Review of the perturbation methods in the Meteorological Service of Canada (MSC) Global Ensemble Prediction System (GEPS)

MSC is running a 20 member GEPS (0.45x0.45L40) up to Day 16 operationally (32 days on Thursdays at 00Z). This system is using an Ensemble Kalman Filter as a data assimilation scheme (Houtekamer et al. 2014) to generate the initial conditions for the medium-range ensemble forecasts. The forecast perturbations are coming from set a multi-physical parameterizations as well as two stochastic approaches (physical tendencies perturbations and the stochastic energy back-scattering scheme).

• Context: In preparation of the arrival a new supercomputer and the associated opportunity to increase model resolution, an evaluation of the different perturbations methods of the MSC Global Ensemble Prediction System is performed.
• Goal of the study: review the source of spread in the system.
• Method: several experiments with individual source removed incrementally