Researchers have found that the most tornado-prone region of the country is shifting east, but they can't say how much global warming is to blame.

Rescue efforts resumed on Monday in Lee County, Alabama, after a series of powerful tornadoes carved a deadly path through its rural communities on Sunday. At least a dozen touched down across Alabama and Georgia over the weekend, leveling more than 1,000 homes and other structures. Lee County Sheriff Jay Jones told reporters that it looked as if someone "took a giant knife and scraped the ground." The storms left 23 people dead, dozens injured, and countless still missing, making Sunday the deadliest day for tornadoes in almost a decade.

The American South is no stranger to tornadoes, which under the right weather conditions can occur anywhere on Earth, according to the National Oceanic and Atmospheric Administration. But the weekend's twisters were outside of what's historically been the most tornado-prone region of the United States, an ill-defined stretch of the southern plains known as Tornado Alley that covers ground between Texas, Iowa, Kansas, and Nebraska.

It's too early in the year to draw any conclusions about what this might mean for the 2019 tornado season, which peaks in the late spring, but the Weather Channel noted last year that the 2018 season was unusual, with tornadoes in both Iowa and Louisiana far outpacing occurrences in Tornado Alley. A record nine tornadoes touched down in Connecticut last year; Illinois and Pennsylvania were both hit with surprising, off-season twisters; and Wyoming was hit with its first severe tornado in more than 20 years.

It's not just tornadoes: The U.S. seems to have entered an era of increasingly erratic and extreme weather of all kinds, from record-breaking hurricanes, to endless wildfire seasons, to extreme heat waves—all linked, in some way, to climate change. Which raises the question, could climate change also be influencing the behavior of tornadoes?

According to Noah Diffenbaugh, a professor of earth system science at Stanford University, there's been a "scientific revolution" over the last 15 years in researchers' ability to detect the influence of global warming on individual extreme events such as heat waves or droughts. But the same cannot be said for tornadoes.

One reason is that the long-term, observational record for tornadoes is much less reliable than it is for other climate variables, such as temperature or precipitation. "We've been observing temperature objectively for a long time with networks of thermometers and weather balloons and then satellites," Diffenbaugh says. "Tornados are harder to observe remotely, and certainly before the radar era we have a much less-developed, older record that's based on first person accounts from tornado spotters or from destruction where tornadoes occurred."

Without long-term data researchers can't draw any conclusions about trends over time, or evaluate what the chances are that an individual, extreme event was just another case of natural variation or was made worse by global warming.

"Experimentation is a pillar of scientific inquiry, and we can't stick the Earth in a lab and run controlled experiments," Diffenbaugh says. "And so we use climate models and weather models to run the kinds of experiments that we would run in petri dishes if we were studying something smaller than planet Earth, and then compare those results to what has happened historically. ... [W]ith respect to tornadoes, we have limitations both in the observational record and in our modeling capabilities."

But researchers are slowly closing those gaps now, thanks to radar observations of tornadoes over the past couple decades, which they can pull together to look for patterns. Diffenbaugh points to the work of Columbia University's Michael Tippett, whose research has shown that tornado outbreaks, or clusters of tornadoes occurring together within a short timespan, have become more common and more powerful since the late 1970s—though Tippett and his colleagues did not say that climate change played a role in this change.

In a separate study published last year, researchers at Northern Illinois University and NOAA’s National Severe Storms Laboratory set out to determine if the total number of tornados across the U.S. had changed over time. They looked at tornado frequency data dating back to 1979, and found that, since then, the number of twisters has been rising in the southeastern states of Mississippi, Alabama, Arkansas, Missouri, Illinois, Indiana, Tennessee, and Kentucky, and falling in the traditionally tornado-prone states of Texas, Colorado, and Oklahoma. In other words, while the number of tornadoes has stayed relatively stable over the last four decades, Tornado Alley has shifted eastward.

"When you close your eyes and picture a tornado, your first thought is maybe Texas or Oklahoma or Kansas—it's not in Alabama and Georgia or Mississippi," says Victor Gensini, an assistant professor at NIU and lead author on the study. "We've shown that those areas are just as ripe for tornado activity as the central Great Plains is."

Tornadoes tend to form under very specific atmospheric conditions—wind shear, or differences in wind speeds and direction; atmospheric instability; and moisture all fuel storms, for example—and there are reasons to believe that climate change is making those conditions more likely. For example, Gensini says that storm-scale models suggest that, as the Earth warms, "updrafts in storms will get stronger, and if you have a stronger updraft you're much more likely to get severe weather, including tornadoes." But that just means that climate change could potentially create the opportunity for more severe storms and tornadoes to form, it doesn't guarantee that storms will occur, Gensini says.

And whether or not climate change is influencing the frequency or location of tornadoes may not be the most relevant question when it comes to mitigating the impact of the destructive phenomenon. "We get caught up on the climate aspect, but the real issue going forward with tornadoes—and hail storms and hurricanes and insert your favorite natural disaster—is the fact that we have more human exposure," Gensini says.

"What we've documented in our research is we're having about the same number of tornados but we're having an exponential increase in losses, and really the only thing that describes that is population density," he says. "We have more targets to hit."

Indeed, the eastward shift of tornadoes toward more densely populated regions where homes and structures weren't built with twisters in mind could lead to increasingly destructive and deadly storms. For example, though it's not clear yet how many of the fatalities over the weekend occurred in mobile homes, overall, history shows that an unexpectedly high number of deaths during tornadoes occur in the weak-framed structures.

"Research shows that mobile home fatalities account for nearly half of all tornado fatalities," Gensini says, "yet mobile homes only account for 7 percent of the United States' housing stock. That right away tells you that we have a disproportionate share of fatalities in mobile homes."

Strengthening building codes or implementing safe-shelter requirements for mobile home park residents could both help to mitigate the loss of life associated with tornadoes, Gensini says, but just as important is outreach to the at-risk communities. Forecasting extreme weather is improving all the time, but it's always a challenge to get those forecasts into the hands of vulnerable populations in time—and to overcome the human tendency toward complacency in the face of rare events like tornadoes.

"Forecasting what humans are going to do when they get the warning information is sometimes harder than forecasting actual weather," Gensini says. "I can make the best forecast in the world, but if you don't take that information and act on it, then is it really a good forecast?"