120 Four Years of Real-time Probabilistic TC Prediction with the COAMPS-TC Ensemble

Tuesday, 17 April 2018
Champions DEFGH (Sawgrass Marriott)
Jonathan R. Moskaitis, NRL, Monterey, CA; and P. A. Reinecke, W. A. Komaromi, and J. D. Doyle

Since 2014, NRL has run COAMPS-TC ensemble forecasts in real-time for select Atlantic, Eastern North Pacific and Western North Pacific TCs, altogether constituting a sample of nearly 2000 forecast cases. Over the four years of forecasts the ensemble configuration has remained constant, with an unperturbed control member accompanied by 10 perturbed members that have initial condition variability introduced at both the synoptic and vortex scales, as well as lateral boundary condition variability.

Utilizing the aforementioned forecast sample, we assess the quality of ensemble-based probabilistic forecasts of rare events, such as episodes of rapid intensification and rapid over-water weakening. For TC position, we validate how well the ensemble-based TC forecast position covariance matrix characterizes the leading direction of position variability and its magnitude. We also assess the ability of the ensemble to discriminate between forecast scenarios with differing amounts of position and intensity uncertainty, and investigate the covariance between position and intensity predictions. Finally, we present the usual first-moment statistics and draw conclusions regarding the future prospects of regional dynamical model ensemble prediction of tropical cyclones.

- Indicates paper has been withdrawn from meeting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