5D.6 The Sensitivity of Hurricane Irma’s Track Forecasts As Determined from ECMWF Ensemble Forecasts

Tuesday, 17 April 2018: 9:15 AM
Heritage Ballroom (Sawgrass Marriott)
Ryan D. Torn, SUNY, Albany, NY

Although Hurricane Irma’s track forecast errors were smaller than historical averages, there were critical track uncertainties 3-5 days prior to Irma’s Florida landfall that had a significant impact on intensity, wind, and surge forecasts. As a consequence, it was of great interest to understand how the uncertainty associated with various tropical and midlatitude features impacted the steering flow and hence Irma’s track. Regions of large sensitivity indicate where to take additional observations, either via reconnaissance aircraft (i.e., G-IV) or via 6-h sampling by rawinsondes (if the region is over CONUS). This study investigates the sensitivity of Hurricane Irma’s position forecasts at the time of the South Florida landfall to the steering flow for forecasts initialized at various points in Irma’s life. The sensitivity is determined using the ensemble-based sensitivity method applied to the 51-member ECMWF ensemble forecasts. Preliminary results indicate that early track forecasts of Irma were sensitive to a midlatitude trough to the east of Irma, which was associated with the southwestward motion from 2-5 September. Once Irma moved well west of this feature, the most sensitive region shifted to the Bahamas, associated with the southward extent of the midlatitude trough over the Southeast United States. Within 3 d of landfall, the main sensitive region moved into the Gulf of Mexico and may have been related to how Hurricane Katia’s outflow interacted with the aforementioned trough in the Southeast United States.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner