Evaluation of Hurricane Forecast Skills of NCEP GFS Retrospective Experiments for the FY2016 Implementation

NCEP is planning to upgrade its global forecast systems (GFS) in the spring of 2016. This upgrade includes changes in both data assimilation and the forecast model. For data assimilation, the current operational hybrid ensemble Kalman filter (ENKF) three-dimensional variational (3D-VAR) analysis will be replaced by a hybrid ENKF 4D-VAR analysis. Hurricane relocation will be run hourly instead of three-hurly. All-sky cloud radiances, AVHRR winds and aircraft moisture will be assimilated. Bias correction will be applied to aircraft observations. For the T1534 semi-Lagrangian Global Spectral Model (GSM), land surface physics will be updated to reduce summertime warm bias over the Great Plains found in the current operational model. A few other minor changes were also introduced to improve model performance. NCEP EMC is currently conducting a set of retrospective experiments to assess the performance of the new system. This presentation evaluates the forecast skills of hurricane track and intensity over the Atlantic, Eastern Pacific and Western Pacific basins in the past three years from May 2013 to the early spring of 2016. Forecast of cyclone genesis in the tropics will also be examined.