Restoring confidence in NHC

There are no threat areas in the tropical Atlantic to discuss, and none of the computer models are forecasting tropical storm formation over the coming week. In the Pacific, an exceptionally large Category 4 typhoon, Man-Yi, will pass close to Okinawa on Friday, and hit Japan on Saturday. Winds at Okinawa have been as high as 50 mph with gusts to 70 mph.

Some links on Man-Yi, sent to me by Jim Edds:

Live camera feed from Southern Okinawa:

click on the pic arrow on the bottom right.

Okinawa radar.
click on Okinawa to zoom in - awesome shot

Camera feed with some audio.
click on the first "no image" box then select "report 17/66"


Latest satellite image of Typhoon Man-Yi, courtesy of NOAA.

Restoring confidence in the NHC
Interim National Hurricane Director Dr. Ed Rappaport has two immediate tasks--restoring morale fractured by Bill Proenza's turbulent 6-month tenure, and restoring public confidence in the Hurricane Center's ability to do their job. With the steadying influence of Dr. Rappaport, a highly respected and talented hurricane scientist, I expect that the staff of NHC will put out their best hurricane forecasts ever this season. Aiding in this endeavor will be the availability of a new hurricane tracking, intensity, and storm surge model called the HWRF--Hurricane Weather and Research Forecast Model. In addition, several of the other reliable models used by the forecasters, such as the GFS and GFDL, have had upgrades since last hurricane season. Furthermore, the Air Force Hurricane Hunters will be carrying the SFMR instrument for the first time, which can measure winds speeds at the ocean surface everywhere the aircraft fly.

Forecasters at the National Hurricane Center made their best track forecasts ever for storms in the Atlantic in 2006. The mean track errors for 12 to 72 hour forecasts were 15% - 20% lower than during 2001-2005. Track errors for Atlantic storms have improved about 50% in the past 15 years (Figure 1), a remarkable achievement that has undoubtedly saved lives and hundreds of millions of dollars. The track error in 2006 for a 24 hour forecast was 58 miles; 112 miles for a 48 hour forecast; and 171 miles for a 72 hour forecast. Track errors for 96 and 120 hour forecasts were 236 miles and 305 miles--the second best on record (2003 set the record). NHC's long-range 120 hour forecasts had a significant bias to the west of 94 miles--about double the bias of what the computer models were forecasting. Thus, when the models correctly called for systems to recurve out to sea, NHC human forecasters tended to resist following what the models were saying.



Figure 1. Track forecast skill since 1990 in the Atlantic for the official NHC forecasts. Track errors are given in nautical miles (100 n mi = 115 miles). Skill is rated compared to a "zero skill" forecast using NHC's CLIPER5 model. The CLIPER model (short for CLImatology and PERsistence) is a model that makes a forecast based on historical paths hurricane have taken, along with the fact that hurricanes tend to keep moving in the direction they are going (i.e., their current motion persists).

Intensity forecasts
Intensity forecasts since 1990 have shown little or no improvement, and 2006 was no exception (Figure 2). One encouraging result was the emergence of the GFDL's intensity model as the best intensity model for 2006. This is the first time that a non-statistical model has made the best intensity forecasts. With the major improvements that were added for the 2007 version of the GFDL, plus the availability of the HWRF model, I am hopeful that this year will see the first noticeable improvement in intensity forecasts since 1990.



Figure 2. Intensity forecast skill since 1990 of the official NHC Atlantic forecasts. Intensity errors are given in knots (10 knots = 11.5 mph).

Jeff Masters