I think the question is a little misleading. If you are taking long exposures of an stellar spectrum then the only "noise from the atmosphere" you have to account for absorption, refraction and dispersion, which is basically what Andrew answered (for more details see the works of Vacca, Cushing or Rayner, 2003, where they plot the spectrum that you are asking for I think).
However, if you are actually taking short exposures and want to correct for the real noise that produces changes on stellar spectra because of climate changes, probably talking of red noise, which have been a concern for years for photometry and is today being used in spectroscopy (see the work of Pont et al, 2005 for a short introduction).
If no answer is useful for you, maybe you'll have to give us details in what you want (and please, if you saw this on a magazine, post the link to the article or to some preprint that we could see in order to help you. For example, I'm not actually subscribed to the Astronomy magazine so I really don't know what figure caught your mind).