Mechanisms for varying non-LTE contributions to OH rotational temperatures from measurements and modelling. I. Climatology

Rotational temperatures Trot from OH line intensities are an important approach to study the Earth's mesopause region. However, the interpretation can be complicated as the resulting Trot are effective values weighted for the varying OH emission layer. Moreover, the measured Trot only equal kinetic temperatures Tkin if the rotational level population distribution for the considered OH lines is fully thermalised. In many cases, this basic condition of a local thermodynamic equilibrium (LTE) does not seem to be fulfilled. In order to better understand the non-LTE temperature excesses ΔTNLTE and their variations, we used Trot measurements based on 1526 high-resolution spectra of the UVES spectrograph at the Very Large Telescope at Cerro Paranal in Chile in combination with Tkin weighted for the OH emission layer based on 4496 nighttime temperature and OH emission profiles from the SABER radiometer onboard TIMED taken at a similar location. Both data sets were linked via climatologies consisting of the nighttime and seasonal temperature variations. The study focusses on the non-LTE effects at the vibrational level v=9, which is directly populated by the OH-producing hydrogen-ozone reaction and therefore especially prone to incomplete thermalisation of the rotational level population. In comparison to the less critical v=3, the ΔTNLTE climatology showed clear and strongly variable temperature excesses of several kelvins with minima in the evening around the equinoxes and a reliable maximum in the second half of the night around the turn of the year. The Trot non-LTE contributions are positively correlated with the effective OH emission height and volume mixing ratio of atomic oxygen. A significant anti-correlation is found for the air density. Thus, especially variations in the OH emission layer altitude and shape, which are related to changes in the layer-weighted chemical composition and density, are important for the amount of ΔTNLTE.