JP1.12 Development and testing of an operational mesoscale ensemble system

Monday, 1 June 2009
Grand Ballroom Center (DoubleTree Hotel & EMC - Downtown, Omaha)
Teddy Holt, NRL, Monterey, California; and J. Hansen, C. Bishop, and M. Sestak

To provide effective mesoscale ensemble prediction of high-impact weather relevant to the Navy, the Naval Research Laboratory (NRL) and Fleet Numerical Meteorology and Oceanography Center (FNMOC) have jointly developed and beta-tested with operational inputs a new mesoscale ensemble prediction system. This system, utilizing the Coupled Ocean Atmosphere Mesoscale Prediction System* (COAMPSĀ®), allows us to (1) explore the ability of COAMPS to simulate high-impact weather from a probabilistic approach, 2) explore mesoscale predictability of Navy-relevant high-impact phenomena, 3) explore initial condition perturbation methods for mesoscale ensemble prediction, and 4) investigate representation of model error using mesoscale stochastic and perturbed physics. The development of this mesoscale ensemble system has been closely coordinated with the Navy/Air Force Joint Ensemble Forecast System (JEFS), a joint program that will provide probabilistic mesoscale forecasts for the Navy and Air Force.

The system uses a computationally inexpensive Ensemble Transform (ET) method for generating high resolution initial perturbations for regional ensemble forecasts. The method provides initial perturbations that (a) have an initial variance consistent with the best available estimates of initial condition error variance, (b) are dynamically conditioned by a process similar to that used in the breeding technique, (c) add to zero at the initial time, (d) are quasi-orthogonal and equally likely, and (e) partially respect mesoscale balance constraints by ensuring that each initial perturbation is a linear sum of forecast perturbations from the preceding forecast. Results indicate that the ET technique dynamically selects growing perturbations that are consistent with user provided estimates of analysis error variance. In addition, ET ensembles show skillful means and useful spread-skill relationships. Results will be shown for ensemble forecasts over an extensive period (1-6 months) for two areas: (1) the JME domain (45- and 15-km resolution) centered over Korea, and (2) the RIMPAC domain (45-, 15-, and 5-km) centered over Hawaii.

---------------

*COAMPSĀ® is a registered trademark of the Naval Research Laboratory.

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