Short-to-medium range forecasting of midlatitude weather systems using a mesoscale global model with uniform and variable resolution

Most numerical studies of mesoscale structures of midlatitude weather systems are conducted with models in which the horizontal resolution is focused over a limited region (using grid-nesting or variable resolution). With such models, the predictability of weather systems critically depends on the quality of the lower-resolution solution outside the domain of interest. Because of the always increasing computer power, it is now possible to integrate a global uniform-resolution model with a mesoscale resolution. In this study, a 0.28 degree (~ 30 km at midlatitudes) version of the Global Environmental Model (GEM) is used to simulate two severe weather events: a synoptic-scale winter storm that occurred in northeastern United States, and a mesoscale convective system that propagated over the central United States. The short-to-medium-range predictability of these events is investigated by examining forecasts up to five days before the events. This predictability is also investigated in the context of a variable-grid configuration which is currently used operationally for regional forecasting at the Canadian Meteorological Center.