GMOS Long-slit Reductions: It's an ADVENTURE!

Rewritten based on experience reducing LMC173-1 February 2025

Some resources:

• The Cookbook.
• In IRAF/gemini/gmos do a "gmosexample longslit"
• But best: The turorials (which is what this is basically modeled on).

Installation

• Just use the gemini 1.16.1 package that comes with the standard NOIRLab IRAF (or will, come Spring 2025).

Getting Sorted Out

• Download the data. (If the data are still propriatory, be sure to list the program name and enter the key/password in "My data" in the pulldown on the upper right side.)
• Put all of the cals, science, and bias frames in a subdirectory called "Raw"
• Make an observing log. The difficulty is that some header words are in [0] and others are in [1]. This gets around this. Here's a handy script: makelogs.cl. It creates obsLog.txt and (unlike the manual) retains the keywords. I found it handy to then write down in a notebook what the pertinent frames were.
• Make input text file lists.
  • The major issue is that because of all the MEF stuff it's hard to just make a list of file names.
    • Edit script biasSelect.cl. Creates bias.txt.
    • Edit script compSelect.cl Creates flats.txt, arcs.txt, etc.
    • Edit script sciSelect.cl Creates science.txt
    • Let us imagine that the data consistes of multiple nights of observations of a single star, taken as 3 consecutive spectra with both a flat and an arc. One should take flats.txt, arcs.txt, and science.txt and create fn1.txt, fn2.txt, fn3.txt....,an1.txt, an2.txt, and3.txt...sn1.txt,sn2.txt,sn3.txt
  • ALTERNASTIVELY you can just do a bunch of hand editing
    • files S*%.fits%% > arcs.txt
    • copy arcs.txt flats.txt
    • copy arcs.txt science.txt
    • and edit like crazy.
  • Or You can also just do something like:
    • gemlist "S20150503S" "275-278,280 > bias.txt

Everything all at once

Why, yes, there is a script for doing everything It will need editing. Below is what it really should say:

• set rawdir="Raw/"
• Make master bias frame:
  • gbias @bias.txt Zero rawpath=rawdir$ fl_over+ fl_trim+ fl_inter- fl_vardq+
• Do the basic processing (overscan, trim, and bias subtraction). Note that how some of the flags interact are...subtle...and so pay attention here.
  • unlearn gsreduce
  • #Flat
  • gsreduce @flats.txt rawpath="rawdir$" fl_bias+ fl_flat- fl_gmosaic- fl_cut- fl_gsappwave- bias="Zero.fits"
  • #Science
  • gsreduce @science.txt rawpath="rawdir$" fl_bias+ fl_flat- fl_gmosaic- fl_cut- fl_gsappwave- bias="Zero.fits"
  • #Arc. Reduce AND mosaic it.
  • gsreduce @arcs.txt rawpath="rawdir$" fl_bias+ fl_flat- fl_gmosaic+ fl_fixpix- fl_gsappwave- fl_over+ bias="Zero.fits"
• Do the wavelength solution on the mosaic'd arcs and then transform the arc.
  • gswavelength gs//@arcs.txt fl_inte+ fl_addf- function="spline3"
  • gstransform gs//@arcs.txt wavtran=gs//@arcs.txt
  • implot the resulting "t" frame and make sure everyting is nicely lined up. For instance, implot tgsblah.fits[SCI] and then :l 150 o :l 350.
• Do the QE corrections for the flat and the science exposures. These aren't mosaic'd yet. At this point we need a one-to-one correspondsponce between the science frames and the arcs. It's time to break things to use the little text files created above. If there are three observations and then a flat and an arc, that works: just make sure you have corresponding lists.
  • #QE the science exposure: Need the wavelength frames as reference because the QE is a function of wavelength. Note tha the gqe does take a NY minute or two.
    • gqe gs//@sn1.txt ref=gs//@an1.txt
    • gqe gs//@sn2.txt ref=gs//@an2.txt
    • gqe gs//@sn3.txt ref=gs//@an3.txt
    • etc.
  • #QE the flats:
    • gqe gs//@fn1.txt ref=gs//@an1.txt
    • gqe gs//@fn2.txt ref=gs//@an2.txt
    • gqe gs//@fn3.txt ref=gs//@an3.txt
    • Actually since there's a one-to-one correspondance already, you could just do the following as long as everything is in the correct order. It's a big if: gqe gs//@flats.txt ref=gs//@arcs.txt
• Make the flat-field. I don't like interpolating over the chip caps so I prefer the second of these: fl_fixpix- Note. At least with the 1200 grating and our high S/N data I don't find that it matters.
  • gmosaic qgs//@flats.txt fl_fixpix-
  • gsflat mqgs//@fn1.txt FlatN1.fits order=23 fl_over+
  • gsflat mqgs//@fn2.txt FlatN2.fits order=23 fl_over+
  • gsflat mqgs//@fn3.txt FlatN3.fits order=23 fl_over+
• Reduce the science exposure, mosaic'ing it in the process:
  • gsreduce qgs//@sn1.txt flat="FlatN1" fl_over- fl_bias- fl_trim-
• Apply the wavelength calibration. Note the extra "gs" at the start that got added abover:
• gstransform gsqgs//@sn1.txt wavtran=gs//@an1.txt
• gstransform gsqgs//@sn2.txt wavtran=gs//@an2.txt
• gstransform gsqgs//@sn3.txt wavtran=gs//@an3.txt