LES of Mountain Rotors and Stable Boundary Layers During T-REX: Effects of Sub-grid Land-Surface Heterogeneity

Accurate prediction of both quiescent conditions and strong wind events over mountainous terrain is a challenge because of land-surface heterogeneity, among other factors. The resolution of land-surface data available from satellites, ranging from several hundreds of meters to a kilometer, has far outpaced computational limits for operational and climate forecasting, where grid sizes typically range from 4-12 km for local forecasts to 100 km for global models. Atmospheric numerical codes typically assign the dominant soil and vegetation types in a given grid cell to the entire cell, creating significant errors in flux calculations in complex terrain where elevation, soil and vegetation type, and snow cover can vary substantially over small scales.

To allow full utilization of available land-surface data, tiled land-surface models have been developed to compute separate surface energy budgets and fluxes from multiple land surface types within a grid cell. While previous tiled models used a relatively limited number of soil/vegetation combinations, we implemented a tiled land-surface model that allows for the occurrence of all physically possible sub-grid combinations of soil and vegetation types. Our implementation in the Advanced Regional Prediction System (ARPS) also provides for fractional snow cover. At each time step, soil variables (temperature, moisture, surface humidity), surface radiative flux and net radiation, and surface fluxes of heat, momentum, and moisture are calculated with a column model for each sub-grid class, with all classes within a grid cell drawing on the same atmospheric variables. A sum of the surface fluxes for each grid cell, weighted by the fraction of each type, is then calculated as input to the model dynamic core.

Obervational data collected during the Terrain-Induced Rotor Experiment (T–REX) in the Owens Valley, California during March and April 2006 have been used to examine the degree to which land-surface heterogeneity effects are important under a variety of atmospheric conditions. We study the influence of surface flux heterogeneity on simulations of both stable boundary layers and convective flows in mountainous terrain, where fluxes are known to impact the timing of slope winds and along-valley winds, and may affect transitions between non-wave and wave conditions and the boundary layer separation necessary for rotor formation. The model sensitivity to the tiled land-surface representation is evaluated by applying both tiled and standard ARPS to simulations of stable boundary layer and strong rotor cases during T-REX. Predictions from the NAM model are used to provide initial and boundary conditions at 12 km resolution, and one-way nesting is used to reach a horizontal resolution of 350m in the Owens Valley. Both simulations utilize high-resolution land-surface data-sets, including 1km daily snow analysis, 250m greenness fraction (NDVI), 1km soil type, and 30m vegetation type. In the standard ARPS simulation the dominant soil and vegetation type and average greenness fraction and snow-depth values are used, while in the tiled model fractional snow cover, and average snow depth and greenness fraction are calculated for each soil-vegetation type.

Simulation results are compared with surface meteorological tower time series data and to rawinsonde data. The impact of the tiled model is strongest near the ground under stable conditions, but effects, albeit often small, are seen throughout the atmospheric boundary layer in both stable boundary layer and rotor cases.