At this stage of data reduction, the dispersion axis must be wavelength calibrated. Ideally one would use the arc frame obtained at the beginning of the night. One would identify the arc lines in the spectrum and since their wavelength is known, a one-to-one correspondence between pixel number along the dispersion direction and wavelength could be obtained by fitting a polynomial to the pairs (pixel number,wavelength). This calibration would be copied to the dispersion axis and used in all the subsequent frames. Performed in this way, the spectral images could have been wavelength calibrated at the observation level.
A couple of points must be brought to the attention though. Firstly,
when using the echelle (high resolution spectroscopy) not very many
lines are present in the arc spectrum. Two to three lines is a typical
number, sometimes only one is available!. From this, only a very
unsatisfactory wavelength calibration can be achieved, if at
all. Fortunately, there are lines in the spectra
of the astronomical sources that can be used in wavelength
calibration, namely OH sky emission lines and telluric absorption lines.
Usually these lines are considered to be at best
superfluous and at worse detrimental
. Secondly, especially
at high resolution, the wavelength calibration might change during the
night due to flexure of, or in, CGS4. If this occurs a single arc spectrum
taken at the beginning of the night (even if it had a reasonable number
of lines) is
not good enough to wavelength calibrate all subsequent frames. The
description of how the spectra presented here were wavelength calibrated
will be given in Sections 3.2 and 3.3, dedicated to
the particulars in data
reduction for data obtained in different observing runs and at different
wavelengths.