Canadian Global Ensemble Forecast System

Canadian Global Ensemble Forecast System
Name	Canadian Global Ensemble Forecast System
Developer	Environment and Climate Change Canada
Initial release	2006
Latest release	2020s
Programming language	Fortran, C
Platform	High-performance computing

Contents

Canadian Global Ensemble Forecast System The Canadian Global Ensemble Forecast System is an operational probabilistic forecasting suite produced by Environment and Climate Change Canada for global numerical weather prediction. It provides ensemble-based predictions used by national services, regional centres, and private firms to support planning for aviation, marine, emergency management, and energy sectors. The system integrates advancements from international collaborations with agencies such as European Centre for Medium-Range Weather Forecasts, United States National Weather Service, Met Office, and research institutions including University of Toronto, McGill University, and University of British Columbia.

Overview

The system produces multiple forecasts to represent uncertainty, delivering analyses and forecasts for variables like geopotential height, temperature, wind, and precipitation to stakeholders including Nav Canada, Transport Canada, Canadian Armed Forces, Hydro-Québec, and insurers. It runs on supercomputers maintained by Compute Canada and uses numerical schemes and physical parameterizations developed in partnership with groups such as Meteorological Service of Canada, Canadian Meteorological and Oceanographic Society, and international research labs like NOAA Geophysical Fluid Dynamics Laboratory and National Center for Atmospheric Research. Products are disseminated via portals maintained by Environment and Climate Change Canada, academic consortia, and commercial vendors partnering with IBM and Amazon Web Services for archival and delivery.

Development traces to ensemble concepts advanced at European Centre for Medium-Range Weather Forecasts and operational ensemble adoption by United States National Weather Service in the 1990s. Early Canadian ensemble experiments involved collaborations with Météo-France, Deutscher Wetterdienst, and university groups at McGill University and Université de Montréal. Major milestones include implementation of stochastic physics inspired by research at Met Office and transition to an updated global model core following community work at Canadian Centre for Climate Modelling and Analysis and the Canadian Institute for Theoretical Astrophysics collaborations. Funding and oversight have included agencies such as Canada Foundation for Innovation, Natural Sciences and Engineering Research Council of Canada, and provincial research initiatives in Ontario and Québec.

The system is built on a global circulation model using spectral and grid-based discretizations developed in coordination with the Meteorological Service of Canada research division and academic partners including McGill University and University of Toronto. Physics packages incorporate microphysics schemes informed by studies at National Center for Atmospheric Research, convective parameterizations related to work at Princeton University, and planetary boundary layer formulations similar to those tested at University of Reading. Dynamical cores draw on research from European Centre for Medium-Range Weather Forecasts and Deutscher Wetterdienst, while coupling to ocean and sea-ice components references models used at Bedford Institute of Oceanography and Fisheries and Oceans Canada. Numerical treatments for advection, diffusion, and time-stepping reflect methods pioneered at Institut Pierre-Simon Laplace and Los Alamos National Laboratory.

Data assimilation merges observations from satellites such as Geostationary Operational Environmental Satellite and Polar-orbiting Operational Environmental Satellite, radiosonde networks including those coordinated by World Meteorological Organization, aircraft reports from NAV CANADA flights, and surface observations curated by Meteorological Service of Canada stations. The system employs hybrid variational and ensemble-based assimilation methods influenced by research from European Centre for Medium-Range Weather Forecasts, NOAA, and academic groups at University of Oxford and Massachusetts Institute of Technology. Reanalysis datasets and boundary conditions reference outputs from ERA-Interim, ERA5, and collaborations with international centres such as Met Office and JMA.

Ensemble generation uses perturbations from initial condition uncertainty, stochastic physics perturbations inspired by Met Office research, and ensemble spread control with methods similar to those employed at European Centre for Medium-Range Weather Forecasts. Typical configurations include multiple members run at differing resolutions and perturbation strategies developed with inputs from Cornell University, University of Washington, and University of Colorado Boulder. The ensemble supports coupled atmosphere–ocean–ice runs drawing on expertise from Bedford Institute of Oceanography and Fisheries and Oceans Canada, and interacts with downstream regional ensembles used by provincial centres in Alberta, British Columbia, and Ontario.

Verification metrics follow best practices from World Meteorological Organization task teams and use continuous rank probability score, Brier score, and anomaly correlation as in studies from European Centre for Medium-Range Weather Forecasts and NOAA. Independent verification campaigns have compared performance with ensembles from Met Office, ECMWF, and United States National Weather Service, with peer-reviewed evaluations produced in journals such as Monthly Weather Review, Journal of Climate, and Quarterly Journal of the Royal Meteorological Society. Performance assessments also incorporate case studies for extreme events examined by research groups at University of Toronto, McGill University, and University of British Columbia.

Operational users include Meteorological Service of Canada forecasters, Nav Canada for aviation planning, Transport Canada for marine advisories, and emergency management agencies in provinces such as Ontario, Québec, and British Columbia. Applications extend to energy demand forecasting for companies like Hydro-Québec and renewable integration studies performed with partners at National Research Council Canada. The ensemble also supports climate services, hydrological modelling with groups such as Environment and Climate Change Canada's hydrology divisions and academic collaborators at University of Waterloo and McMaster University.

Limitations include sensitivity to observational coverage over polar regions studied by Canadian Ice Service and representation of convective processes examined at Princeton University and Colorado State University. Future developments target higher resolution runs, improved coupling with ocean and ice models informed by Bedford Institute of Oceanography, increased use of machine learning methods from Vector Institute and MILA – Quebec AI Institute, and expanded collaboration with international centres including European Centre for Medium-Range Weather Forecasts, Met Office, and NOAA. Planned investments involve supercomputing upgrades through Compute Canada and algorithmic advances researched at University of Toronto, McGill University, and University of British Columbia.

Category:Meteorological models