| description abstract | The vector motion of severe tropical cyclones (including storm, hurricane/typhoon stages) is forecasted by a numerical scheme which involves two steps: a. Numerical geostrophic steering of the center of the cyclone using the U.S. Navy Fleet Numerical Weather Facility's (FN WF) operationally produced smoothed isobaric height fields, called SR. The tropical perturbations are steered in 1-hr. time steps up to 72 hr., using winds derived from the SR analysis dated closest to warning time. SR 500 mb. in the Pacific and SR 700 mb. in the Atlantic gave the most accurate forecasts on tests of 10 northwest Pacific typhoons and all five north Atlantic tropical storms and hurricanes in the period Aug. 15?Nov. 1, 1965. Forecasts were made twice daily, 0600 and 1800 gmt, during this period using the best track information. b. Next, the numerical-steering prediction is objectively modified to adjust for bias (i.e., deficiency in both zonal and meridional motion) by utilizing errors made in the most recent. 12- and 24-hr. numerical-steering forecasts. Several modes of adjustment are employed; the most recent 12- (12- and 24-) hr. numerical-steering bias yields the most accurate correction of subsequent Atlantic (Pacific) forecasts out, to periods of 72 hr. The optimal Naval Post-graduate School (NPGS) technique produces forecast errors ranging from an average of 4.2 kt. for 12-hr. forecasts to 6.2 kt. for 72-hr. forecasts. The U.S. Navy's official forecast accuracy is excelled by the NPGS scheme for all time periods. Stratification of error statistics by area, trajectory, and stage of storm, intercomparison with ESSA's NHC-64 technique, discussion of merits and deficiencies of the research program relative to operational forecasts, and current experiments at FNWF are discussed. | |
