The REPS is based on the limited-area version of the Global Environmental Multiscale model (GEM-LAM) with 20 members at 33-km resolution. Initial conditions are provided by the Ensemble Kalman filter (EnKF), and boundary conditions are obtained from the global EPS, both operational at the Canadian Meteorological Centre (CMC), and stochastic perturbations are applied to the sub-grid-scale physical tendencies and to surface parameters. The high-resolution models include 2.5-km and 1-km versions of GEM-LAM integrated for 16h, twice a day, with improved cloud microphysics, geophysical fields, and radiation and cloud-radiation interactions, as compared with the system (15-km and 2.5-km models) currently operational at CMC. Particular emphasis is put on mesoscale forecasts of important weather elements over complex terrain in winter conditions, such as cloud, fog, visibility, mixed precipitation types, and wind gusts.
Finally, several new and original tools are used to adapt and refine forecasts near and at the surface. Microscale 2D surface systems (with 100-m and 20-m grid sizes) cover the Vancouver Olympic venues using forcings from the 1-km model. Based on a similar strategy, a single-point model will be implemented, using surface observations as forcing, to better predict surface characteristics at each station of the special observing network set up for the Vancouver Olympics. The microscale 2D surface models better represent surface processes, and thus lead to better predictions of snow conditions and near-surface air temperatures.
The configuration of the experimental numerical prediction system will be presented at the conference, together with preliminary results from the practicum forecaster training period of January-March 2008.