In this study, we analyse 24 years of atmospheric temperatures from the mesopause region (~87 km altitude) derived from ground-based spectrometer observations of hydroxyl airglow at Davis station, Antarctica (68° S, 78° E). These data are used to quantify the effect of the solar cycle and the long-term trend due to increasing greenhouse gas emissions on the atmosphere at this level. A record-low winter-average temperature is reported for 2018 and comparisons are made with satellite observations.

Line emission from hydroxyl (OH) molecules at altitudes of about 90 km strongly contributes to the Earth's night-sky brightness and is therefore used as an important indicator of atmospheric chemistry and dynamics. However, interpreting the measurements can be ambiguous since necessary molecular parameters and the internal state of OH are not well known. Based on high-quality spectral data, we investigated these issues and found solutions for a better understanding of the OH line intensities.

Atomic oxygen is one of the most important trace species in the mesopause region. A common technique to derive it from satellite measurements is to measure airglow emissions involved in the photochemistry of oxygen. In this work, hydroxyl nightglow measured by the GOMOS instrument on Envisat is used to derive a 10-year dataset of atomic oxygen in the middle and upper atmosphere. Annual and semiannual oscillations are observed in the data. The new data are consistent with various other datasets.

Light emission from energetic hydroxyl radical, OH*, is a prominent feature in spectra of the night sky. It is routinely used to determine the temperature of the atmosphere near 90 km. This note shows that the common practice of using only a few emission features from low rotational excitation to determine rotational temperatures does not account for the bimodality of the OH population distributions and can lead to large systematic errors.

Satellite overlap is often carried out as a check on the stability of the data collected. We looked at how length of overlap influences how much information can be derived from the overlap period. Several results surprised us: the confidence we could have in the matchup of two records was independent of the offset, and understanding of the relative drift between the two satellite data sets improved significantly with 2–3 years of overlap. Sudden jumps could easily be confused with drift.