The KNMI Parameterization Testbed: Motivation, Configuration and Preliminary results

Single column model (SCM) simulations have become a valuable and relied-upon tool in the development and evaluation of the sub-grid physics of general circulation models (GCM). In this method, the complete sub-grid model of a GCM is independently integrated in time, using prescribed large-scale forcings and boundary conditions. Numerous SCM inter-comparison studies have been organized in recent years, for example by the various GEWEX Cloud System Studies (GCSS) working groups. Typically, an idealized case is constructed based observational data from a field experiment, after which simulations are made with both SCM and high-resolution cloud resolving models. The focus of these efforts has been on subgrid model components such as turbulence schemes, convection schemes and cloud schemes, inspired by the significant uncertainty in numerical weather and climate predictions still associated with these model components. The outcome of these projects has resulted in demonstrable improvements in GCM performance.

A clear advantage of the SCM technique is that the model physics is studied in an isolated mode from the larger-scale circulation, which enhances model transparency and facilitates the identification of both problems and progress. On the other hand, this means that any interaction between parameterized scales and the larger scales is absent. Another potential shortcoming of the SCM method is that the available case studies often represent so-called 'golden days', reflecting observed situations in nature that are thought to be prototypical for a weather regime of interest. However, one can question the representativeness of such scenarios; they might reflect unique situations that rarely occur. As a result, subgrid models might be optimized for rare situations, while the performance for the many more intermediate situations remains unassessed.

These issues have inspired efforts to evaluate sub-grid parameterizations on a more continuous basis, thus automatically covering many different weather regimes. One such project is the recently initiated KNMI Parameterization Testbed. Various SCMs are evaluated daily against continuous observational datastreams from the Cabauw meteorological site in the Netherlands, for a range of relevant meteorological parameters. The larger-scale forcings required to drive the SCMs are obtained from short-range forecasts with the Regional Atmospheric Climate Model (RACMO), initialized with the ECMWF analysis. During the SCM simulations, the model thermodynamic and kinematic state variables are continuously relaxed towards the "background" (i.e. RACMO) state, on a 6-hour timescale. This relaxation timescale is small enough to curb excessive model drift, but large enough to leave the SCM sufficient freedom to generate its own unique state. The short range of these forecasts (3 days) justifies direct comparison to local measurements, as the model initial state (i.e. the ECMWF analysis) is still close.The emphasis of the evaluation lies on the atmospheric boundary layer.

Preliminary testbed results will be presented, including various branches of the ECMWF cycle 31r1 subgrid model. Model performance for various encountered scenarios will be discussed, among which the GABLS3 stable boundary layer case. Weather regimes for which models typically have problems will be identified. This could guide further future research by the meteorological modelling community.