How accurate are pre-season hurricane landfall forecasts?

Last week, we ran a story on the April forecasts for the 2013 Atlantic hurricane season (June 1 – Nov. 30). Colorado State’s Philip Klotzbach and Bill Gray, WeatherBell’s Joe Bastardi and others are calling for a very active year with above-average numbers of hurricanes. Toward the end of his post, Jason pondered the accuracy of these preseason predictions, pointing out that forecasters offered a bearish outlook last year – a call that subsequently busted as the third-highest number of tropical storms on record developed.

The landfall prediction game is a bit difficult to score given the different ways forecasts are presented, but an analysis  through the years reveals a mixed record. WeatherBell’s Joe Bastardi (formerly of AccuWeather) has more often than not overpredicted the number of landfalling storms. Gray and Klotzbach’s forecasts for landfalling hurricanes have been strong, but they haven’t skillfully predicted the drought in major hurricane strikes.

This year may prove challenging for forecasters if an El Nino event unexpectedly takes hold during the summer, suppressing hurricane activity. Gray and Klotzbach believe that neutral ENSO conditions will remain in place instead, and that the negative (or cold) phase of the Pacific Decadal Oscillation (PDO) and warmer-than-normal ocean temperatures in the tropical and North Atlantic will stimulate a hyperactive 2013 hurricane season.

The International Research Institute and NOAA/Climate Prediction Center’s latest roundup of model forecasts for ENSO shows a neutral event – that is, neither El Niño nor La Niña – continuing through the summer.

The International Research Institute and NOAA/Climate Prediction Center’s latest roundup of model forecasts for ENSO shows a neutral event – that is, neither El Niño nor La Niña – continuing through the summer.

Bastardi cites the past more prominently in his forecast, arguing that “multiple major hits” are likely along the U.S. coast thanks to a combination of the positive (or warm) phase of the Atlantic Multidecadal Oscillation (AMO)– a decadal signal favoring the East Coast, Bastardi writes – and, again, the negative PDO. He expands on his argument, stating that the coincident warm AMO/cold PDO cycle in the 1950s drove up the tally of storm hits on the Northeast coastline.

WeatherBell graphic juxtaposing the number of decadal Northeast U.S. storm landfalls against the summer phase of the PDO. Note that the 1950s – a decade marked by negative PDO conditions – featured the highest number of storm hits within the region.

WeatherBell graphic juxtaposing the number of decadal Northeast U.S. storm landfalls against the summer phase of the PDO. Note that the 1950s – a decade marked by negative PDO conditions – featured the highest number of storm hits within the region.

Searches for Bastardi’s preseason landfall forecasts yield specific numbers since 2006. His predictions were spot-on for 2008 and last year; he nailed the exact number of landfalls (three) in 2008 and virtually did the same in 2012 (when there were two landfalls – I am counting Sandy as a landfall since it delivered hurricane-force winds to the coastline). Verifications are not so good for all other years, as Bastardi overestimated the number of hurricane landfalls by five in 2006, 2007, 2010 and 2011, and by three in 2009.

*Bastardi and his AccuWeather colleagues presented the forecast from 2006-2010, while Bastardi collaborated on the outlook with his WeatherBell team in 2011 and 2012.

*Bastardi and his AccuWeather colleagues presented the forecast from 2006-2010, while Bastardi collaborated on the outlook with his WeatherBell team in 2011 and 2012.

The “half” numbers of hurricane landfalls are used because, for the 2007, 2011 and 2012 seasons, Bastardi offered a forecast range (i.e. “6 or 7” landfalls in 2007). Green bars and associated numbers represent the difference between the forecast and verified totals.

Gray and Klotzbach’s landfall forecasts issued in April are based on a probability scheme,  making them challenging to assess. Moreover, their probabilities for landfalling hurricanes have almost always been higher than both the 100-year (1900-2000) average and 50%.

The number of yearly hurricane (H) and major hurricane (MH) landfalls that occurred along the U.S. coastline since 1999.

The number of yearly hurricane (H) and major hurricane (MH) landfalls that occurred along the U.S. coastline since 1999.

The way I prefer to judge Gray and Klotzbach’s forecasts, therefore, is to consider the seasonal number of landfalls and then determine if their probable call for one or more landfalls was accurate. Hurricanes have struck the U.S. coastline in all but four years since 1999, so that fact would seem to validate Gray and Klotzbach’s accuracy. It’s interesting to note that the one year – 2012 – for which Gray and Klotzbach called for landfall probabilities lower than the climatological average, two hurricanes – Isaac and Sandy – hit the nation.

How about Gray and Klotzbach’s track record on major hurricane (Category 3, 4 and 5) landfalls? Here, their performance lags. The probabilities, much like they were forecast for all hurricanes (Categories 1-5), have almost always been higher than the long-term average of 52% (which, in itself, suggests a greater than 50/50 chance of a major hurricane striking the U.S. coastline in any given season). Take another look at the table above. A major hurricane landfall has rarely occurred in the past 14 years. By that measure alone, Gray and Klotzbach bullish calls have not verified well.

For the total number of hurricanes and major hurricanes per season, Gray and Klotzbach have done fairly well with their April predictions. Eleven of Colorado State’s 18 forecasts have come within three hurricanes of nailing the seasonal total and, on four occasions, they have only missed the cumulative number by one. They hit the mark in 2008, which, as Bastardi’s record also attests, was a good year for preseason prognosticators. Of course, Gray and Klotzbach posted a few wayward forecasts as well: for 1995 (a five-storm underestimate), 2005 (undershot the total by eight) and in 2012 (fell short by six hurricanes).

Gray and Klotzbach's forecast number of hurricanes is plotted against the observed total.

Gray and Klotzbach’s forecast number of hurricanes is plotted against the observed total.

On major hurricanes, Gray and Klotzbach have made six forecasts that failed to fall within two storms of the final tally. This should not obscure the fact that they’ve also posted four perfect forecasts and six others that approximated the seasonal total within one storm.

Gray and Klotzbach's forecast number of major hurricanes is plotted against the observed total.

Gray and Klotzbach’s forecast number of major hurricanes is plotted against the observed total.

As outside observers, we have the luxury of sitting back and grading forecasts that most of us do not make on our own. That’s why it’s important to appreciate the craft that these gentlemen – Gray and Klotzbach, Bastardi and several others – try to refine each and every year. Gray and Klotzbach speak to the inherent difficulty of their endeavor with the following words from their 2013 seasonal forecast:

“It is also important that the reader appreciate that these seasonal forecasts are based on statistical schemes which, owing to their intrinsically probabilistic nature, will fail in some years. Moreover, these forecasts do not specifically predict where within the Atlantic basin these storms will strike. The probability of landfall for any one location along the coast is very low and reflects the fact that, in any one season, most U.S. coastal areas will not feel the effects of a hurricane no matter how active the individual season is.”

Also on Capital Weather Gang

A year of outbursts from the sun