Turbulence Kinetic Energy Characteristics in the Inn Valley—A Model Evaluation Study with High-Quality Turbulence Measurements

Mountainous terrain covers more than 50% of the Earth's surface and is hence an important contributing factor to local weather all around the globe. Modern mesoscale numerical weather prediction (NWP) models operate with horizontal resolutions where topography is already partly resolved. Therefore, the right representation of both terrain and boundary-layer structure is of great importance for the correct simulation of atmospheric processes in complex terrain. Unfortunately, there are still shortcomings, often associated with misrepresented model terrain, sparse input data, or inappropriate parameterization schemes. To disentangle these various effects we evaluate the state-of-the-art NWP model COSMO (setup by MeteoSwiss) in the Inn Valley, Austria. A measurement data pool in the valley is provided by the so-called i-Box stations, which include two remote sensing systems in the city of Innsbruck (Doppler Lidar & HATPRO temperature profiler), and six turbulence flux-towers at representative sites in mountainous terrain, such as the valley floor, various slopes, and the mountaintop. In this contribution, we especially focus on turbulence kinetic energy (TKE) and the various terms in its budget equation (shear, buoyancy, dissipation). As an example we discuss clear-sky days, when thermally-driven flows are dominating. The TKE structure is highly variable between the various i-Box sites (order 10 km or less apart from each other), depending on the location, time of the day, slope angle, exposure to incoming radiation, and the influence of the up-valley wind. In general, the model underestimates TKE at all sites in the valley. However, good model performance is found before noon, when a growing mixed layer is present in the valley and when TKE generation is buoyancy-dominated. In the afternoon, the daytime TKE maximum occurs at the same time as the maximum wind speed of the strong up-valley wind, and is associate with shear generation. During the evening transition, the TKE in the model drops to negligibly small values, while in the measurements, moderate TKE (on the order of 0.5 m² s^-2) is present also during nighttime. Thus the model's 1D turbulence scheme exhibits problems with simulating both the daytime and nighttime TKE structure correctly – even though the modeled budget terms are astonishingly well captured over most of the time. This is related to the fact that many turbulent exchange processes in complex terrain are fully three-dimensional. This is a motivation for using 3D TKE schemes when simulating boundary-layer processes in mountainous areas.