The Third GABLS SCM Intercomparison and Evaluation Case

For the third GEWEX Atmospheric Boundary Layer Studies (GABLS) a case with realistic geostrophic forcings and dynamical tendencies is set up such that a quantitative comparison with observations should be possible. From the Cabauw data archive a relatively simple clear sky boundary layer case has been selected for a full day starting July 1, 2006. The case shows a clear decoupling at sunset, a well-defined inertial oscillation resulting in a low-level jet during the subsequent night and a well-defined morning time transition. In total 18 state of the art but different single column models from various research groups participated in the intercomparison study. Straightforward comparison of the model results shows a large variation in night time minimum temperature. Most of the models attain lower temperatures then observed. Also a large variation in wind speed maximum is observed. All models underestimate the speed of the nocturnal low-level jet.

The development of the atmospheric boundary layer results from rather complex interactions between turbulence, radiation and surface exchange, in particular for stable conditions. Contrary to previous GABLS intercomparison studies, for the current study the full interactions between these model components are taken into account. In this way we anticipated to find realistic results, but it appears that the differences among models and between the models and the observations are rather difficult to interpret. Models may differ from observations because they don't mimic the characteristic boundary conditions of the site well enough or they may differ because they mis-represent more fundamental physical processes. In this study we aim to discriminate between model representation of the site and of the physical processes by running one of the models many times with different parameter settings. It is, for example, shown that by varying the coefficient that regulates the thermal coupling between the atmosphere and the surface, this single model spans up a large part of the variation that is observed among the 18 models in this intercomparison study.