Achievement of USWRP Hurricane Landfall Research Goal

Abstract

he purpose of this brief communication is to report that the first research goal of the U.S. Weather Research Program (USWRP) Hurricane Landfall (HL) focus has been achieved. This research goal—to produce tropical cyclone track forecast guidance products with an improvement in accuracy of 20% over a 5-year period—was set following a Hurricane Landfall workshop during November 1997 (reported in the April 1999 Bulletin). Although this 20% goal was considered to be challenging in view of the relatively slow improvement since major advancements in dynamical models had been introduced in 1994 and 1995 (see chapter 4 of the WMO technical document, Global Perspectives on Tropical Cyclones), it was believed that the goal was achievable via advances in understanding of tropical cyclone motion, further improvements in dynamical models, and better use of existing and new observations. The overall goal of the USWRP is to improve forecasts of significant weather impacting the United States; this particular effort was a 5-year research goal, as it was recognized that it might take some additional time for the hurricane specialists at the Tropical Prediction Center/National Hurricane Center (TPC/NHC) to learn how to make effective use of the improved guidance. It should be noted that the degree of difficulty in the track forecast varies in different scenarios (generally easier in the deep Tropics and more difficult when the storm moves poleward, interacts with baroclinic features, and recurves into the midlatitude westerlies). Thus, the annual average error in a tropical cyclone basin with a relatively small number of forecasts may in part reflect the differences in difficulty. Nevertheless, the annual average track error at 72 h will be used as the metric for two of the dynamical models used at TPC/NHC, and a technique called CLIPER (named for a statistical blend of climatology and persistence) will be used as an indicator of the degree of difficulty. For the 1995–2003 period to be discussed here, the average CLIPER 72-h error was about 355 n mi, with a range from about 290 n mi during 1999 to almost 400 n mi during 2001 (Fig. 1). The number of 72-h forecasts also varies, with a range from only 39 during 1997 to 231 during 1995 (Fig. 1). This number is for a homogeneous sample that requires all of the three guidance products and the official forecast to be available so that a fair comparison can be made. Early in the period, the dynamical model guidance was only available at 0000 and 1200 UTC, but by the end of the period all of the guidance was also available at 0600 and 1800 UTC. Because the 0000 UTC dynamical model guidance is not available to the TPC/NHC until after the 0000 UTC forecast is issued, the comparisons here are with interpolated tracks from the previous forecast, which is what is actually available to the hurricane specialist.