OH level populations and accuracies of Einstein-A coefficients from hundreds of measured lines

OH airglow is an important nocturnal emission of the Earth's mesopause region. As it is chemiluminescent radiation in a thin medium, the population distribution over the various roto-vibrational OH energy levels of the electronic ground state is not in local thermodynamic equilibrium (LIE). In order to better understand these non-LIE effects, we studied hundreds of OH lines in a high-quality mean spectrum based on observations with the high-resolution Ultraviolet and Visual Echelle Spectrograph at Cerro Paranal in Chile. Our derived populations cover vibrational levels between v = 3 and 9, rotational levels up to N = 24, and individual Lambda-doublet components when resolved. As the reliability of these results critically depends on the Einstein-A coefficients used, we tested six different sets and found clear systematic errors in all of them, especially for Q-branch lines and individual Lambda-doublet components. In order to minimise the deviations in the populations for the same upper level, we used the most promising coefficients from Brooke et al. (2016) and further improved them with an empirical correction approach. The resulting rotational level populations show a clear bimodality for each v, which is characterised by a probably fully thermalised cold component and a hot population where the rotational temperature increases between v = 9 and 4 from about 700 to about 7000 K, and the corresponding contribution to the total population at the lowest N decreases by an order of magnitude. The presence of the hot populations causes non-LIE contributions to rotational temperatures at low N, which can be estimated quite robustly based on the two-temperature model. The bimodality is also clearly indicated by the dependence of the populations on changes in the effective emission height of the OH emission layer. The degree of thermalisation decreases with increasing layer height due to a higher fraction of the hot component. Our high-quality population data are promising with respect to a better understanding of the OH thermalisation process.