The are many known atmospheric absorption features (or telluric absorption lines) in the wavelength range accessible to HIRES. Features of moderate strength occur at wavelengths beyond about 5900 Angstroms, (although the features below 6270 Angstroms are sometimes weak enough to be ignored.) There are also features below 3200 Angstroms (not usually observed with HIRES).

The strongest lines occur in the A-band (7600-7630), in the B-band (6860-6890), and also between 7170 and 7350 Angstroms. These lines are unresolved with HIRES (as is also the case with arclamp lines). HIRES is sensitive enough that the strongest features (e.g. A-band) can be seen in quartz lamp exposures due to the air contained within the HIRES "room" (usually negligible).

Telluric lines will vary in wavelength relative to the object spectrum due to the motion of the earth (see Heliocentric Velocity Correction ). These lines will also vary in strength with airmass.

Note that due to the heliocentric effect, it is possible that at some times of the year your favorite object narrow absorption line may fall within a telluric feature, and at other times of the year it may fall between features.

Start with the object spectrum and a suitable bright "standard" star with few intrinsic emission or absorption lines (e.g. a hot B type SAO star, it need not be a flux calibration standard) and do the following:

1. Continuum divide (or normalize) the star so that the continuum level is 1.0.
2. Set the regions of the star spectrum which do not contain atmospheric lines to exactly 1.0. (See List of Telluric Lines .)
3. Shift the wavelength scale by (A pixels) so that the telluric lines in the star line up with the lines in the object spectrum.
4. Apply an exponential correction to the flux level of the star: (new star flux) = (old star flux)^(B), where B is a function of the airmass in the object and star: (B = (object airmass)/(star airmass)).
5. Divide the object spectrum by the star spectrum.
6. If the result is unacceptable, multiply the object spectrum by the star spectrum, and repeat the last three steps varying the values of A and B until you get an acceptable result.

Note that it is difficult (impossible) to get a perfect correction (sometimes even an adequate correction). There is a limit to how well you can divide these lines which appear very sharp in HIRES spectra.

More documentation to follow...