Statistical Postprocessing of Operational NWP Output—A Canadian Retrospective and Perspective (Invited Presentation)

Operational statistical postprocessing schemes have been running at the Canadian Meteorological Center (CMC) since the 1970s. As in the US, the earliest techniques were of the “Perfect Prog” formulation, but the landmark development of MOS in the US convinced us to steer development in that direction, leading to our first operational implementation of MOS in 1986 (wind). When changes to the operational models became more frequent, we were forced to switch back to perfect prog methods for the decade of the 1990s, until we solved that problem with an “updateable MOS” system, implemented in 2000. This system has been stable and running for nearly two decades now. Current development efforts have focused on switching from station-based MOS prediction to grid-based prediction.

This presentation will describe the highlights of the Canadian experience with statistical postprocessing in the operational context from the 1970s to the present, including examples and applications to ensembles. The presentation will conclude with a look toward the future.