Mesoscale modelling of Low Level Jets in the North Sea

Low Level Jets (LLJ) produce extraordinary shear distributions across the rotor swept area of wind turbines, dramatically affecting the power production and loads. LLJs arise when strongly stably stratified conditions occur and turbulence is almost suppressed. This is a challenging scenario for numerical models, which were formulated in the majority of the cases for turbulent atmospheric flows. Hence, the ability of mesoscale models to forecast those events needs to be thoroughly analyzed.

In this study, five different turbulent flux parameterizations in the Weather Research and Forecasting model are confronted and compared with LiDAR wind profiling measurements and multi-height sonic anemometry at the FINO1 platform, located 40 km off the German shore in the North Sea. Two cases were the amplitude of the jet core and location significantly differ are selected. The performance is evaluated considering turbulent flux profiles, eddy diffusivities and wind shear. We consider one first order closure scheme: Yonsey University (YSU), and four one-and-a-half order (or TKE closure schemes): Mellor-Yamada-Janic (MYJ), Mellor-Yamada-Nakanishi-Niino (MYNN), Quasi-normal Scale Elimination (QNSE) and Bougeault-Lacarrère (BouLac).

In general, we find the PBL schemes to provide high diffusivity and thus, to have difficulties to generate strong shear conditions as the observed ones. TKE closure schemes can better reproduce the LLJ features but still present some problems that seem to come from enhanced mixing close to the surface partially added to alleviate problems such that run away cooling. QNSE and MYNN are the most suited PBL schemes to simulate LLJs based on the offshore conditions at FINO1. We found BouLac parameterization to be unable to handle such strong stability due to a deficiency on the stability correction term of the eddy diffusivity. Fine vertical resolution improves the location of the jet core but has a less evident influence on its magnitude.