Scientists Are Probing Tornadoes With Drones to Save Lives

By Brian Kahn on at

This spring saw one of the worst tornado outbreaks in recent years. More than 300 twisters touched down across the US, scarring the landscape, causing widespread damage, and at least eight fatalities.

The widespread danger also offered an opportunity for a group of scientists looking to help ensure people in tornadoes’ paths have more time to get out of harm’s way. The outbreak represented a vast trove of data waiting to be collected. Armed with a suite of high tech radars, car rooftop-launched fixed wing drones, laser, and even an aerial assist from Hurricane Hunter aircraft, these researchers spent weeks on the road chasing down storms all with the goal of figuring out why some storms spawn tornadoes and others don’t. The end goal is to improve forecasts that can help save lives.

“You can’t forecast a phenomenon you don’t understand,” Adam Houston, an atmospheric scientist at the University of Nebraska who helped put together the research team, told Gizmodo. “The aim of this project is to improve the understanding of supercells so forecasters will have more knowledge.”

Supercell storms are distinguishable from other thunderstorms by their large-scale rotation. Photo: Greg Lundeen/Wikimedia Commons

Supercells are storms that can spawn tornadoes, damaging winds, hails, and walls of rain. They look like motherships and are a favourite photographic target of storm chasers. But the thing is, not all supercells spit out tornadoes, and scientists aren’t exactly sure why that is.

“You can’t forecast a phenomenon you don’t understand.”

The US is home to three-quarters of all the tornadoes that form around the world in any given year. It also has a dense network of weather stations, radar, and other means of observing storms. But that network is fixed in place and while tornadoes and supercells do occasionally buzz by or even right next to all those weather surveillance systems, it’s a rare occurrence and doesn’t provide nearly enough data for forecasters to work with. While forecasts have improved—the lead time for forecasting a tornado has gone from three minutes to fourteen minutes over the past four decades—having even more advance warning could obviously help save lives.

Researchers load instrumentation onto a tracker vehicle before a drone’s takeoff.Photo: Roger Laurence (Courtesy University of Colorado Boulder)

That’s why Houston helped put together Project TORUS, basically a roving pack of weather monitors strapped to trucks and SUVs that allowed researchers to get up close and personal with nature’s most violent weather. But of course if you really want to understand what’s going on in the atmosphere, it helps to sample it directly. So Project TORUS’ panoply of high tech weather monitors also includes fix-wing autonomous drones that can be sent into the storm while driving using a pneumatic launcher.

This merry band of weather monitoring equipment spent five weeks this spring prowling Tornado Alley in search of storms. Each day, the group of 60 researchers, students, and assistants would wake up (usually at an interstate hotel) and begin going over weather data. The meteorologists would create a forecast, identify an area where severe weather was likely to pop up that day as well as the next, and brief the team before decamping to said location, sometimes driving hundreds of miles to get there. After arriving, they would wait into the late afternoon when heat blossomed, forcing air to rise and knock into the colder atmosphere above spawning massive thunderheads and, occasionally, tornadoes. Then it was wash, rinse, repeat the same process the next day. If it sounds like the premise to “Twister,” well, it is basically. Just don’t call them storm chasers.

Photo: University of Colorado Boulder

“We usually say we’re severe storm researchers,” Eric Frew, an aerospace engineer at the University of Colorado who has spent 15 years refining the drones used as part of Project TORUS, told Gizmodo. He and Houston both said their interest wasn’t in getting as close to possible to tornadoes for a money shot to sell to the nightly news. Rather, it was to collect the best data possible.

To do that, the group had a unique plan of attack. Previously Houston and Frew had worked on a project flying one drone along the right flank of supercells. The group would launch from the ground and then the drone would autonomously follow within a half mile of a GPS tracker on an SUV, continuously measuring temperature, pressure, humidity, wind direction, and wind speed as it went.

That’s good and all, but taking measurements from the right flank only tells part of the story. Frew likened a supercell to a fist with the thumb stuck out. If you do that with your right hand, the right flank is that outside of your thumb. Looking at that leaves the whole other side of your fist obscured. In the case of supercells, that inside part of the thumb—or in supercell parlance, the left flank—is one of the most important but least understood areas for tornado formation. Gathering data on how it actually works to create tornadoes is crucial, so the team devised a new drone-launching system that would streamline the process and set out this spring with three separate drones.

Photo: University of Colorado Boulder

In addition to observing the right flank of the storm, the team also launched drone to monitor the left flank and another 30-40 miles in front of the storm. Together, the three drone missions created some of the most detailed measurements of supercells ever gathered.

The group got within two miles of a tornado on one pass, but tried to keep their distance from the heart of areas where tornadoes were likely to form to avoid recreating “Twister’s” final scene and getting caught right under a tornado. The measurements in the vicinity, though, can provide vital clues about supercells and the conditions when tornadoes do (or don’t) form.

“We usually say we’re severe storm researchers.”

Despite no getting tossed about directly by tornadoes, the drones encountered their fair share of rough weather. Frew said they flew through areas buffeted by 75 mph winds on some of their 20 flights this spring. Though their bodies are largely made of strong foam, the drones were able to withstand the winds, hail, rain, and other harsh weather in the atmosphere. But the worst conditions they encountered were actually much closer to the ground. In fact, they were the ground.

“The landing is the most violent part,” Frew said, noting that the drones usually came down in rough fields or gravel roads without landing gear since it would add unnecessary weight. “Hitting the ground will damage these systems.”

That’s why he and the aerospace team ensured the unmanned aerial vehicles all had interchangeable parts. If one sensor snapped off or a wing contorted, there were spare parts and backup drones to quickly pick apart and replace it.

Photo: University of Colorado Boulder

After five weeks of travelling down interstates, dirt roads, and every other drivable surface in between, the band of researchers went back to their respective institutions in Colorado, Texas, Nebraska, and Oklahoma to begin to look at reams of data they’ve collected. Project TORUS is a multiyear process, and the crew will be back out seeking storms again next spring.

Ultimately the data they’re collecting will be examined for hints at why some storms form tornadoes and others don’t. It will also be compared against data collected at fixed weather stations to see if there are any correlations forecasters can use to improve warning times for tornadoes. An ancillary goal of the project is also serving as proof of concept of what a future weather monitoring system for the U.S. could look like.

“The technology we’re using for this project’s unmanned aircraft part could be used in theory in a next generation meteorological surveillance network,” Houston said.