Evaluating the COSMO model without knowing the truth

Ronny Petrik

Max-Planck-Institute of Meteorology, Ozean in the Earth System, Hamburg, Germany

Two different diagnostic methods are applied for evaluating the physical quality of the mesoscale model COSMO. Firstly, a finite volume diagnostic approach (FVDA) is introduced to determine the local budgets of various quantities like energy, water mass and total mass in a predefind control volume. The monitoring of conservation properties is essential for model development and for the investigation of the hydrological cycle, as well.
Secondly, a spectral cosine fourier analysis (SCFA) makes a spectral decomposition of e.g. the wind field in a region of interest. Doing so, one can make a statement about the ability of a model to reproduce energy- and enstrophy cascades.

The application of the FVDA to the COSMO model reveals large errors in energy and total mass conservation, but a good performance for the water mass. It is shown, how physical processes and numerical schemes contaminate the local budgets. Regarding this fact, it is demonstrated how to construct a saturation adjustment technique for COSMO to reduce these errors.

The advantage of the both diagnostic tools is that any weather or climate model can be evaluated, including the limited area models using lateral relaxation and upper damping zones. Therefore, it is discussed with a realistic test case, which dimension the conservation errors can reach. In addition, the SCFA indicate to less numerical diffusion with upwind biased difference approximations.

Back