Penn State professors Paul Markowski and Yvette Richardson have an unusual job. They launch weather balloons into supercell thunderstorms.

These rotating storms are what most commonly produce tornadoes, and the data these scientists collect could improve their forecasts.

“We’re trying to go where traditional instruments haven’t gone before,” Markowski said. “This is something new.”

The team aims to launch miniature weather balloons about 12 to 16 inches in diameter into the air at the periphery of the storm. Attached to each balloon is a probe that measures temperature, dew point and relative humidity. Markowski hopes his team can use the data to re-create the near-surface temperature field of the environment surrounding and within the storm.

“The horizontal temperature gradient in storms has a huge role in determining buoyancy,explained Markowski. Buoyancy describes the air’s affinity to rise. Hot air balloons are positively buoyant. But if you suddenly chilled all the air inside, it would become negatively buoyant and your balloon would sink. The same premise applies to pockets of air in the atmosphere. If air is buoyant and rising, a tornado is more likely to form.

“Buoyancy can in turn influence vorticity,said Markowski. Vorticity describes the amount of spin in the air present at different levels of the atmosphere. It’s one of many ingredients that interact to ultimately contribute to tornadogenesis, or tornado formation.

“We’re not launching balloons into tornadoes themselves,” said Richardson. “We’re sampling regions that the inflow air will pass through en route to the tornado.”

Launching balloons may seem like a straightforward process, but designing a balloon-carried apparatus that can transmit data in real time is an entirely different challenge.

“It used to be that we’d need a different frequency for each probe we launched,” Markowski said. “When we’re launching up to two balloons per minute, that wouldn’t be [feasible without the new technology.]”

In recent years, however, it’s been possible to partition “time slots” on an individual frequency. “Let’s say you have 100 probes, and you want data from each of them every second,” said Markowski. “If you have precise time-syncing, you could have each one transmit in a different hundredth-of-a-second window.”

It’s sort of like a conference call. If 50 people were in the same phone conversation talking at once, no information would get through. But if the time was divided up and each person took a turn to talk, all the information would get through.

That technology has made an endeavor like that of the Penn State team’s possible. But it comes with a trade-off. “The more probes we launch, the more things we have to keep track of, and the less resolution we can get time-wise.”

The team also had to design a setup in which the balloons would rise slowly. "If they ascend too much, they overshoot key low-altitude areas of the storm, which are the most critical to measure. We have to get it just right,” said Markowski. “The balloons have to actually take off and rise, but not too fast. And even a few raindrops — the weight of a penny or less — can make a balloon . . . stop, or even start to descend.”

“Each probe we launch has a cost,” Markowski said. “Radar you can just scan. There’s nothing to lose by just scanning any storm. But we have to be careful to only launch balloons into worthwhile storms. Once launched, [the balloons] are gone for good.”

The Penn State team’s project, which is sponsored by the National Science Foundation, has been a long time in the making. “Some people have asked why we’re not using drones. The Federal Aviation Administration doesn’t allow people to fly drones into places without visual contact,” he explained. “Besides, drones aren’t really disposable.”

This year the team sampled a number of storms, taking advantage of the active severe season to collect vital data. “We were out there for 11 days this year and got to deploy balloons on multiple strong supercells,” Richardson said. “We went during 2017, but it was a much quieter season.” Markowski says the project is still in the phase of “beta testing” to an extent, but he estimates they’ve launched about 220 probes since the start.

Markowski hopes that the team’s three-dimensional temperature mapping will help make progress on one of the biggest weaknesses of storm modeling: precipitation.

“We can run storm simulations and computer models, but all of them make big assumptions on how to handle precipitation, which can be problematic.” When computers make guesses about temperature, they’re also making conjectures about the distribution of water droplet sizes. “We know there’s a finite amount of water that enters the storm. But what that water ends up as — whether it be big hailstones or tiny droplets of mist — is extremely important in terms of the physics.” Better understanding subtle temperature variations could make much progress in this area.

“These are the sort of things we need to know to better understand tornadogenesis,” Markowski emphasized. “It’s fragile. It’s nonlinear. And . . . this will hopefully help.”