Highly accurate spectroscopic models for O2 A-band absorption are required in numerous satellite- and ground-based remote sensing measurements of atmospheric greenhouse gases. In this study, we quantify the sensitivity of the calculated A-band transmission spectrum to the choice of line profile and model-input line parameter data. The models considered include the Voigt and Galatry profiles for the individual A-band transitions as well as line mixing and collisional induced absorption. We find that a widely used algorithm for calculating the Voigt profile (Humlicek 1979) is subject to a significant truncation error that leads to incorrect, zero-valued absorption cross sections in the far wings. Use of this algorithm leads to increased transmission (~0.02) and introduces an airmass-independent bias of ~4% in pathlength or effective surface pressure compared to more robust algorithms that properly compute the far-wing behavior of the Voigt profile. This easily rectifiable error occurs when the Lorentzian width is less than the Doppler width, which for atmospheric conditions typically occurs for altitudes in excess of 11 km. In order of increasing relative importance, we also show that measurement uncertainty is influenced by: available line parameter data, Dicke narrowing, line mixing, and collisional induced absorption effects.
Geophysical Research Letters
Oxygen A-band, Cavity ring-down spectroscopy, Remote Sensing, Voigt Profile, Galatry Profile, Line Mixing