Further verifications of and experiments to improve the MODIFIED HATRACK scheme for forecasting the motion of tropical cyclones

Abstract

The MODIFIED HATRACK (MODHATR) scheme for forecasting tropical cyclone motion consists of a numerical steering component using geostrophic winds derived from Fleet Numerical Weather Central's SR height field to steer the storm center, and a statistical modification component to correct for bias and improve forecast accuracy. MODHATR forecasts from the 1969 and 1970 North Atlantic hurricane seasons are analyzed, and average errors presented and compared to earlier years' results. MODHATR forecasts are shown to be superior on the average to OFFICIAL forecasts, NHC-67, and TYRACK forecast schemes for forecast intervals to 48 hours, with relative accuracy of MODHATR decreasing with time. Results of an experiment to improve the statistical correction for bias are reported. A level- and mode-selection scheme is investigated which offers some promise of improving forecast accuracy at late forecast intervals. A comparison is made between warning-time and synoptic-time initial-position errors showing synoptic-time positions to be more accurate for initiating MODHATR forecasts.