Why Derechos Are So Devilishly Difficult to Predict

Why Derechos Are So Devilishly Difficult to Predict

At 8:30 am on Monday, Mark Licht was sitting in his home outside of Ames, Iowa, on a conference call with other agronomists and meteorologists from around the state. Iowa had been having a dry spell, the western half of the state stricken with severe drought. What farmers needed was a big storm, thought Licht, a cropping systems specialist at Iowa State University. The meteorologists on the call told him that one was just getting started in South Dakota and Nebraska. But, they said, it didn’t look like it would have the energy to make it into Iowa. Everyone crossed their fingers and hung up.

Around 10:15 am, Licht got an email from the group; the storm looked like it might be sticking together after all.

Less than an hour later, he heard storm warning sirens blaring from the closest town. He went outside. It was sunny, barely a cloud in the sky. The air was still and the humidity suffocating. “That’s weird,” he thought. But when he checked radar he saw a huge mass barreling in his direction at about 60 mph.

He got his family into the basement, and 10 minutes later the storm was on top of them. Rain so heavy you couldn’t see more than a few feet ahead. Winds so fierce they could shear a tree in half. When Licht and his family emerged about 45 minutes later, the steel shed where their cars were parked had completely collapsed. “We were smack in the middle of one of the more devastated storm paths,” says Licht. “It’s going to be a long process to deal with the damage, but we’re lucky it wasn’t worse.”

Iowa knows summer storms. But the one that tore across the Midwest on Monday, traveling 770 miles in 14 hours, leveling 10 million acres of crops, twisting grain silos, and knocking out power for hundreds of thousands of people for days, was a rare type of storm known as a derecho.

The term means “straight ahead” in Spanish, and was coined in the late 1800s by a Danish physics professor at the University of Iowa, who used it to describe the “straight blow of the prairies,” in contrast to the circular winds associated with tornadoes. Today, for a derecho to meet the National Oceanic and Atmospheric Association’s definition, it must travel at least 240 miles and move at speeds of at least 58 mph. This week’s derecho hit top speeds around 112 mph in Cedar Rapids, Iowa, about two hours due east of Licht’s home.

“Derechos are these long-lived, fast-moving walls of super-thunderstorms,” says Paul Huttner, a meteorologist who watches the weather for Minnesota Public Radio. They’re regular but not common events, occurring in the Midwest one or two times a year. Derechos come in two varieties: linear and progressive. Monday’s storm was a progressive derecho, which moves faster, is more compact, and packs more of a punch than its more spread-out sibling. And this one, says Huttner, “was a real doozy.”

Derechos are often referred to as “inland hurricanes,” because the experience on the ground—whipping, destructive winds; sideways pelting rain—is similar to being inside a Category 1 hurricane. But they’re completely distinct weather phenomena. Derechos are fueled by different factors and behave more like a stampede of wildebeests than a bloat of hippopotamuses. And unlike the path and severity of hurricanes, which scientists have gotten good at forecasting days or even weeks in advance, predicting where and when a derecho will form remains one of the most challenging tasks facing meteorologists.

The difficulty comes down to two things: patchy data and the complexity of the storm itself. Derechos can arise at random, almost out of nowhere, says Victor Gensini, a meteorology professor at Northern Illinois University. On many days during the year, the ingredients are all present for a derecho—but they’re missing something to start the storm. “In other words,” he says, “you need a trigger.”

It could be a cool pool of air left by a previous storm, high terrain, or a sudden sea breeze. In the summer months, these triggers can be especially difficult for meteorologists to detect. That’s in part because the fast-flowing global air current known as the jet stream slows down during that time, as the temperature contrast between the poles and the equator shrinks. The result is that instead of distinct and regular waves of warm and cold fronts being forced eastward across the US, the atmosphere is more stagnant. And small, random disturbances can more easily kick up warm air from the surface. If there’s enough water vapor in the air to buoy the updraft, it will keep rising and accelerating, triggering a thunderstorm.

Sometimes these disturbances come from other storms that have recently passed through—the cold air they leave behind can be the seeds of the next storm. “That’s why the weather changes so rapidly in the summer without warning,” says Gensini. These storms actually modify the environment around them, often determining the next day’s weather.

And sometimes, it’s the lack of a storm that determines the next day’s weather. That’s what happened in the run-up to this week’s derecho.

