OH airglow observations with two identical spectrometers: benefits of increased data homogeneity in the identification of the 11-year solar cycle-, QBO-induced and other variations

Abstract. Hydroxyl (OH) radical airglow observations have been performed at the environmental research station ‘Schneefernerhaus’ (UFS, 47.42° N, 10.98° E) since October 2008, and with continuous operation since July 2009. The instrumental setup relies on the parallel operation of two identical GRIPS (Ground-based Infrared P-branch Spectrometer) in order to achieve maximum completeness and homogeneity.

After the first decade of observations the acquired time series are evaluated with respect to the main influences on data quality and comparability to those on other sites. Data quality is essentially limited by gaps impacting the completeness. While technical failures are largely excluded by the setup, gaps caused by adverse meteorological conditions can systematically influence estimates of the annual mean. The overall sampling density is high, with nightly mean temperatures obtained for 3382 of 4018 nights of observation (84 %), but the average coverage changes throughout the year. This can bias the annual mean up to 0.8 K if not properly accounted for.

Sensitivity studies performed with the two identical instruments and their retrieval show that the comparability between the observations is influenced by the annual and semi-annual cycle as well as the choice of Einstein-A-coefficients, which influence the estimate of the annual cycle’s amplitude.

A strong 11-year solar signal of 5.9 ± 0.6 K / 100 sfu is identified in the data. The OH temperatures follow the F10.7cm value with a time lag of 90 ± 65 days. However, the precise value depends on details of the analysis. The highest correlation (R²=0.91) is achieved for yearly mean OH temperatures averaged around 4th February and the F10.7cm solar flux leading ahead with 110 days. A prominent two-year oscillation is identified between 2011 and 2015. This signal is linked to the quasi-biennial oscillation (QBO) leading to a temperature reduction of approximately 1 K during QBO westward phases in 2011, 2013, 2015 and a respective 1 K increase in 2012 and 2014 during QBO eastward phases. The amplitude of the semi-annual cycle shows a similar behavior with the decade’s minimum amplitudes (~2.5–3 K) retrieved for 2011, 2013 and 2015 and maximum amplitudes observed in 2012 and 2014 (~4 K). The signal appears to disappear after 2016 when the solar flux approaches its next minimum. Although it appears as a rather strict 24-month periodicity between 2011 and 2015, spectral analyses show a more or less continuous oscillation with a period of approximately 21 months over the entire time span, which can be interpreted as the result of a non-linear interaction of the QBO (28 months) with the annual cycle (12 months).