"When we have tornadoes, we have lots of them," said Greg Carbin, warning meteorologist for the Storm Prediction Center in Norman, Okla. "It's boom or bust."
The 12-month period before last May set a record for the fewest significant tornadoes. But two years earlier, the nation also set a record for the most in 12 months.
Tornadoes can occur any time of year. All it takes is the right conditions. It varies by location, but the season tends to run from spring to mid-summer, with the biggest hotspots in Texas, Kansas, Oklahoma, Florida and Nebraska.
In the past few decades, the U.S. has averaged annually about 1,250 tornadoes of all sizes. Longer trends also show more tornado clusters recently. The number of days with at least one significant tornado is going down.
In the early 1970s, there was a single tornado about 150 days a year. Now it's about 100, said Harold Brooks, a meteorologist at the National Severe Storms Lab in Norman.
But the number of super-busy days with 30 tornadoes or more has gone up. In the 1970s, they came once every two years or so. Now we get those days three times a year, Brooks said.
"Something has been happening and we're not sure yet why," he said.
One possibility raised, but far from proven, is climate change.
Studies have shown through computer simulations that as the world warms, the United States should see slightly more thunderstorms that provide the instability needed for tornadoes, said Purdue University meteorology professor Jeff Trapp, who co-authored one of the studies.
But the other key ingredient in tornado formation is wind shear, created by different wind speeds and direction at high and low altitudes. Wind shear provides the rotation and organization for tornadoes, Trapp said.
While storminess should increase with warming, overall wind shear is projected to decrease, Trapp and Brooks said. However, because wind shear is expected to decline only on certain days, global warming could possibly mean fewer days with tornadoes, but more on those days, Brooks said.
Or the tornado clusters "could be random" with no cause just chance because tornadoes are small events that naturally vary, said meteorology professor Howard Bluestein at the University of Oklahoma.
Carbin calls tornado bunching "a puzzle that keeps going."
National Weather Service's Storm Prediction Center: http://www.spc.noaa.gov/
National Climatic Data Center on tornado averages and history: http://1.usa.gov/1fQRsBj
Seth Borenstein can be followed at http://twitter.com/borenbears