Small clouds drift below the resolution of current instruments, yet matter more to climate models than anyone fully understood
In the long effort to understand Earth's atmosphere, a four-kilogram satellite built by Israeli and German researchers stands ready to attempt something no instrument has done from orbit: map the interior of small clouds in three dimensions. Launching from California in June, CloudCT carries seven years of collaborative science and a method borrowed from medical imaging — CT scanning adapted for the sky, guided by artificial intelligence. The clouds it seeks are easy to overlook, yet they quietly shape the planet's energy balance and remain one of the deepest sources of uncertainty in climate prediction. What begins as a single small machine may, if it succeeds, grow into a constellation that changes how humanity reads the atmosphere.
- Small clouds — too fine for current satellites to resolve — are quietly distorting climate models, and scientists have long known the gap without having the tools to close it.
- A 4-kilogram satellite, compact enough to hold in two hands, is now the vessel for a solution: AI-driven tomography that reconstructs cloud interiors in three dimensions from multiple orbital angles.
- The engineering obstacles are severe — the satellite must autonomously orient itself toward specific cloud formations and, eventually, fly in precise formation with nine companion satellites without constant ground control.
- CloudCT is tested and waiting in California, weeks away from a launch that will determine whether the concept survives contact with the real environment of orbit.
- Success unlocks a ten-satellite constellation planned for 2027, one that could sharpen long-term climate forecasts by finally giving modelers a clear view of the clouds they have been forced to guess at.
A four-kilogram satellite built by Israeli and German scientists is ready to leave Earth. CloudCT will launch from California in June, carrying seven years of collaborative research and a fundamentally new way of seeing clouds — one small enough to hold in two hands, yet representing a potential shift in how climate science observes the atmosphere.
Traditional remote-sensing satellites miss small clouds entirely, the ones that drift below the resolution of current instruments. Yet these clouds matter more to climate models than was fully appreciated until recently. CloudCT is designed to see them — mapping their internal structure in three dimensions and measuring the microphysics of individual cloud droplets with a precision never before attempted from orbit.
The project brings together Ilan Koren and Yoav Schechner from Israel's Weizmann Institute and Technion, and Klaus Schilling from Germany's Center for Telematics, backed by a prestigious European Research Council Synergy grant. Their core innovation borrows from medicine: just as a CT scanner reconstructs a patient's interior from images taken at many angles, the team adapted this logic for clouds, using AI to build volumetric maps of cloud structure from multiple orbital perspectives. The AI also evaluates the reliability of its own measurements — a crucial safeguard.
The technical demands are steep. The satellite must autonomously orient itself toward specific cloud formations and, eventually, coordinate with nine companion satellites in precise formation flight — all without constant ground intervention. If June's launch succeeds and the sensors perform as designed, the path opens for a full ten-satellite constellation in 2027. That network could meaningfully tighten climate forecasts by resolving the small clouds that currently represent one of the largest sources of uncertainty in long-term predictions. The real test is about to begin.
A four-kilogram satellite built by Israeli and German scientists is ready to leave Earth. CloudCT, as it's called, will launch from California in June, carrying with it seven years of collaborative research and a fundamentally new way of seeing clouds.
The satellite is small enough to hold in two hands, yet it represents something larger: a shift in how climate scientists might observe the atmosphere. Traditional remote-sensing satellites miss the small clouds entirely—the ones that drift below the resolution of current instruments, the ones that matter more to climate models than anyone fully understood until recently. CloudCT is designed to see them, to map their internal structure in three dimensions, to measure the microphysics of individual cloud droplets with a precision that has never been attempted from orbit.
The project emerged from seven years of intensive work by researchers at Israel's Weizmann Institute of Science and the Technion, partnering with Germany's Center for Telematics. Ilan Koren, a leading atmospheric scientist at Weizmann, leads the effort on the Israeli side. Yoav Schechner at the Technion and Klaus Schilling at the Center for Telematics round out the core team. The European Research Council awarded them a prestigious Synergy grant—the kind of funding that backs moonshot science—and the work has already produced publications in top-tier journals.
The innovation at the heart of CloudCT borrows its logic from medical imaging. Computed tomography—the CT scan that doctors use to see inside a patient's body—works by taking many images from different angles and reconstructing a three-dimensional picture. The team adapted this principle for clouds. By positioning a small satellite to observe cloud fields from multiple angles and using artificial intelligence to process the data, they can build a volumetric map of cloud structure that reveals details invisible to conventional satellites. The AI component also allows the scientists to assess how reliable their own measurements are, a crucial safeguard in research.
The technical challenges are formidable. A four-kilogram satellite must autonomously orient itself toward specific cloud formations. It must coordinate with nine other satellites—the constellation planned for next year—to fly in formation through space. Schilling, who specializes in small-satellite engineering, describes this as one of the central difficulties: getting tiny guidance and control systems to execute precise maneuvers and maintain formation flying without constant ground intervention. This autonomous formation flying is the key to making the whole system work.
If the June launch succeeds, if the satellite's sensors function as designed and the AI algorithms perform in the harsh environment of orbit, the path opens for deployment of the full ten-satellite constellation in 2027. That network would transform the resolution and detail available to climate modelers. Small clouds contribute significantly to Earth's energy balance—they reflect sunlight, they trap heat, they influence precipitation patterns—yet they remain a major source of uncertainty in long-term climate forecasts. Better observations of them could tighten those forecasts, reduce the error bars, make predictions more reliable.
Right now, CloudCT is on the ground in California, tested and ready. In a few weeks, it will ride a rocket into orbit and begin its work. The real test—whether a four-kilogram machine can see what larger satellites cannot—is about to begin.
Notable Quotes
The mission addresses significant sources of uncertainty that currently limit long-term climate models and forecasts— Ilan Koren, Weizmann Institute
Precise aiming and coordination between 10 tiny satellites flying in formation in space are significant challenges for such small guidance and control systems— Klaus Schilling, Center for Telematics
The Hearth Conversation Another angle on the story
Why does a satellite need to be so small? Wouldn't a bigger instrument see more?
Size is actually the constraint that forces innovation here. A small satellite can be launched cheaply, and you can afford to build ten of them instead of one. But more importantly, the small size forces you to be clever about what you measure and how you process it. That's where the AI comes in.
So the AI is doing the heavy lifting that a larger instrument would do with brute force?
Exactly. The satellite collects images from different angles as it moves, and the AI reconstructs the three-dimensional structure from those images. It's like how a CT scanner works in a hospital—many two-dimensional slices become one three-dimensional picture.
And these small clouds that current satellites miss—why do they matter so much for climate?
They're everywhere, and they're invisible to the instruments we have now. But they reflect sunlight and trap heat. If you're trying to predict how Earth's climate will change, you need to know what those clouds are doing. Right now, that's a huge source of uncertainty.
What happens if the June launch fails?
Then you learn what went wrong and you try again. But the team has been working on this for seven years. They've published the science. The engineering is solid. The real question is whether the technology works as well in orbit as it does in the lab.
And if it does work?
Then next year they launch nine more satellites, and suddenly you have a network that can see clouds the way doctors see inside a patient's body. That changes what climate scientists can do.