GFS Forecast Skill for Long-Lived Tropical Disturbances
Open Access
- Author:
- Klees, Alicia Marie
- Area of Honors:
- Meteorology
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Jenni Evans, Thesis Supervisor
Paul Markowski, Thesis Honors Advisor - Keywords:
- GFS forecast skill
tropical disturbances
Wavetrak - Abstract:
- Tropical cyclones are high-impact weather phenomena that numerical weather prediction models struggle to forecast. To improve the skill of a model in forecasting tropical cyclogenesis, the first step is to evaluate how the model forecasts long-lived tropical disturbances, as many of these may provide the incipient vortex for tropical cyclogenesis given favorable conditions. The skill of the Global Forecast System (GFS) model at forecasting tropical disturbances during August and September 2011 in the North Atlantic Ocean is examined in this project. Disturbances identified in GFS forecasts are verified using disturbances found in Wavetrak 850 hPa vorticity analyses. Vortex contingency tables are used to document the skill of GFS in capturing disturbances identified in the Wavetrak analyses and to document additional disturbances forecasted by GFS that were and were not observed in the Wavetrak analyses. Characteristics (major and minor axis lengths, size, vorticity, and eccentricity) of the disturbances identified in the Wavetrak analyses matched or unmatched with GFS forecasted disturbances are compared to identify the types of systems that GFS did and did not capture. Additionally, the vorticity and size of verified GFS forecasted disturbances are evaluated against the vorticity and size of the corresponding Wavetrak disturbances to determine the skill of GFS in forecasting those particular characteristics. While nearly 60% of GFS forecasts verify (are seen in Wavetrak), GFS only forecasted around 20% of the disturbances observed in Wavetrak analyses. Further, GFS does not have good skill in forecasting the vorticity of disturbances. Future work could compare the synoptic-scale environments for matched and unmatched disturbances to discern why GFS misses forecasting a plethora of observed disturbances.