From the GABLS's result to the ARPEGE NWP system: an easy ride?

The first GABLS experiment (Holtslag, 2003) provided a clear framework for 1D and LES intercomparison on a stable boundary layer (SBL).

Several operational and research models participated to the exercice. The main result of this intercomparison (Cuxart et al, 2006) is that the operational models mix more than the research models to a lesser extent for the operational turbulent kinetic energy (TKE) schemes.

The global model ARPEGE uses a first-order approach with the stability function derived from Louis (1979) and an interactive mixing length. This version too is not able to reproduce correctly the Ekman spiral. The low level jet does not exist due to the excess mixing on the wind and on the temperature in the SBL.

Accordingly, a TKE scheme (Cuxart et al, 2000) has been implemented in the ARPEGE system for a cross validation between ARPEGE-CLIMAT, ARPEGE-NWP and the limited area model ALADIN. This implementation requires new choice with respect to the different constraints for the tunable parameters of the TKE scheme but not only.

The interaction with the microphysics, the shallow convection and the deep convection raises new questions and probably changes the missing processes and the compensating errors.