Drones Target Weather Blind Spot for Military and Trading Advantage

(Bloomberg) — In late May 2024, Steve Piltz and his weather forecasting team in Tulsa, Oklahoma, faced a dilemma: they didn’t have enough data to make the call on whether some unremarkable thunderstorms could morph into a violent tornado.

A professor from Oklahoma State University plugged that gap, sending a drone up 1,600 feet (488 meters) into the night sky. The moisture and wind readings he obtained were enough for Piltz’s National Weather Service team to upgrade its public warning. Hours later a deadly EF3 tornado — the third-strongest category — touched down in Rogers County, unleashing a 24-mile (39 kilometers) trail of destruction.

The drone transmitted crucial intelligence from the so-called boundary layer — the few kilometers of the atmosphere that’s closest to the ground. It’s where much of the weather that affects humans — fog, low-level clouds, wind and storms — form, but it’s also a poorly observed region, with traditional ground stations, weather balloons and satellites leaving significant blind spots in coverage.

Now Swiss firm Meteomatics is at the forefront efforts to penetrate that layer using automated drones. Its Meteodrones make repeated vertical flights, feeding temperature, humidity, pressure and wind readings into forecast models. There are significant regulatory and operational challenges to overcome, but boundary layer drones are set to automate a crucial element of weather forecasting that hasn’t changed in decades.

The data collected provides critical information to military planners and air traffic controllers on storms, snow and ice. But the granular forecasts could also give power traders an edge in a market that’s increasingly sensitive to the output from wind and solar farms.

Traders — eager to eliminate costly forecast errors and find profitable opportunities — are among the firm’s biggest customers, according to Brad Guay, a meteorologist with Meteomatics. Something as basic as a temperature inversion that delays the sun burning off cloud cover can offer a crucial trading edge in determining the supply and demand of renewable energy.

“You see these quite small weather events that can make all the difference,” said Guay. “A single blown forecast can cost the traders millions of dollars in a single day.”

The appetite for finer-grained weather data has been stoked by a new generation of AI forecasting models trained on historical observations. But the AI models are only as sharp as the data that feed them, and satellites and other systems designed to make large-scale weather observations aren’t good at recording small-scale atmospheric events, according to Tyler Bell, a meteorology professor at the University of Oklahoma, who operates a separate weather drone platform called CopterSonde.

“We call that terra incognita,” or unknown land, said Bell, describing the boundary layer. Drones, he adds, are the ideal technology to fill the gap as they can monitor wind, temperature, humidity and pressure. “They’re cheap, they’re relatively robust. They have everything you want in one package.”

Mercuria Energy Group Ltd. is exploring whether drone data collected from the boundary layer could boost the accuracy of weather forecasts it uses to predict supply and demand in key US power markets, including the Electric Reliability Council of Texas and the southern hub of the Southwest Power Pool, said Jason Lynn, a meteorologist at the firm.

“The notion of more boundary layer information improving a short-fuse forecast is an easy one to accept, theoretically. But it turns out nobody has the damn datasets to prove it,” Lynn said.

Better and more frequent direct measurements there could help drive higher resolution forecasts in the 48- to 72-hour range, which are vital for predicting volatile atmospheric changes like cloud cover and precipitation that have downstream effects on grid load, he said. Mercuria started discussing the possibility of tapping data from Meteomatics two years ago, and recently approached Bell about using measurements recorded with his CopterSondes. One idea is a network of drones that collect boundary layer data upstream of wind farms, Lynn said.

“Wind is the target parameter that jumps out the most,” he said, especially during winter months in coming years as power demand soaks up excess electricity generation. “All of a sudden, SPP South and Ercot become really, really interesting given the wind generation dependency of pricing in those two markets.”

Military Strategy

Penetrating the boundary layer also has potential strategic importance for the military as geopolitical tensions increase.

The biggest field test of Meteomatics’s weather drones is being undertaken by the Norwegian military, whose territory has become a staging ground for NATO’s northern flank since Finland and Sweden joined the alliance. Over the winter, 14 nations assembled 30,000 soldiers to test whether they could fight effectively as a single force in harsh Arctic conditions.

When a snowstorm moved in faster and more ferociously than the Norwegian Meteorological Institute predicted during one exercise in February, visibility was reduced to almost zero at the takeoff site for military surveillance drones. But a weather visualization tool fed with Meteodrone data from the boundary layer allowed the army to adapt.

“The takeoff area got hit real bad with thick snow and low cloud coverage,” said Kenneth, a mission support specialist with the Norwegian army’s drone squadron. “We were able to relocate them to a site where there was actually better weather so they can take back control of the drone and be able to land it safely.”

Bloomberg agreed for this story to withhold the last names of personnel working in Norwegian military intelligence.

The army is now testing whether adding more boundary layer data could make weather forecasts even more accurate. Early testing is promising, especially with predicting icing conditions that are a major risk in military drone flights, said Nils Håheim-Saers, a senior engineer with NORCE, an independent research institute overseeing the three-year project.

“There’s no other sensor which can detect icing and then feed that observation back into the models,” he said.

Demonstration Flight

Beyond Norway, agencies and researchers in the US, Germany, France, Switzerland and Peru are also testing Meteomatics equipment.

The company’s six-rotor weather drones are custom-built to fly in extreme conditions, including 90 kilometer-per-hour winds, and have built-in de-icing features that allow them to fly at -45C (-49F).

The drones are designed to launch from climate-controlled ground stations that house them between flights and serve as a communications hub. Before launch, the ground station’s clamshell doors open, allowing the drone to ascend from a launchpad. When the drone returns, an automated system re-centers the drone on the launchpad and connects it for recharging before the station’s doors whir shut.

The US’s National Oceanic and Atmospheric Administration came together with Meteomatics and Oklahoma State University researchers for a drone demonstration flight this month. NOAA is bidding to plug the data gap on the boundary layer, while also catching up with European global weather models that currently set a benchmark.

The agency has yet to decide how Meteomatics’s drones will fit in, according to Taylor Jordan, NOAA’s assistant secretary of commerce for environmental observation and prediction.

“I see all of these new technologies as supplementing and augmenting our capabilities,” he said.

To be sure, there are unresolved challenges.

In Germany, government forecaster DWD is bumping up against regulatory limits in its testing of Meteomatics drones. European Union rules still require a human to manually authorize and oversee each drone flight before launch, while the proximity of Berlin airport to the testing site prevents full exploration of the boundary layer, according to DWD’s Ben Skodda.

“You still need somebody to check that everything is fine,” Skodda said. “You cannot just schedule it to fly every half an hour.”

NOAA faces similar restrictions, with the agency still needing to work out Federal Aviation Administration allowances and line-of-sight rules with regulators, according to Jordan.

DWD has also so far been frustrated in its efforts to test Meteomatics’s sensors itself, partly because they are integrated into drones that are too big to fly in the weather service’s wind tunnel. Skodda said DWD has also had trouble getting Meteomatics to slow the climbing rate of the drone, which is too fast for optimal testing against data collected by weather balloons that ascend at a slower pace.

Guay said Meteomatics is working with DWD and other agencies to resolve problems with directly testing the drone’s sensors, but said the faster climb rate is a key part of the drone’s design, and that its sensors and data processing are tuned to its flight speed.

Despite the challenges, Guay said advances in drone battery life, cost and navigation are putting fully automated weather observation within reach.

“We’re kind of at the cusp of rethinking the global observing network,” he said. “In the past, we collected measurements twice a day because that’s essentially how we’ve done it since World War II, but it doesn’t always reflect the meteorological reality.”

–With assistance from Peter Champelli.

©2026 Bloomberg L.P.