6B.1 Short-range ensemble forecasts of high-impact weather from MOGREPS

Wednesday, 27 June 2007: 12:00 AM
Summit B (The Yarrow Resort Hotel and Conference Center)
Kenneth R. Mylne, Met Office, Exeter, Devon, United Kingdom; and N. Bowler, S. John, K. Robertson, T. P. Legg, and A. Arribas

The UK Met Office introduced a new quasi-operational short-range ensemble prediction system, MOGREPS, in September 2005, and following successful trials is now planning to make MOGREPS fully operational during 2007. MOGREPS uses a regional model covering the North Atlantic and Europe at 24km resolution, with lateral boundary conditions supplied from a global ensemble with a grid-length of ~90km in mid-latitudes. MOGREPS uses an ETKF (Ensemble Transform Kalman Filter) to provide initial perturbations suitable for short-range use. It is believed to be the first quasi-operational implementation of the ETKF in a full NWP system making use of the full set of observations used in the 4d-VAR system which generates the control analysis. MOGREPS also employs three different stochastic physics schemes to help account for model error. For operational use MOGREPS is supported by a comprehensive web-based display system which allows forecasters real-time access to a wide range of probabilistic forecast products animated in time. Site-specific products are generated through an operational database system allowing statistical post-processing for generation of optimised products for customers.

A key aim of MOGREPS is to aid forecasters in prediction of high impact weather events. This paper will present the MOGREPS system and focus on performance for high-impact events. Objective verification of high-impact events is difficult due to the rarity of such events, but results for some moderately severe events will be presented as a guide to the capability of the system. Early results indicate that the ensemble provides skilful probability forecasts for a range of parameters, and in particular has greater skill in the short-range than is available from existing medium-range ensemble prediction systems. This is believed to be due to the use of a higher resolution model and perturbation systems designed for short-range performance. Feedback from operational forecasters has also been extremely positive. Case studies for one or two more significant events will be shown to illustrate the benefits of an ensemble system in highlighting the risks which may be missed by deterministic forecast systems.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner