ALTHOUGH SOME ASPECTS OF TROPICAL CYCLONE FORECASTS HAVE IMPROVED CONSIDERABLY DURING THE PAST DECADE, THEY ARE STILL PLAGUED BY SIGNIFICANT ERRORS. WHILE TODAY S 48-H POSITION FORECAST IS AS ACCURATE AS A 24-H FORECAST 10 YEARS AGO, FORECASTING CYCLOGENESIS REMAINS A CHALLENGE, AND INTENSITY FORECAST SKILL SIGNIFICANTLY LAGS THAT OF TRACK FORECASTS. THERE HAS BEEN LITTLE IMPROVEMENT IN OUR ABILITY TO PREDICT HURRICANE INTENSITY, AND PREDICTIONS OF RAPID INTENSIFICATION AND DECAY REMAIN PARTICULARLY PROBLEMATIC. THE INITIAL PROPOSAL FOR THIS GRANT WAS TO USE ASSIMILATION OF OBSERVATIONS FROM NASA S GLOBAL HAWK (GH) UNMANNED AIRCRAFT SYSTEM TO IMPROVE ANALYSES AND FORECASTS OF TROPICAL CYCLONES. IN PARTICULAR, WE PROPOSED USE OF AN ENSEMBLE KALMAN FILTER (ENKF) TO ASSIMILATE OBSERVATIONS FROM THE HURRICANE AND SEVERE STORM SENTINEL (HS3) EXPERIMENT. ENKF ANALYSES ARE USEFUL NOT ONLY BECAUSE THEY ACCURATELY CAPTURE THE INITIAL VORTEX, BUT ALSO BECAUSE THEY CONTAIN PROBABILISTIC INFORMATION THAT EXPRESSES FORECAST UNCERTAINTY ON A CASE-BY-CASE BASIS. THE PERTURBATIONS OF ENKF ANALYSES CAN ALSO BE USED TO INITIALIZE ENSEMBLE FORECASTS.