Inside the Room Where They Control the Weather Satellites

The headquarters of the European Meteorological Satellite Agency in Darmstadt, Germany (“city of science”!), was easy to spot: It is shaped like one of their early weather satellites, with a central cylinder and protruding wings.

In the gardens outside, large-scale models of their space-borne fleet are lined up among the shrubbery like cocktail tables at a wedding. They are ungainly looking things. Unlike airplanes, whose graceful lines and smooth skins help them slip through the atmosphere, they have delicate protuberances and pocked exteriors. One looks like an engine dipped in gold, another a washing machine with its casing blown off.

It was good to see them there nonetheless. The thing about weather satellites is that they work out of sight. We see them before launch, half under construction, blown out under fluorescent lights and pored over by technicians in bunny suits. Or we see them in artists’ renderings, those sci-fi looking images of spacecraft zooming along in orbit.

But I went to Darmstadt to see them in a new way, in the moment when they come closest to Earth.

“You know, people are always criticizing the weather predictions,” said Yves Buhler, EUMETSAT’s director of technical and scientific support, when I met him in his sunny corner office. A French rocket scientist, he was dressed like it: crisp white shirt, spread collar, breast pocket full of fine-tipped pens. “But globally, it has become much more accurate. And it has become much more accurate, also, in the medium range—so a week, two weeks. Why is that? Because the satellite observations are providing a uniform coverage of the earth. There’s no black hole of an area.”

Today’s weather forecasts are better than ever before, thanks to continual advances in observing the atmosphere and using those observations for fine-grained computer simulations—“weather models”—capable of confidently predicting future skies, 5, 6, 7, even 8 days in advance.

For today’s weather forecasts, the global view is everything. The outlooks that show up on your smartphone apps, in the television studio screens of on-air meteorologists, or the dispatch centers of airlines, all originate from supercomputer weather models—which are voracious consumers of data. And the best, most global data comes from satellites. But not all weather satellites are equal.

Two categories of weather satellites are flying around Earth today: geostationary orbiters and polar orbiters. The geostationary, or GEOs, orbit in the same direction as Earth’s rotation, making them appear motionless in the sky. They provide constantly updated information about a single area of the atmosphere, and provide the pretty pictures we associate with weather satellites.

The polar, or low Earth orbiters, known as LEOs, fly low and fast. They circle the planet from north to south and south to north, overflying a different geography with each orbit and cutting a pattern around the globe like an orange peeled with a knife. They specialize in quantitative data, sucking up numerical readings of things like temperature and humidity, and feeding them, by the millions, to the supercomputer weather models. When it comes to meaningful impacts on forecasting, especially more than a couple of days in the future, they are the champs.

But numbers are hard to see, so it’s the geostationary satellites, the GEOs, and the dramatic images they produce that tend to suck up all the air.

Similarly, not every nation’s weather satellites are the same. This decade, the United States’ geostationary satellite program—GOES—is in the midst of an $11 billion makeover, run by Lockheed Martin. That money pays for the life of four satellites, the first two of which launched in 2016 and 2018, but the number is still shocking, especially when placed alongside the entire annual budget of the National Weather Service, which hovers around a billion dollars annually.

Put more starkly, American weather satellites cost more to fly than the entire forecasting system they support. That expense can be seen as a testament to the importance of satellites to today’s weather forecasts, but it is also a clue to the system’s bureaucratic complexity.

New American weather satellites have been prone to delays, mishaps, and congressional funding cuts, leading to frequent handwringing over a “satellite gap,” in which old satellites fail before new ones are ready to take their place. In 2018, the latest GOES (GOES-17) had trouble keeping one of its primary instruments cool in the sun, rendering it useless at certain times of the day and year. The problem was 97 percent fixed with software and operational changes and the cooling system is being redesigned, but the problems are hard to ignore.

Complex systems have complex problems, but it doesn’t have to be this way. While the US system is mired in bureaucratic complexity in both development and operation, the European Meteorological Satellite Agency, EUMETSAT, keeps its structure simple. It is an independent organization funded and overseen by the meteorological services of 30 nations. Its 450 employees are housed in the single campus in Darmstadt and run under a single leadership. They have ten working weather satellites in orbit and have plans to launch a dozen more. Crucially, the data their polar orbiters produce is as essential (and in some cases more) to the global weather models than their American counterparts.

