A tornado touches down near Elie, Manitoba, Canada, in June 2007
A tornado touches down near Elie, Manitoba, Canada, in June 2007. Credit: Justin Hobson, courtesy Wikimedia, CC-BY-SA-3.0

A quirky thing about tornadoes is that over the past 30 years, their reported frequency—at least going by U.S. reports—has increased 30%. Of course, the last 30 years also correlate with the rise of Doppler radar, population growth, and cell phones.

Despite the headline-grabbing statistic, scientists have concluded that the frequency of tornado strikes isn’t actually increasing. Instead, better reporting is filling in gaps where tornadoes might have otherwise gone unreported.

In the United States, experts agree that a handful of tornadoes go unreported each year. In Canada, where the vast majority of the population lives within 160 kilometers (100 miles) of the U.S. border, storm trackers miss quite a few more.

A map of Canada showing confirmed and probable tornadoes from 1980 to 2009
The map of Canada illustrating confirmed and probable tornadoes over 30 years shows that reported twisters mostly—but not entirely—track with population density in the country. Credit: Dave Sills, Environment Canada

“If you look at a map of tornado occurrence in Canada and you look at a map of population density in Canada, you’d say, ‘Well it’s kind of the same thing’,” says Greg Kopp, acting dean of engineering at Western University in London, Ont., Canada.

Of course, tornadoes are not converging on densely populated areas. It’s simply that sparsely populated areas have fewer ways to report a tornado.

Tornadoes rarely show up on radar since the radar beam is too big to resolve a feature as small as a twister. So most of the time, the only way to confirm a tornado is to see it.

Dave Sills, a severe weather scientist with Environment Canada, undertook a study about 5 years ago looking at a new set of data from the Canadian Lightning Detection Network, which can detect lightning across 95% of the country, correlating it with data on verified tornadoes and population density.

Where there’s a thunderstorm, there’s the possibility for a tornado. So Sills and colleagues tried to hash out exactly how many tornadoes were likely occurring but never detected.

“Missing” tornadoes, most of which hit remote logging camps or entirely uninhabited areas, offer tantalizing troves of lost data for forecasters.

The result? Although about 60 Canadian tornadoes are verified each year, it’s likely that more than 230 twisters touch down in the country during that time period.

The “missing” tornadoes, most of which hit remote logging camps or entirely uninhabited areas, offer tantalizing troves of “lost” data for forecasters. These scientists think they can get a better understanding of tornado risk to humans and infrastructure by tracking all the twisters in the country.

Northern Tornadoes Project

The Northern Tornadoes Project, started in 2016, attempted to identify all the missing tornadoes in northwest Ontario. The following year, the team attempted to track every tornado in all of Ontario, as well as significant tornadoes elsewhere in the country.

In 2019, the project is aiming to get data on every single tornado that touches down in Canada.

“It’s ambitious…but I think we’ve got it together now, and we’re going to give it a shot,” Sills says.

In forested areas, the team is using satellite data to look for hints of damage like blown-down trees and stripped-away vegetation.

“If we see a hint of damage, we’ll fly aerial surveys to resolve the trees and branches,” says Kopp.

The team is working to figure out how to gather data on not just whether a tornado occurred and what its path was but what its intensity was.

In the prairies, tornado tracking gets much more complex. Strong twisters, such as the one that struck Alonsa, Manitoba, in 2018, are so powerful that they leave a visible track through the grass. But smaller tornadoes don’t show up on satellite imagery in the plains, as the team found after an outbreak in southern Saskatchewan in July 2018.

“We had to rely on eyewitness accounts and farmers with cellphone videos, because [the tornadoes] didn’t leave marks that we could identify. We’re going to have to engage with storm chasers in some way; otherwise, we might still miss some,” Kopp says.

The project has already yielded some interesting results.

The region northeast of Lake Superior, for instance, is sparsely populated, and few tornadoes are reported there. Sills had wondered whether Lake Superior’s frigid temperatures were reducing convective activity, and thus storms and tornadoes, or if “there are tornadoes there and we just don’t know about them.”

Data from the Northern Tornadoes Project found very few tornadoes in that region, showing that the lake “shuts down convective activity. Thunderstorms are unable to really get going, so we don’t have too many tornadoes in that region,” says Sills.

The team hopes to use the data to provide ground truths to forecasters.

“When something happens that forecasters didn’t understand, like a tornado occurs with a storm they didn’t expect, having that data allows them to look at the relationships between the meteorology…and the tornadoes.”

”A lot of times, [forecasters] issue tornado warnings in areas of the north and never get any reports back….Verification data gives them a better idea of how well they’re performing, and once you know how well your forecasts are doing, you can work on improving the accuracy,” Sills says.

“When something happens that they didn’t understand, like a tornado occurs with a storm they didn’t expect, having that data allows them to look at the relationships between the meteorology…and the tornadoes.”

Public Awareness

In addition to providing better data to forecasters, the Northern Tornadoes Project aims to bring increased awareness to the public.

“In some regions of the country, there may be the attitude that ‘we don’t get tornadoes here,’” especially if many tornadoes are unreported, says Sills.

Canada has two “tornado alleys” (one extending from Central Alberta to Northwestern Ontario, and the other in Southern Ontario), but tornadoes can form anywhere. A richer trove of data could simply get the general public to take tornadoes seriously.

Jana Houser, an assistant professor of meteorology at Ohio University, studies how tornadoes form and how topography affects them. She’s not involved in the Northern Tornadoes Project but speculates that a fuller data set on tornado strikes would help climatologists investigate long-term tornado trends.

“If you don’t have the reporting numbers, then you don’t have confidence in the [answer to the] question of where tornadoes form. There might be geographic, subtle changes, or certain areas are seeing tornadoes earlier in the year.”

Houser adds that the meteorology community has “a pretty good understanding of the environments that are needed to produce tornadoes” but that there are some “subtleties….As a community, we are investigating why one supercell produces a tornado while the next one over, which is what appears to be a similar environment with a similar structure, doesn’t end up producing tornadoes. That’s the largest gap in our knowledge.”

It’s that gap that the Northern Tornadoes Project really hopes to tackle.

“A lot of what severe weather forecasters do depends on their library of conceptual models,” Sills says. “They do a lot of training on recognizing patterns in radar. Anything we can bring to the table that increases their knowledge in those areas and [helps them] to recognize different patterns really helps.”

—Rachel Kaufman (rk@readwriterachel.com; @rkaufman), Editor at NextCity.org

Correction, 25 March 2019: This article was updated to reflect that the data set from the Canadian Lightning Detection Network included tornado and population density data; that there are 60 verified tornadoes a year in Canada, not reported; and that Canada has two tornado allies.


Kaufman, R. (2019), Before Canadian scientists can study tornadoes, they have to find them, Eos, 100, https://doi.org/10.1029/2019EO118983. Published on 22 March 2019.

Text © 2019. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.