The Impact of Satellite-derived Winds on Gfdl Hurricane Model Forecasts

A series of experimental forecasts are performed to evaluate the impact of satellite-derived winds on numerical hurricane track predictions using the GFDL model. Over 100 cases are examined from 10 different storms covering 3 seasons (1996-1998), enabling us to account for the large case-to-case variability in the forecast results when assessing the wind impact. On average, assimilation of the GOES winds reduces track error at all forecast periods. The relative reductions in track error range from ~5% at 12 hours to in excess of 12% at 36 hours. Statistically significant reductions in track error are noted for the 24, 36, and 72 hour forecast periods. A composite analysis of the initial flow fields suggests that the reduction in track error may be associated with the ability of the GOES winds to more accurately depict the strength of vorticity gyres in the environmental flow.