For a journalist eager to see up close how weather satellites work, and looking for a straightforward account of their operations, EUMETSAT is a dream.

Midway through our conversation in his office, Buhler came up short and studied the big watch on his wrist. He spun around to the phone on his desk. “Do you know when the pass is? Yes. That’s perfect. That’s perfect.” Buhler led us through the satellite-shaped building, clicking through electronic locks and shuffling down natural-lit stairwells, saying passing hellos to scientists and engineers in French, English, German, and Italian.

Behind a final set of heavy double doors the control room was a wide, tall space, set up like a Hollywood mission control with task chairs, dozens of screens and large ticking countdown clocks mounted high up on the wall. Technicians kept a close eye on EUMETSAT’s LEOs and the GEOs from adjacent control rooms. Each room had its own personality and rhythm, like the satellites being watched. The technicians in the GEO control room kept a steady vigil; if everything goes well, not much happens. The LEOs are livelier, and their life is more syncopated.

Every 30 minutes one of their LEOs makes a “pass”: the period in each orbit when the satellite flies over the North Pole, allowing it to be in radio communication with its ground station. When Buhler and I walked in, Nico Feldmann, a young ponytailed operations engineer, jumped to his feet. “Twenty-three minutes; Metop-B; over Svalbard!” he barked. It took me a moment to realize that he was joking, playing Spock to Buhler’s Kirk, pretending we were on the bridge of the Starship Enterprise. But then I realized that he was only half-joking. We were really there to meet a spaceship.

The facility in Darmstadt controls its polar orbiters via a fiberoptic connection that runs across Europe and under the Barents Sea to Svalbard, the Norwegian island above the Arctic Circle. From there, a radio connection is made from a dish antenna 33 feet in diameter, sheltered by an egg-shaped dome the size of a cottage. The dish that serves EUMETSAT is one of 31 set on a plateau known as Platåberget, next door to the Svalbard Global Seed Vault, which stores seeds from around the world in case of an apocalypse.

While Buhler, Feldmann and I chatted in the LEO control room in Darmstadt, the antenna in Svalbard rotated on its spindles, quick and smooth like a robot arm, until the massive bowl was aimed at precisely the point on the horizon where Metop-B would appear, rising like a speck of dust in a sunbeam.

Feldmann and his colleagues call that moment “AOS,” a spaceflight acronym meaning “acquisition of signal.” Metop-B orbits Earth 14 times a day, flying nearly north to south, and then south to north (at an angle, or inclination, of 98 degrees), aiming its instruments downward through a narrow swath of the atmosphere each time. A polar-orbiting satellite, by definition, overflies the poles every orbit; but, with the earth spinning beneath it, it crosses the Equator at a different longitude each time.

Each orbit around the earth takes 102 minutes, and the satellite is only visible to the ground station on Svalbard for anywhere from 12 to 15 minutes of that. The day I’m here, the shortest pass would come during the Norwegian night, around two or three o’clock in the morning, when the satellite would be headed for daylight on the other side of the globe. The main task of each pass is to download the gigabytes of observational data typically collected while the satellite is flying around the planet.

Technically speaking, this is called a “full dump,” and it is kind of like trying to download a movie over your neighbor’s Wi-Fi while driving by their house. (Except it works.) “You need to get the data down, and you need to get it down fast,” Buhler said. To reduce the delay between the satellite’s observations and the distribution of the data, there is also a “half dump,” when the satellite flies over McMurdo Research Station, in Antarctica.

Each pass brings a mix of drama and routine. It was why I was interested in this control room and not the one next door. The geostationary satellites are just that: stationary. They appear to float above us, sloth-like in their stability, keeping a watchful eye. That is an illusion, of course; they are actually flying through space at more than 6,700 mph, completing an orbit of the earth once each day—in other words, at the same pace as the planet itself. The GEOs are always in touch. But with the LEOs, each pass has some excitement. If anything went amiss while the satellite was out of range—if an instrument malfunctioned or any temperature or voltage parameter went out of limits—an alarm would sound. Feldmann gave his own analysis of the geostationary spacecraft managed by his colleagues next door. “GEO is boring,” he said. Buhler was more diplomatic. “It’s always interesting to watch what’s happened over those 100 minutes when the satellite has been flying, invisible to our site,” he mused.

