The GEM's operational model, known as the global deterministic prediction system, is currently operational for the global data assimilation cycle and medium-range forecasting, the regional data assimilation spin-up cycle and short-range forecasting. Mesoscale forecasts are produced overnight and are available to the operational forecasters. A growing number of meteorological applications are now either based on or use the GEM model. Output from the GEM goes out to 10 days, on par with the public output of the European Integrated Forecast System.
Ensemble model
The ensemble variant of the GEM is known as the Global Ensemble Prediction System. It has 20 members and runs out 16 days, the same range as the American global forecast system. The GEPS runs alongside the GFS ensemble to form the North American Ensemble Forecast System. A regional ensemble prediction system, covering North America and also having 20 members plus control, runs out 72 hours.
Development
The GEM model has been developed to meet the operational weather forecasting needs of Canada for the coming years. These presently include short-range regional forecasting, medium-range global forecasting, and data assimilation. In the future they will include nowcasting at the meso-scales, and dynamic extended-range forecasting on monthly to seasonal timescales. The essence of the approach is to develop a single highly efficient model that can be reconfigured at run time to either run globally at uniform-resolution, or to run with variable resolution over a global domain such that high resolution is focused over an area of interest.
Mechanics
The operational GEM model dynamics is formulated in terms of the hydrostatic primitive equations with a terrain following pressure vertical coordinate. The time discretization is an implicit two-time-level semi-Lagrangian scheme. The spatial discretization is a Galerkin grid-point formulation on an Arakawa C-grid in the horizontal and an unstaggered vertical discretization. The horizontal mesh can be of uniform or variable resolution, and furthermore can be arbitrarily rotated, the vertical mesh is also variable. The explicit horizontal diffusion is -2 on all prognostic variables. The operational GEM model is interfaced with a full complement of physical parametrizations, these currently include:
The next stage of development of the GEM model is to evaluate the non-hydrostatic version for mesoscale applications where the hydrostatic assumption breaks down. The limited-area version is scheduled to follow. The distributed memory version of GEM is almost completed, it is a major recoding effort that is based upon a locally developed communication interface currently using Message Passing Interface. Research on the performance of different land surface schemes such as ISBA and CLASS is making progress. The strategy is progressing towards a unified data assimilation and forecast system, at the heart of which lies a single multipurpose and multiscale numerical model.
