Evaluation of the diurnal cycle in the atmospheric boundary layer over land as represented by a variety of single column models - the second GABLS experiment

This paper presents the main results from the second model inter-comparison within GEWEX Atmospheric Boundary Layer Study (GABLS). The target is to examine the diurnal cycle over land in today’s numerical weather prediction and climate models for operational and research purposes. The setup of the case is based on observations taken in Kansas, USA in the early autumn with a strong diurnal cycle with no clouds present. The models are forced with a constant geostrophic wind, prescribed surface temperature and prescribed large-scale divergence.
Results from thirty different model simulations and one Large Eddy Simulation are analysed and compared with observations. Even though the surface temperature is given, the models give variable near surface temperatures. This, in turn, gives rise to differences in low-level stability affecting the turbulence and the turbulent heat fluxes. The increase in upward sensible heat flux during the hours after the morning transition is too weak and the growth of the convective boundary layer before noon is far too slow. This also results in too weak winds during the morning hours. The agreement between the models, the LES and observations is best during the late afternoon.
From this intercomparison study, no clear signal emerges of which turbulence closure that gives superior results. For the convective part of the diurnal cycle, the performance of the first order schemes with high vertical resolution (with a few exceptions) seems to be somewhat better and the TKE schemes with high vertical resolution tend to be slightly better during night-time conditions.