison spectrum ( comp1). We assume that you are starting with "fresh" parameters, i.e., the default settings of specred, and explicitly refer only to parameters that we would change.

• Create an "aperture id" table ( apidtablname)for each slitlet mask in the form of "1 1 objectname1", "2 2 objectname2", one per line.

• apall object1,object2,object3 reference"" find- trace+ b_nave=-60 background=fit weights=variance clean=yes readnoise=rdnoise gain=gain apidtab=apidtabname t_nit=1 width=10 Extract the object spectrum from each aperture. In this task, the region for the object within the aperture is set as well as the region to be used for sky subtraction. The "width" parameter should be the base-to-base width of an object spectrum. Once in the task, you will be given a chance to define the apertures using "m" (proceed left to right"), and to adjust the background regions ("b"). After this stage, the spectra will be traced.

  The result is an image which contains the object spectrum for each aperture with variance weighting, the object spectrum without variance weighting, the sky spectrum for each aperture, and the error spectrum for each aperture.

  If you have only very weak signal in your object frames, use the trace of the quartz exposure as the reference, i.e., apall object1,object2,object3 reference=flat1 trace- recenter+ resize+ back=fit width=10

• apsum comp1 output=c1 reference=object1 recent- resize- trace- back- interact-

  apsum comp1 output=c2 reference=object2 recent- resize- trace- back- interact-

  apsum comp1 output=c3 reference=object3 recent- resize- trace- back- interact-

  Extract the comparison spectra from the comparison image using the aperture and trace of each object spectrum.

• identify c1 Run identify on the first comparison spectrum generated by apsum, You'll be presented with the comparison spectrum of the middle aperture: use "m" to mark and identify a few good comp lines and then an "f" to do a fit. Change the fitting function using :function spline3 and the order using :o 3; a "q" returns you to identify. An "l" will find the remaining lines. You can then zoom using "z" and step through the lines using "+" and "-"; delete blends using "d". A new "f" will make a new fit. Next, do solutions for all the other apertures: a "k" or "j" will change to the next aperture up or down. Identify a prominent line and type an "s" at its position to shift the solution. Do a new "f". If need be, reinitialize using a "i" and begin using "m". When all the apertures are correctly identified and fit, exit using "q".

• reidentify c1 c2,c3 newaps- refit+ verbose+ override- This will make the slight adjustments to the comparisons obtained by using the object2 and object3 as the reference image as above.

• refspect object1.ms reference=c1 ignoreap- verbose+

  refspect object2.ms reference=c2 ignoreap- verbose+

  refspect object3.ms reference=c3 ignoreap- verbose+

  This puts the identification of the right comp into the headers of the object spectra.

• dispcor object1.ms,object2.ms,object3.ms samedisp- global+ ignoreap- confirm+ verbose+ This dispersion corrects the data.

• scombine object1.ms,object2.ms,object3.ms output=final.ms reject=avsigclip scale=mode weight=mode Combines the three spectra into one, weighting by the signal. Many other rejection options are available; if you have only two spectra, you can use "ccdclip"; read the help page.