Just in time for Hurricane Preparedness Week, we have a story about an NSF-funded computer model that may greatly improve the accuracy of hurricane forecasts

NSF's Weather or Not!

Just in time for Hurricane Preparedness Week, we have a story about an NSF-funded computer model that may greatly improve the accuracy of hurricane forecasts

National Science Foundation

Transcript

A better forecasting method for hurricane season? This new model improves the accuracy of seasonal hurricanes by 23 percent

Interviewer: Charlie Heck

Interviewee: Xubin Zeng

Charlie: Most of the time, you wouldn’t bet your life on the accuracy of the weather forecast. But during hurricane season in the US--which lasts from June through October--that could very well be the case.

Charlie: With funding from the National Science Foundation, atmospheric scientist Xubin Zeng and a team at the University of Arizona have developed a model that improves the accuracy of forecasts for the North Atlantic and Gulf of Mexico hurricane season by 23 percent.

Xubin: If we can give a more accurate prediction of hurricane activities in the coming season, I imagine, this could affect the federal and state governments. For example, FEMA could be better prepared for the coming hurricane season. And the local governments, particularly over coastal regions, could be better prepared.

Charlie: I’m Charlie Heck at the NSF, co-editor of Science360’s news service. I first spoke with Zeng about the current hurricane prediction models.

Xubin: There are three approaches to do the seasonal forecasting of hurricane activities. The simplest one is statistical measures--using observational data. The second method is using computer models is called dynamic measure. The third measure would be a combination of both, called hybrid measure. Today, most of the weather forecasting centers use hybrid measures for forecasting.

Charlie: In the 21st century the number of hurricanes per season became more variable, with 15 occurring in 2005 but only two in 2013. Forecasting methods relied heavily on the state of the El Nino climate cycle, a 3-7 year cycle.

Charlie: So how is this new model different? Zeng and his team relied on a 40-70 year climate cycle  called the Atlantic Multidecadal Oscillation and the use of ocean surface wind over the Atlantic Ocean. They tested their model by using data from the 63 hurricane seasons to “hindcast” the number of hurricanes that occurred each season from 1900 to 1949.

Xubin: So we have actually talked with the Climate Prediction Center of NOAA. And they have given us the official review of our results. The overall review indicates that our results our outstanding. We are going to issue the forecast in early June. And quite a few organizations have contacted us and are waiting for our prediction. So we are eager to see what will happen.

Charlie: That was atmospheric scientist Xubin Zeng, at the University of Arizona. To learn more about this new forecasting model, the paper, “A new statistical model to predict seasonal North Atlantic hurricane activity,” was published online in the journal Weather and Forecasting on March 25, 2015. You can also search our top stories on news.science360.gov. I’m Charlie Heck, co-editor of Science360’s news service at the National Science Foundation.