Many mesoscale models today use some kind second order closure modeling of turbulence. Usually, the closure model is derived from theory and closure constants are derived from a combination of labo-ra-to-ry and field measurements, both from horizontally homogeneous and stationary conditions. The results in the mesoscale model are usually evaluated in terms of the effects on the resolved scale variables and the effects on the ensemble averaged turbulence by the lack of homogeneity and stationarity is usually never evaluated in any detail - everyone is happy as long as it works.

In this presentation, we take a closer look on the turbulence statistics generated from a second order closure model in a mesoscale model when simulating a highly heterogeneous coastal flow. Highlights of the presentation will be a discussion on what one can really expect to learn from looking at modeled turbulence, an illustration of how observations and model results can be used to form a consensus on the turbulence structure for a special type of flow, and a discussion on how prognostics turbulent kinetic energy is misused in models by forcing it at the boundary by similarity relationships.