P57 Polar WRF simulations over northwest Canada

Klausmann, Alfred M., Exponent

Polar WRF simulations were conducted over the Victoria Island region in Northern Canada. The simulations were motivated by the need to develop three-dimensional meteorological data for air quality studies. The Polar optimized WRF Version 3.4.1 provided by the Polar Meteorology group at Ohio State University was used for the simulations. The Polar WRF model has several improvements for high-latitude applications such as better surface energy balance and surface fluxes as well as being able to handle variable snow and ice thicknesses, and the addition of seasonally varying sea ice albedo. These modifications are done through enhancements of the NOAH land surface model.

The Polar WRF model was run with the three one-way nested domains at 30, 10, and 3 km resolution using a polar stereographic projection centered at 67.25o north and 110.10o west. The 3 km domain was centered over the Coronation Gulf and the Dease Straight. The model was configured with 45 vertical levels. The initial and lateral boundary conditions were from the three-hourly North American Regional Reanalysis data with a spatial resolution of 32 km. An analysis of skin temperature from the NCEP Real Time Gridded Sea Surface Temperature data set showed that this data included unrealistic well above freezing skin temperature during the winter over the narrow waterways south of Victoria Island. The NARR data showed a much better representation of land-water temperatures in this region and was thus used to define the sea surface temperature. The model was configured with the Morrison double moment scheme to handle microphysics, Mellor-Yamada-Janjic planetary boundary layer and ETA surface layer schemes. Subgrid scale convection was parameterized using the Grell-Devenyi 3D ensemble cumulus parameterization scheme on the 30 and 10 km domains while the 3 km nest simulated convection explicitly. Short and longwave radiation processes were handled using the Goddard scheme and Rapid Radiative Transfer Model (RRTM) scheme, respectively. Four dimensional data assimilation (FDDA) was used to force the model integration towards the three-hourly fields from the NARR data on the coarse domain only. The FDDA simulations incorporated three-dimensional analysis nudging while discrete observations were used for validation.

Simulations were conducted using sea ice data sets from the National Snow and Ice Data Center and additional simulations were conducted using the sea surface temperature threshold option for defining sea ice. An evaluation of the Polar WRF model performance for these tests will be presented including comparisons of Polar WRF prediction surface winds and temperatures with available observational data.