As a historic upsurge of desert locusts ravages East Africa, scientists are using a sophisticated air pollution model to anticipate where the destructive pests are being blown by the wind — and where they will strike next.

Following extreme rainfall events that created favorable breeding conditions over a span of 18 months, swarms of desert locusts from the Arabian Peninsula began rampaging across East Africa in early 2020, devouring crops and vegetation wherever they landed. By February, the crisis had reached historic proportions, with 10 countries in the Greater Horn of Africa and Yemen experiencing infestations.

By the second half of the year, more than 42 million people in those countries could face “severe acute food insecurity,” according to the U.N. Food and Agriculture Organization (FAO).

With heavy rainfall in March and April bringing ideal conditions for an additional wave of locust breeding, authorities are desperate for any information that can help them prepare for coming onslaughts.

Now, scientists at the National Oceanic and Atmospheric Administration have teamed up with Keith Cressman, the senior locust forecasting officer at the FAO, to develop a web app that can be used to forecast where the wind will blow desert locusts after they take flight. The app is powered using an atmospheric model called HYSPLIT.

Cressman is already using the app to inform at-risk countries about locust threats, in addition to integrating the model’s findings into the FAO’s desert locust updates.

The app is a novel use of HYSPLIT, which scientists typically use to understand how pollution particles — whether soot from a power plant or ash from a volcanic eruption — spread and disperse in the atmosphere. To learn where pollution is headed, the model can be run forward in time using data from NOAA’s Global Forecast System and other weather forecasting models.

To determine where pollution came from, the model can be run backward using data from a weather reconstruction technique called reanalysis.

In principle, HYSPLIT can be used to track anything that gets transported through the air, including locusts, which are “passive fliers,” according to Cressman. The insects take off into the wind and are then carried by it, traveling up to 93 miles a day.

Several years ago, Cressman started using HYSPLIT to predict where locust swarms were moving based on field observations. But as East Africa’s locust crisis intensified last winter, Cressman realized he could use the model far more effectively if it was tweaked to account for certain idiosyncratic aspects of locust behavior.

Locusts “don’t fly 24 hours a day like a particle,” Cressman said. “They take off in the morning after a certain time and land just before sunset at a certain time. And then they rest and they go again.”

In February, Cressman reached out to NOAA’s Air Resources Laboratory in College Park, Md., to see whether there was a way to easily integrate this daily cycle into his model runs. NOAA agreed to help, and after a few weeks of “very intensive, long discussions,” the lab rolled out an initial version of a web app in March that could model dozens of swarms at a time and predict their location at five-minute intervals, up to seven days in advance.

“All he has to do is enter the location of the swarm on a certain day,” said Mark Cohen of the Air Resources Laboratory, who helped develop the locust forecasting app, “and we’ve created a system that calculates the sunrise and sunset anywhere on Earth, any day of the year” and models swarm movements accordingly.

The app allows Cressman to predict where swarms will land if they’re flying at different altitudes, where they’ll be subject to different wind speeds and sometimes different wind directions. Cressman can also use the model to work out where a recently spotted swarm probably originated.

“So often we will get out-of-the-blue reports of swarms landing on the coast of the Red Sea,” he said. “The concern of the recipient country is how many more swarms are we going to get and where are they coming from.”

With Cressman’s feedback, NOAA is continuing to improve the app. The Air Resources Laboratory would like to add a feature to account for the rare instances when locust swarms take nonstop, multiday journeys across oceans — for instance, leapfrogging across the Indian Ocean from Somalia to Pakistan.

Cohen says his team would also like the model to account for other linkages between the weather and locust biology, like the fact the insects don’t fly when it’s raining.

While Cressman is feeding the model field observations, in principle it could also integrate weather radar data, which can provide additional information on the size of a swarm and the altitude at which it’s flying.

Ryan Neely III, an atmospheric scientist at the University of Leeds who isn’t involved with the FAO’s locust forecasting efforts, said it would be “amazing” to put radar snapshots of locust swarms into a model like this, while adding radar data can be difficult to access in East Africa.

“If we can have the radar data going and say we see the swarm at this height, then we can put that height into the model and make much better predictions,” said Neely, who is helping spearhead a U.K.-based effort to produce near real-time maps of flying insects using weather radar. Cohen agreed radar or even satellite data would provide additional information that could improve the model’s predictive power.

The locust tracking tool’s rollout comes at a critical time. By the end of June, a new generation of locusts will have hatched and matured into “hungry teenagers” in northern Kenya and southern Ethiopia, Cressman said.

That generation will start taking flight in search of food around the same time many East African farmers are harvesting their spring crops. Being able to anticipate where a swarm will strike next is key to helping local authorities conduct pesticide treatments that prevent crop losses.

“I believe it will help tremendously in [improving] the precision of the early warnings,” Cressman said.