Probabilistic Tropical Cyclone Prediction with a High-resolution Regional Model: The COAMPS-TC ensemble

Tuesday, 19 April 2016
Plaza Grand Ballroom (The Condado Hilton Plaza)
Jonathan R. Moskaitis, Naval Research Laboratory, Monterey, CA; and P. A. Reinecke and J. Doyle

In 2014 and 2015 NRL-Monterey ran a real-time 11-member COAMPS-TC ensemble for all Atlantic tropical cyclones in Aug-Oct as well as select Eastern North Pacific and Western North Pacific TCs. The configuration of the model used for the ensemble was the same as operational deterministic COAMPS-TC, except that the nested-grid horizontal resolutions were increased to 27/9/3 km from 45/15/5 km and GFS was used as the parent global model rather than NAVGEM. Ensemble perturbations to the initial state were applied at the synoptic scale and also specifically to the initial TC vortex position, intensity, and radial structure. Here, we present probabilistic and deterministic validation results based on the 2-year sample of real-time COAMPS-TC ensemble predictions. Deterministic validation results highlight the relative performance of the ensemble mean forecast and unperturbed control member forecast. However, our primary focus will be on probabilistic forecast validation metrics for TC position and intensity, with special attention given to assessment of ensemble-based probabilistic predictions of rapid intensification. Finally, we examine the potential of a multi-model high-resolution regional TC ensemble prediction system by combining the real-time Atlantic basin COAMPS-TC ensemble forecasts with the corresponding real-time ensemble forecasts from HWRF and GFDL.
- Indicates paper has been withdrawn from meeting
- Submission entered in competition