Climate change is real. And scientists are confident that we’re going to see more heat waves, droughts, and deluges as the atmosphere continues to warm. Hurricanes may be more frequent, too, or more powerful—either way, a rising sea will make them more devastating.

The behavior of tornadoes, however, has proven more difficult to predict, a fact thrown into sharp relief by the deadly twister that hit Moore, Oklahoma, on Monday with only a 16-minute warning. Researchers says it's simply too difficult to predict with certainty how a changing climate will affect the frequency of their wrath. 

In part that’s because tornadoes, compared with other weather events, are relatively small. A heat wave or a drought or even a hurricane affects hundreds or thousands of square miles; tornadoes, famously, can level a house and leave the neighbor’s untouched. Even now, advanced weather-forecasting models work with grid squares that are several miles large and can at best say whether a particular thunderstorm might be likely to spawn a tornado several hours or days in advance.

There’s also not much quality historical data on tornadoes. The Fujita scale wasn’t invented until the 1970s, at which point the National Oceanic and Atmospheric Administration went back and retroactively rated tornadoes based on historical records of the damage they wrought. The scale was changed again in 2007, but it’s still essentially a way of measuring damage, so only tornadoes that pass through populated areas get recorded. Then there’s the compounding factor that tornadoes vary a lot from year to year—from June 2010 through May 2011, we had over 1,000 EF1 or stronger tornadoes, a record high, and from May 2011 through April 2013, we had 217, a record in the opposite direction. A short historical record, inconsistent reporting, and wide variability: it all makes for a noisy data set.

Because of these limitations, researchers have been focusing on how climate change will affect the atmospheric conditions that give rise to tornadoes.

Tornadoes need a few basic ingredients. First, you need warm, humid air beneath a layer of cool, dry air. Second, you need those layers to be traveling at different speeds or in different directions, a phenomenon called wind shear. These two conditions happen a lot in the Plains states, as the jet stream pushes cool, dry air from the Rockies over slower-moving humid air from the Gulf of Mexico. When a disturbance like a cold front or a low-pressure system causes the two layers to interact, the hot layer tries to rise, and you get a rotating column of air that can turn into the sort of violent thunderstorm that sometimes spawns tornadoes.

Climate change will affect these two ingredients in opposite ways, says Harold Brooks, a researcher at the National Weather Center in Norman, Oklahoma. On one hand, warmer air can hold more moisture than cool air can, so moisture content will increase with global temperatures. On the other hand, wind shear is expected to decrease as the poles get warmer. Researchers have debated for several years which factor will win out. Brooks says that most recent models point to higher moisture content resulting in more strong thunderstorms, but the lower wind shear means a smaller fraction of them will spawn tornadoes. Whether there will be so many more thunderstorms that they end up creating more net tornadoes, despite the lower wind shear, is unclear. “We’re not sure which will win out,” says Brooks, but either way expect more violent storms.

Stanford University scientist Noah Diffenbaugh says that recent models show higher moisture content winning out over decreasing wind shear, resulting in a higher likelihood of severe thunderstorms that could spawn tornadoes, but he cautions that the research is still in the early stages.

“The response of tornadoes to global warming is one of a small handful of outstanding and really important unknowns in terms of how global warming will likely impact humans,” says Diffenbaugh.

That’s partly due to a historical gap between the people who study climate and the people who study extreme weather, which has started to close only recently, according to Adam Sobel, who studies atmospheric science at Columbia University. The reason for it is partly technical and partly cultural: until recently scientists didn’t have the computational power to model relatively small weather events like hurricanes and tornadoes, and the people who studied climate and the people who studied extreme weather tended to belong to different institutions. That started to change with hurricanes a decade ago, but with tornadoes it’s just starting to happen, thanks in part to more powerful modeling tools and greater cross-discipline communication.

“The community is aware we need to understand it,” Diffenbaugh says, “and we’re working hard.”