In the spring of 2005 Frank Grundahl and myself attended a workshop in Uppsala where we learned how to reduce Échelle spectra. The REDUCE code is written mostly in IDL by Nikolai Piskunov (Uppsala) and Jeff Valenti (STScI). I have successfully reduced spectra from UVES@UVES and achieved S/N ratios that were 30% higher than a reduction that Jens Viggo Clausen made with IRAF.

This week I have been reducing new data from the FEROS@ESO-2.2m spectrograph. Everything has run smoothly except for the wavelength calibration. I thought I'd share some information here that other people may use in their reduction of FEROS spectra.

The first thing I did was to find the approximate wavelength range of one Échelle order. I did this by identifying the water vapor lines in the spectra. There are several groups but I picked the ones around 6275 ångström in relative order number 25 (the order just next to the one containing H-alpha at 6562.79 ångström). There are also two strong Fe lines in the order that I used to find the approximate scale (ångström/pixel) in order to infer the wavelength range in that spectral order. In the program wavecal.pro I could not immediately recognize the lines in Thorium/Argon (Th/Ar) atlas. I then swapped the input array so wavelength increased from left to right instead, and then I could recognize the Th/Ar lines. I was then able to improve on the wavelength scale and go on to the next Échelle order.

My next problem was that I couldn't get the 2D solution to work even after fitting 10 orders manually with a 1D solution. The orders that I had fitted 1D solutions to manually all had residuals of exactly 0.0 m/s in the 2D fit. I think the problem was that I didn't supply wavecal.pro with the right absolute Échelle order. However, I was able to deduce it (within ±1 order) from the documentation on FEROS.

The image below is a screenshot from the wavecal.pro widget (right click to get full resolution image). I have indicated the absolute order with yellow text and also the wavelength range it covers (accurate within ±0.5 ångström). The red arrows mark the boxes where you have to type the absolute Échelle order of the base order (ie. relative order = 0). I may add, as a footnote, that there are more than 37 Échelle orders on the CCD, any you may be able to extract up to 39 or even 40 (partial) orders with the REDUCE program. However, the S/N in the orders at the top/bottom of the CCD will be quite low, so I don't recommend to bother trying to include them.

No comments.