Forecasting the motion of northeastern Pacific tropical cyclones.

Abstract

Several operational methods of forecasting the motion of tropical cyclones over the northeast Pacific Ocean area are evaluated for three 1970 and nine 1971 hurricanes. OFFICIAL forecast accuracy is shown to excel that of TYRACK (1971 only) and HATRACK (both years). The MODIFIED HATRACK (MOD-HATR) forecast scheme developed by the Navy in Monterey, California, comprising a numerical steering component (HATRACK) and a statistical modification (correction for bias in HATRACK) is applied to the same 1970-71 operational data, with the result that the MODHATR accuracy, using 850-mb steering, is shown to be superior to HATRACK and TYRACK while only slightly inferior to the OFFICIAL forecasts. Specifically, MODHATR forecast accuracy lies in the range 7 to 5 kt for 12 to 72 hour forecasts, respectively, while the ratios of OFFICIAL to MODHATR errors range from 1 to .7 in 1970 for forecasts 12 to 48 hours and from .9 to .8 in 1971 for 12 to 72 hour forecasts, respectively. The number of 1970 forecasts evaluated in the test averaged 35 per forecast interval while the 1971 forecast sample ranged from 160 at 12 hours to 45 at 72 hours.