High-Resolution COAMPS-TC Ensemble Performance for the High-Impact Hurricanes of 2017

Hurricanes Joaquin (2015), Matthew (2016), Harvey (2017), and Irma (2017) were all high-impact, major (category 3+) tropical cyclones (TCs) of the last few Atlantic hurricane seasons. Each of these storms presented unique track and intensity forecast challenges. The performance of the 11-member COAMPS-TC ensemble in terms of accurately depicting the forecast uncertainty, as well as its ability to correctly capture the verifying track and intensity amongst the envelope of possibilities is evaluated. For each system, a handful of ensemble members correctly predicted landfall intensity and location from 96-120 h lead time. The relative sensitivity of the forecast to vortex initial condition (IC), environmental IC, and environmental boundary condition (BC) perturbations to wind and potential temperature is evaluated by running the ensemble with each perturbation technique activated individually, then compared against the operational forecast run with all three perturbation techniques active simultaneously.

The effectiveness of each of the three perturbation techniques in generating spread in both track and intensity is found to vary as a function of forecast lead time, and also varies from storm to storm. Overall the spread-skill relationship of the ensemble is found to be quite good for track, but underdispersive for intensity, especially at longer lead times. Lastly, the sensitivity of the forecast to number of ensemble members is also evaluated. Preliminary results suggest that increasing the size of the ensemble produces a more realistic probability distribution for track, intensity and wind speed probabilities, but does not significantly affect the spread-skill score.