Metop-B’s approach to Svalbard from the far side of the earth was indicated by a red-numbered LED countdown clock on the wall. “The first activity associated with a pass is usually 12 minutes before, when we establish the connection to the ground station,” Feldmann said. The moment approached. We waited. A machine chirped. “There it is,” Feldmann said. “Now we have 12 minutes to send commands to the spacecraft.”

“And to get the data down,” Buhler added, pointing his finger up. We all watched as a column of boxes on the monitor turned green. “And the telemetry looks ... nominal,” Buhler said, relieved, using the space lingo for “normal.” “Telemetry” refers to the basic health of the satellite and its systems, things like temperatures and voltages. For each pass of the satellite, the two key values Feldmann monitored were the duration of “TM” and “TC,” standing for telemetry, meaning the health data received from the satellite; and telecommand, indicating the ability to send commands back. Those transfers happen over relatively low S-band frequencies.

The juicy stuff—which is to say the “science data”—comes over X-band, which is a higher frequency microwave band. Feldmann pointed at another column of green rectangles. “If those are green, it means science data is coming down.” We watched as the numbers ticked up, bit by bit. Feldmann went down the list of acronyms of the different instruments, which measure things like temperature, humidity, cloud cover, and water vapor: ASCAT. GOME. GRAS. IASI. AMSU. Buhler quietly mouthed their names along with him, like a dad at a spelling bee. By this point, we were 1.8 gigs into the dump. If the data being downloaded were a movie, it was one only some experimental filmmaker of the future could imagine: It consisted of 10,000 channels of infrared and radar soundings, shot from space through the clouds. There were five minutes left in the pass.

As Buhler, Feldmann and I talked through the details of Metop-B’s routine, I could appreciate the satellite as a busy working instrument, watching the atmosphere as attentively as any earthbound weather station. It tucks away its images in its solid-state memory banks and then zaps them down to the surface. The data itself isn’t a snapshot, a single click in time, but more like an unspooling filmstrip, as the high-flying robot traverses the atmosphere, pushing through space with its nose pointed down like a bloodhound.

Before I’d had a chance to notice, Metop-B had finished its dump. “Now you see the science data has been downlinked,” Feldmann said in his best David Attenborough voice. “We have just a little bit over one minute left to send it commands.”

But we had nothing to tell the satellite. Everything was green, everything was nominal. In another part of the building, EUMETSAT’s computers had already begun to send the observations out to the world through their data links, paying special attention to the most eager customers: the operators of the weather models, hungry for the latest measurements of the atmosphere. There was a pleasing symmetry to it: The satellite sucked up its observations of the whole earth, and EUMETSAT sent them back out to the whole earth.

When the first weather satellite, TIROS 1, launched in the spring of 1960, President Dwight Eisenhower made a deceptively simple statement: “The earth doesn’t look so big when you see that curvature.” What seemed to surprise him, and really everyone, was the extent to which this new view of the whole planet belonged to the planet. Polar orbiters were the game-changing observatories of today’s forecasts; they were too small to see with the naked eye, but I now had a new vision of how they circled above.

For its part, Metop-B was back below the Norwegian horizon, a happy robot, flying alone, doing its thing. Beside the clock counting down to the next pass was the spacecraft’s odometer: the number of orbits completed. On that afternoon, Metop-B was in the midst of its 10,754th revolution around the earth, a journey that began with its launch in 2012. In less than an hour it would come back over the horizon. Its rhythm is tied to our lives, to the spin of our planet that defines our hours.

Adapted from The Weather Machine by Andrew Blum. Copyright © 2019 by Andrew Blum. Reprinted courtesy of Ecco, an imprint of HarperCollins Publishers.

When you buy something using the retail links in our stories, we may earn a small affiliate commission. Read more about how this works.

More Great WIRED Stories

• Amazon cloned a neighborhood to test its delivery robots
• The best gear to help you clean up your act (and house)
• Fans are better than tech at organizing information online
• Gritty postcards from the Russian hinterland
• My glorious, boring, almost-disconnected walk in Japan
• 🎧 Things not sounding right? Check out our favorite wireless headphones, soundbars, and bluetooth speakers
• 📩 Want more? Sign up for our daily newsletter and never miss our latest and greatest stories