LAUSR

Land surface heterogeneity exerts a strong control on atmospheric boundary‐layer (ABL) evolution by spatially varying the distribution and partitioning of surface energy fluxes and triggering secondary circulations. The representation of this physical process in numerical weather prediction (NWP) models is especially affected in the terra incognita as the model grid resolution approaches the length‐scale of the largest eddies in the boundary layer. We explore these effects for a mesoscale strip‐like land surface inhomogeneity in land cover, soil moisture or a superposition of both embedded in an elsewhere homogeneous landscape. The study is conducted with the numerical weather prediction model ICON (ICOsahedral Nonhydrostatic), using the default operational level 2.5 Mellor–Yamada turbulence closure (MY) and a large‐eddy simulation (LES) configuration as a benchmark. While simulations with the default ABL scheme approach the LES reference when refining the spatial grid towards finer resolution, the model generates artificial circulations leading to ABL height oscillations when the horizontal grid resolution (∆x) approaches the ABL height zi . The effect of these model‐induced circulations on the state of the boundary layer is even present with weak thermal heterogeneity (∆H) under low background wind speed ( vx) but diminishes with increasing background wind speed. The tuning of the asymptotic turbulent mixing length‐scale l∞ in the operational ABL scheme helps in reducing the amplitude of the oscillations, thereby reducing the artificially induced circulations due to thermal heterogeneity which might act as unintentional trigger for clouds and precipitation. Based on the tuned synthetic model data from sensitivity runs, we propose a new parametrization for a 2‐D l∞ as a function of ∆H , zi/∆x and vx , which is otherwise held as a constant in the ABL scheme.