On Sunday night, Patrick Marsh, the science support chief at the National Weather Service Storm Prediction Center in Norman, Oklahoma, was working a forecasting shift alongside four other colleagues. (In pre-pandemic times they would have been in a mission control room together; now they sit in separate rooms rigged up with video monitors and talk to each other over Google Hangouts.) Their models were telling them that a decent-sized thunderstorm was going to develop across the Dakotas and move into Minnesota. They expected it to take the warm moisture from the surface and swap it with cool air from higher up in the atmosphere. This flip-flopping takes some of the water vapor out of the atmosphere, stabilizing it.

“But that’s not what happened,” says Marsh. The storms on Sunday never materialized. So the atmosphere stayed volatile, a huge amount of energy just primed for the next disturbance to tap into. Watching this potential continue to build overnight, the Storm Prediction Center upgraded the risk of severe weather for some parts of Iowa from marginal to enhanced. Early that morning, a new storm system began to form. But it wasn’t until a few hours before the winds in Iowa began gusting to 90 mph that Marsh realized they were about to have a derecho on their hands.

“We see these kinds of environments happen multiple times a year, but something almost always goes wrong that prevents a storm from evolving into a high-end situation,” he says. By mid-morning, his team wasn’t seeing any of those typical failure modes. “That’s when we knew this was going to be a problem.”

Though the derecho eventually grew large enough for satellites to chart its destructive path, the processes that drove its initial evolution were happening at a scale too small for the National Weather Service’s monitoring network to pick up. The NWS has more than 1,000 surface locations from which it collects information. Data from high up in the atmosphere is far patchier. The only way to get that intel is with weather balloons, about 200 of which are launched daily around the US—half at 8:00 am Eastern, and the other half 12 hours later.

“All we have is these snapshots two times a day at 100 points for the entire country, which makes it really hard to get good wind data in the upper levels of the atmosphere,” says Andrew Mercer, a meteorologist at Mississippi State who studies derecho formation. “What our models really need is much more frequent sampling, especially in the hours leading up to the event.”

That’s because one of the key ingredients for derecho development is strong vertical wind shear. Imagine a flagpole extending from the ground 10 miles into the air, with flags placed every few hundred feet. Under ideal conditions for derecho formation, those flags at every level would be blowing at different speeds. That’s vertical wind shear. It’s not the only indication that a derecho is coming—the exact atmospheric alchemy that leads to one is still not totally understood—but it’s an important input for weather models.

(Here’s another X factor: Scientists don’t really know how increased moisture impacts derecho development—sometimes water-enhanced downdrafts fuel them, sometimes they short-circuit. Why? Who knows?!)

Forecasters can request additional balloon launches to try to get that wind shear information. The National Weather Service did that on Monday morning, says Marsh. And the picture it returned—an incredibly unstable atmosphere conducive for strong winds making it down to the surface—told Marsh a derecho was imminent. “It ruled out another failure mode and told us this thing is not going to go away,” he says.

Improving derecho prediction then hinges on more frequent data collection from the upper reaches of the atmosphere. But it’s not exactly practical to send massive fleets of instrument-laden balloons skywards every hour. (And would be a huge headache for the commercial flight industry.) So, the National Oceanic and Atmospheric Administration has recently begun experimenting with autonomous drones to make these critical measurements.

Scientists like Mercer think the problem can’t be solved just by putting more stuff in the sky. His lab has been collaborating with the National Weather Service to build an entirely different kind of prediction model. Rather than the decades-old forecasting workhorse models based on math and statistics, these employ machine learning, running a large number of high-resolution simulations, and teaching the system to identify and predict different characteristics of a storm. Will it be primarily wind-dominant or hail-dominant? Will it be tornadic or non-tornadic?

The research is still in the early stages, and likely years away from being operational. But the idea is to let the machines determine the processes that influence whether a storm dies out or goes derecho, and to pinpoint at what point in time those processes commit a weather system to form one kind of storm versus another That could theoretically help improve some of the rather crude representations of these processes currently used by numerical models.

Mercer recognizes that these types of models will also be hampered by scarce data from the skies. “There are always going to be limitations if we don’t have the upper wind data,” he says. But even with better methods of forecasting, there will never be a lack of information for scientists to measure. There’s always another storm coming.


More Great WIRED Stories

Leave a Reply

%d bloggers like this: