Simulations of the planetary boundary layer (PBL) over the wintertime, 2-m thick, Arctic pack ice have been done in one-dimension offline tests and in three dimensions with the Penn State/NCAR Mesoscale Model (MM5). The long-range purpose of these simulations is to scale-up the point measurements of surface fluxes obtained during the Surface Heat Flux of the Arctic Ocean (SHEBA) year to global circulation model-scale grids using the nesting capability of the MM5. In the short range, this requires obtaining excellent simulations of the boundary layer thermodynamic and kinematic structure as revealed by validating the model output with the extensive observations available near the SHEBA site. The MM5 model configuration has been optimized for the Arctic environment with choices of domain boundaries, low-level vertical resolution, and model physics, with development of a more sophisticated surface parameterization shown to be key. The initial testing of the model is done in the most simple environmental conditions with no solar radiation and mostly clear skies by simulating the week of Jan. 14-20, 1998, at which time the SHEBA site was at 75°N and 151°W in the Beaufort Sea. The validation of the model PBL structure and forcing is done with hourly data from the NOAA/ETL sodar and the 5-level Atmospheric Surface Flux Group 20-m tower site, and with12-hourly rawinsondes, a cloud radar, and a lidar. At the lowest heights, the tower data is used in preference over the radiosonde and sodar data. At heights where both sounding and sodar wind data exists, the sodar data is assumed to be correct, and tower humidity data is used to correct the sounding humidity profile. Along with basic meteorological parameters, the validation data include direct covariance measurements of the turbulent fluxes of heat, moisture, and momentum, and the measurements of the near-surface broadband shortwave and longwave, incoming and outgoing, radiative fluxes.

The results of the initial tests show: 1) the RRTM radiative scheme performs best, but could be improved with information on the aerosol concentration profiles because the moisture profile is less dominant in this dry environment compared to lower latitude environments, 2) atmospheric surface-layer schemes could be improved by using a multi-layer approach rather schemes based on the Monin-Obhukov Similarity Theory, 3) the surface parameterization needs to have a multi-level ice and a multi-level snow model in order to insulate the Arctic boundary layer from the warm ocean and provide the proper timescales to the PBL changes in response to changes in the forcing, and 4) a sensitivity to existing PBL schemes is observed but it is much less than the sensitivity to the surface model sophistication. The simulations and validations upon which these conclusions are based will be discussed. In addition, simulations using realistic spatially varying ice conditions will also be discussed if available by the time of the conference.