A University of Arizona-led hurricane forecasting team predicts 19 named storms and 10 hurricanes – including five major hurricanes at Category 3, 4 or 5 – during this year's Atlantic hurricane season.

Average annual hurricance activity since 1980 is 13 storms and seven hurricanes, two of them major. Forecasters base averages on the last 30 years of data.

The primary reason this season is expectd to be very active is that sea surface temperatures in the Atlantic Ocean are the warmest they've been since 1993, said forecast creator Xubin Zeng, director of the university's Climate Dynamics and Hydrometeorology Center, professor of atmospheric sciences and the Agnes N. Haury Endowed Chair in Environment in the Department of Hydrology and Atmospheric Sciences.

Zeng and his co-author and former graduate student Kyle Davis developed the model, and it has proven to be accurate in the past, as documented in two scientific papers.

This is the first time Zeng and Davis have issued their predictions in April. The researchers usually issue their forecast in June, but felt it was important to release their findings early to allow government agencies and the general public time to prepare.

"Hurricanes occur every year. You cannot change that, but you can get better prepared for the season," Zeng said. "Preparedness will be more challenging this year because all our resources have been devoted to fight the COVID-19 pandemic."

This year's Easter tornado outbreak killed more than 30 people in the American Southeast. While the spring tornado activities are not related to hurricane activities in the fall, Zeng noted one similarity: Hurricanes are generated by warm tropical Atlantic waters, and the recent tornado outbreak was powered by warm waters in the Gulf of Mexico.

Zeng and Davis' model combines, among other things, seasonal forecasts of sea surface temperature, wind, pressure, humidity and precipitation from the European Centre for Medium-Range Weather Forecasts with machine learning and the researchers' own understanding of hurricane activities.

"It's like cooking," Zeng said. "You need good ingredients: A good forecasting model – that's the European model, our own insights and physical expertise, and then the fancy machine learning. These three things together can make a good dish."