Underestimation of Mechanically Generated Turbulence in a Traditional PBL Scheme over Complex Terrain: A TEAMx Case Study for the Inn Valley

Mesoscale numerical weather prediction (NWP) models typically employ planetary boundary layer (PBL) schemes to represent subgrid-scale turbulence, including 1.5-order parameterizations that explicitly predict turbulent kinetic energy (TKE). One crucial assumption of these models is that horizontal gradients, such as horizontal shear, can be neglected in the TKE tendency terms. This assumption may not hold in complex terrain, where the interaction between terrain-induced flows and the orography itself can create substantial horizontal mixing.

We investigate this limitation using a high-resolution (Δx=500 m) WRF model simulation employing the traditional 1.5-order Mellor–Yamada–Nakanishi–Niino (MYNN) PBL scheme. We focus on a valley wind case in the Inn Valley, Austria, which occurred on 29 June 2025 during the TEAMx campaign. The event was characterized by clear skies and weak synoptic forcing, favoring the development of a convective boundary layer and thermally driven daytime up-valley winds with substantial mechanical mixing.

The simulations are compared against observational data from four Doppler wind lidars and several ground measurement stations in the valley. The evolution of the wind system is represented reasonably well by the model, but the peak strength of the valley wind is underestimated. Observations from one of the lidars show that the PBL scheme appears to underestimate TKE when the turbulence is dominated by mechanical production. This bias may result from the lower wind speeds or an incomplete representation of TKE production in the PBL scheme, with potential interactions between the two factors. An estimate of the horizontal subgrid-scale diffusion suggests that accounting for the currently neglected horizontal shear production in the TKE equation could lead to an improved TKE representation.