Abstract : 1A.5
Very high resolution windfields simulations over a steep alpine ridge: mean flow and turbulence characteristics

Francoise Faure, Michael Lehning, Christophe Hug, Stephen Mobbs, Lewis Huw
Swiss Federal Institute for Snow and Avalanche Research, SLF

Snow transported by wind is an important factor for avalanche danger as well as ecology and hydrology in mountainous environments. The topographic modification of airflow causes snow erosion and accumulation zones. Knowledge on flow and turbulence in steep mountains is also useful for the assessment of potential damage to forests and infrastructure, exploitation of wind energy and erosion of soils. This phenomenon is highly dependent on the airflow mean and turbulence characteristics, in particular Turbulent Kinetic Energy (TKE), which is also an important parameter in the SLF Snowdrift model Alpine3D. In this study, high resolution windfields over the steep Gaudergrat ridge, located in the region of Davos, Switzerland, are computed using the Advanced Regional Prediction System (ARPS) in its Large Eddy Simulations (LES) configuration. The advantage of LES in simulating turbulent flow is that the turbulence is explicitly resolved up to the filter size. Previous microscale windfield simulations over the Gaudergrat ridge, applying a homogeneous initial state constructed from a single sounding showed that Initial and boundary conditions have a strong influence on the flow stability. Consequently, ARPS is now coupled to the Meteoswiss Model aLMo : aLMo outputs are interpolated on the ARPS croarsest domain, which resolution is 1600m. This first step provides initial and boundary conditions forcing, then ARPS is run using four nested domains to bridge spatial resolutions from 7km (aLMo-grid size) to 25 m, the finest resolution. The four nested grids have been constructed using the SwissTopo Digital Elevation Model. In comparison with the single sounding initialisation and periodic boundary conditions as used before, the nesting method from aLMo permits to obtain a much more stable flow, hence one full day can be simulated. This is of particular importance to reproduce snowdrift events of several days. A study of the mean flow and the turbulent kinetic energy (TKE) is presented for 2 days of the Gaudergrat Experiment 2003 (Gaudex 2003). These days, although summerly, are comparable with commonly observed wind conditions during snowdrift events at the Gaudergart location. The Gaudex took place during the summer 2003 at the Gaudergrat ridge in collaboration with the University of Leeds. During this measurement campaign, 30 measurement sites were operating, including automatic weather stations, sonics, sodars, and a radiosounding once a day. Numerical results show a good agreement with measurements for the mean flow characterics, whereas turbulence characteristics were more difficult to reproduce. Thermally induced changes in wind directions have been observed but are less pronounced in the model. Thermal effects will be much less important during the winter. The turbulent kinetic energy obtained from the model is much less than the one measured, a common feature for current mesoscale atmospheric models, which are over-dissipative in the sub-grid parameterizations. Also the spatial patterns of TKE are more pronounced in the measurements than in the model. TKE is higher in the lee slopes due to flow separation. This behaviour migth indicate that common parameterizations of TKE will fail in complex terrain. Future work will investigate, in how far the results from the drifting snow simulations are influenced by these model deficiencies.