An Eight-Year Experiment in Improving Forecasts of Hurricane Motion

Abstract

The National Hurricane Center and the National Hurricane Research Laboratory joined forces in a n effort t o improvc techniques for forecasting hurricanc motion in the spring of 1959 when the latter moved its headquarters from West Palm Beach to Miami into offices adjacent to those occupied by the principal hurricane forecast office in the United States. Results now available from verification of forecasts made during the period 1954 through 1966 show that there has been a significant improvement in the accuracy of hurricane forecasts during the period of increased cooperation between the research and operational forecasting groups. This improvement is indicated by B reduction in the mean error of hurricane forecasts of approximately 10 and 12 percent, respectively, for the two principal hurricane forecast areas near the eastern coasts of the United States.

Trends in Forecasting the Tracks of Atlantic Tropical Cyclones

“…The eight-year period 1959-66, cited by Dunn et al (1968) as being associated with a substantial reduction in Atlantic tropical cyclone forecast errors, was a period when mid-level analysis was optimal. At that time, the NASA Manned Space Program with its associated down-range rawinsonde network was at its peak, Ocean Station Vessels were more abundant, frequent trans-Atlantic aircraft operating near 500 mb were present, and diligently hand-drawn analyses techniques were still practiced at the National Hurricane Center.…”

“…There have been major changes in factors (Gentry, 1981) that presumably should have had a profound positive or negative effect on tropical cyclone forecasting. In this connection, Dunn et al (1968) document a 10-12% decrease in 24 h forecast errors (excluding storms east of about 62°W) over the eight years, 1959-66. These improvements were due, according to those authors, to better cooperation between researchers and forecasters beginning in 1959, to the introduction of objective prediction models shortly thereafter, and to the existence of an expanding (at that time) upper-air observational network.…”

“…Although projections of storm tracks are made to 72 h, 24 h is generally considered to be the most important because, at this time, specific preparedness efforts must immediately be taken by interests in the expected path of the storm. Dunn et al (1968) conclude that there was a 10-12% decline in the 24 h tropical cyclone position forecast errors over the eight-year period 1959-66. Those authors attribute the improvement to better cooperation between operational forecasters and researchers, as well as to the use of objective guidance.…”

The National Hurricane Research Project: 50 Years of Research, Rough Rides, and Name Changes

“…The 1960 Atlantic hurricane season was the last one to occur without satellite coverage, because it fell between the operational periods of Television Infrared Observation Satellite (TIROS) platforms. It was not surprising that by the time Hurricane Donna was discovered, it was already at hurricane strength (Dunn 1961). NHRP flew eight sorties into Donna.…”

“…They modified it in 1967 (NHC-67), in time for an eight-year study of numerical computer forecast techniques (Dunn et al 1968) produced by forecasters and NHRL researchers. NHRL also continued to improve its computer simulations of mature tropical cyclone circulation (Rosenthal and Koss 1968) and to model tropical cyclogenesis (Rosenthal 1964 (Holland 1970).…”