Single-column modeling versus mesoscale modeling for an episode with three diurnal cycles in CASES99

From GABLS2 and previous intercomparison studies it is obvious that single column models are a useful tool to study specific parts of the physics. For selected cases where the dynamic forcing plays a minor role, results from a single column model can be useful for the development of the physics. 1D experiments gain insight in the performance of the physics but how much we can learn from them in the perspective of correct dynamic forcings and the selection of the case-study?

Following the GABLS2 single-column intercomparisons, we evaluate a 3-D version of HIRLAM on its performance in the atmospheric boundary layer and compare the 3D results with the single-column model results.The single-column model was run according to the specifications from the GABLS2 experiment. HIRLAM uses the maximum horizontal resolution for a hydrostatic model (10 km) and a vertical resolution which is close to the operational setting. A sensitivity experiment with a higher vertical resolution has been conducted with the single-column model. First a single 3D-run has been made where soil moisture was obtained from climatology. A second 3D-run was also made in which the soil moisture has been adapted to surface observations by a surface analysis scheme over a larger number of cycles. These two experiments show the importance of a correct division between sensible and latent heating for almost all parameters within the boundary layer.

Results from 3-D HIRLAM divert significantly from the single column version. Wind and temperature profiles from the 3-D runs are in better accordance with the observations. For instance the representation of the Low-Level Jet is surprisingly good in the 3D experiment. We try to clarify the differences between the 3D and 1D experiment.

The selection of a period with only small dynamical tendencies is crucial for the comparison of 1D modeloutput with observations. The geostrophic wind is an important factor for the development of the windprofile in the nocturnal boundary layer. It is clear from the 3D study that the dynamic forcing, and thereby the change in the geostrophic wind, is not small in this experiment. Therefore it makes sense that results from the 1D model differ from the 3D model. Consequently it is doubtful to compare 1D results directly with observations. The use of other cases with smaller dynamic forcing will enable a more direct comparison of the observations with the 1D model output and may help us to learn more from 1D model intercomparisons.