2005 National Hurricane Center forecast performance

May 23, 2006

The "other" news from the NHC yesterday was the release of its annual verification report (PDF), which shows the accuracy of the Hurricane Center's forecasts and compares it to an average of the past ten years.  It also shows how the various forecast models performed against each other and the NHC forecast.

The accuracy of the track forecasts produced by the NHC was down slightly from last year's record setting performance.  This was despite an indication that the 2005 storms were more "well-behaved" (theoretically easier to forecast) than those of 2004.  The average errors for the 24, 48, 72, 96, and 120  hour time periods were 60, 106, 156, 156, 220, and 286 nautical miles respectively.  While higher than last year, it is lower than the 5 year and 10 year averages.

As far as individual storms goes the NHC was at its best during Dennis. The track errors through four days were less than 100 nautical miles and the error at  five days was  154 nm. Of the storms that survived long enough to have five day forecasts, Epsilon was the storm that gave the forecasters the most difficulty, with an average 3 day forecast error of 152 nm and a 5 day error of 544 nm.

 The story for intensity forecasting remains the same as before, where little to no progress is being made.  The errors for 24-120 hours were 11, 16, 20, 20, and 22 knots. Overall this is in line with the five and ten year averages. On average, these errors under-forecast the strength of storms.  That is a significant difference from the ten year average, which has (practically) zero bias, that is, on average the forecasts underrated storms just as often and as much as they overrated them. 

With regards to forecast models, the only guidance that bested the NHC's track forecasts averages was the guidance that consists of a consensus of the models.  The FSU Super-Ensemble, which was the best performer last year got beaten by the GUNA, which is a simple average of the GFDL, UKMET, NOGAPS, and GFS (formerly AVN) models. The two were nearly identical in their accuracy through 72 hours, however, the GUNA performed slightly better at 96 and 120 hours, which gave it the crown.  

As far as intensity forecasting goes, no guidance performed better than the NHC forecast (last year the Super-Ensemble edged out the NHC at most time periods). While it didn't top the humans, the SE did perform better than the individual guidance products at most periods. 


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: