How Does NASA's NISAR Satellite “See” Through Clouds and Polar Darkness to Track Antarctica's Ice?

NASA and ISRO's NISAR satellite builds a sharp radar picture by combining thousands of pulses fired from a moving satellite into the equivalent of one antenna 19 kilometers (12 miles) long. Its radar waves are more than ten thousand times bigger than the water droplets inside a cloud, so they pass through cloud cover and total darkness almost undisturbed. In August 2025 this let NISAR scan a mountain called Nunatak Zaterjavshijsja, buried under flowing Antarctic ice in a shape scientists nicknamed “the hummingbird,” revealing crevasses and ice structure no ordinary camera could see. NASA and ISRO released the picture in July 2026.

In July 2026, NASA and India's space agency ISRO published a radar portrait of Antarctica that nobody had gone looking for. It came from a joint satellite called NISAR, and it stopped scientists mid-scroll. It showed a shape buried under a river of flowing ice, in a place with no roads and no research station. What turned up was a hummingbird. Its beak trails into a mountain ridge. Its feathers are cracks in the ice. Its eye is a lone peak, a mountain called Nunatak Zaterjavshijsja. The peak pokes through a glacier dragging it out to sea.

NASA did not paint this picture. A satellite hundreds of kilometers overhead drew it with radio waves alone, using a dish JPL describes as about as wide as a school bus is long.

That is the strange part. A camera needs light bouncing off its subject, plus a clear line of sight. Antarctica, for much of the year, offers neither. NISAR stands for the NASA-ISRO Synthetic Aperture Radar. It returns images sharp enough to trace individual crevasses, in darkness and under cloud, and it reads structure meters into the ice itself.

A school-bus-sized dish sounds big, until you learn what this job actually demands. To resolve detail this fine from orbit, JPL says NISAR's L-band radar would need a traditional antenna 19 kilometers (12 miles) long. So how does a 12-meter dish do the work of one more than 1,500 times longer? How does it work through darkness and thick cloud? How does it see inside solid ice? It is not one trick. It is three, stacked on top of each other, and each one starts with a problem.

What Is Synthetic Aperture Radar, And Why Do Satellites Need A Trick At All?

Start with how any radar works. A satellite fires a burst of microwaves at the ground, called a pulse. A fraction of a second later, some of that energy bounces back. Time the round trip, and you know how far away the ground is. ScienceABC has covered the fundamentals of pulse-echo radar in more detail. NISAR starts from that same basic idea.

The trouble is turning an echo into a picture. Any antenna, however large, spreads its beam as it travels. A flashlight lights a big, dim patch of wall instead of a tight, bright spot, and a radar dish does the same thing. The wider the dish, the tighter the beam, which is why radar engineers want the biggest antenna they can fly.

Here is how badly they lose that fight. NASA works the numbers for a satellite 1,000 kilometers (620 miles) up carrying a 10-meter antenna, which is close to the largest anyone has ever flown: the beam still spreads until its footprint on the ground is 10 kilometers (6 miles) across. (NISAR itself flies lower, at 747 kilometers, with a 12-meter dish.) Everything inside that footprint, mountain, crevasse, and flat ice alike, blurs into one smear.

So the problem is not that engineers were stingy with the antenna. They already built about as big a one as a rocket can carry, and it is still more than 1,500 times too small. Fixing that the obvious way means flying a dish kilometers across, which no rocket built has ever come close to launching. So radar engineers reach for a trick instead. The rest of this story is one satellite's workaround for that fact.

How Does A Moving Satellite “Grow” A Giant Antenna Out Of A Small One?

Here is the workaround, and it is the trick the mission is named for: synthetic aperture radar. NISAR does not fly a giant antenna. It fakes one.

As the satellite races along its orbit, it fires pulses at the same patch of ground over and over, roughly 1,650 times a second in its standard L-band mode. Each pulse leaves from a slightly different point along the flight path, so any one spot on the ice gets sampled thousands of times as the beam sweeps past. Each echo comes back carrying a fingerprint: what NASA calls phase, the alignment of the returning wavelength compared to the original. Because the satellite's distance to that spot changes in a precise, predictable way as it flies over, the phase changes with it, and that record of change is what makes the trick possible.

Picture stitching a hundred quick phone photos of a passing parade into one seamless panorama. Every "photo" here is a radar echo instead. The stitching has to line up each wave to a fraction of its own length, over a flight path kilometers long. The combined data then behaves like one giant antenna recorded it all at once. NISAR's own real antenna is a 12-meter (39-foot) reflector, the largest antenna reflector NASA has ever deployed in space, and SAR processing makes it act like one 19 kilometers long.

The payoff has a lovely twist buried in it. Once you are synthesizing the aperture, resolution stops depending on how high you fly. NASA's worked example turns a 10-meter dish into a virtual 10-kilometer one, and the sharpest detail it can then resolve is about half the real antenna's diameter, or 5 meters. Build a wider dish and the picture gets blurrier, not sharper. That is the opposite of how every camera you have ever owned behaves.

One honest caveat, because it is the part most explanations skip. This trick only sharpens the image in one direction: along-track, the way the satellite is flying. Sideways, in the direction the radar actually faces, antenna size is not the limit at all. There the sharpness comes from how short and how broadband each pulse is, because two objects can only be told apart if they sit more than half a pulse-width apart on the ground. That is why NISAR's numbers are lopsided: about 7 meters along track, and 2 to 8 meters across it.

It is a neat piece of physics with an almost comic footnote. Nobody in the signal-processing lab built a bigger antenna. They just did the math.

Why Does A Radar Wave Slip Past Clouds And Polar Darkness?

Solving the resolution problem still leaves two more. Antarctica spends months in total darkness. It is also, often, buried under cloud. A normal camera, even from orbit, is helpless against either one.

Radar does not care, and the reason comes down to size, not brightness. Synthetic aperture radar is an active sensor. Instead of waiting for reflected sunlight the way a camera does, it brings its own light, in the form of microwaves. That means it works as well at midnight as at noon. Clouds are a separate problem, and the fix is the size of the wave itself.

A wave barely notices anything much smaller than itself. Visible light has a wavelength of about half a micrometer, while a cloud droplet is around 20 micrometers across, some forty times wider. Light stands no chance of slipping past something that much bigger than itself: every droplet scatters it, and scatters all colors about equally, which is why a cloud looks like a solid white wall to your eyes and to any camera. NISAR's L-band radar wave is 24 centimeters (about 9.4 inches) long, more than ten thousand times wider than that droplet, so the droplet barely registers at all. A wave that long simply is not scattered by something that small. An ocean swell rolls right past a small buoy, barely nudging it. A short, choppy wavelet slaps right into that same buoy and bounces off. NISAR's radar wave is the swell. A cloud droplet never had a chance.

Put the two effects together: all-weather, and all-dark. NASA describes SAR as able to observe day or night, in any weather. That is what let NISAR return a usable image over Antarctica at all.

What Does NISAR's Antarctic “Hummingbird” Image Actually Show?

With those three pieces in place, the hummingbird image stops looking like magic. A 12-meter antenna acts like a 19-kilometer one. The satellite works through darkness. It ignores clouds. What is left is an engineering payoff.

The scene shows Nunatak Zaterjavshijsja, a mountaintop in East Antarctica. It sticks up through an ice stream flowing northeast toward the ocean, and as the glacier grinds past the obstruction, the stress heavily fractures the surrounding ice with the deep cracks called crevasses. Those cracks are the "feathers."

The colors are the part worth slowing down for, because they are not a paint job. A radio wave wobbles in a particular direction as it travels, a property called polarization, and NISAR listens for which way the wave is wobbling when it comes back. It sends its pulses out wobbling horizontally. Signals that return still wobbling horizontally, JPL says, most likely bounced off something fairly regular, such as smooth ice. Those are drawn in magenta. Signals that come back wobbling vertically have been knocked sideways along the way, either by refracting down into the snow and ice or by ricocheting off irregular faces like the walls of a crevasse. That reshuffling is called volume scattering, and it is drawn in green. Where both come back strongly, the image goes white, a hint that surface and volume scattering are roughly in balance.

That distinction matters, because the intuitive story, smooth things reflect cleanly and so look bright, is exactly backwards for a side-looking radar. NASA's own primer notes that calm water and other smooth surfaces come out black, because a smooth surface bounces the pulse away from the spacecraft rather than back at it. The magenta here is not "clean reflection." It is a wave that survived the round trip with its wobble unchanged.

Seongsu Jeong, the signal analysis engineer at NASA's Jet Propulsion Laboratory who produced the image, put it this way: "because radar can often see through snow and deep into the ice, NISAR can observe fundamentally different properties of Antarctic ice than can be seen in optical imagery. With NISAR we're seeing what's hidden beneath the surface."

That word “often” is doing real work. This is not a photo of ice reflecting sunlight off its top surface, the way an ordinary satellite picture works. Radar waves can push into dry snow and ice before they scatter back, so part of the signal genuinely comes from inside the ice rather than off the top of it. Some of the picture is still surface: a crevasse wall is an open crack you could fall into, not a buried feature. But the mix of the two is what a camera could never register, and it reads closer to an ultrasound of a glacier than a snapshot of one. Antarctica hides other things below its ice, too, and not all of them are frozen water. Physicists dig kilometers down through that same ice for an entirely different reason. They are hunting ghostly particles called neutrinos.

Scientists did not go looking for a bird, either. A crevasse field, a ridge, and a stray shadow lined up into a shape our brains latched onto fast. It is the same reflex that spots faces in clouds and constellations in stars. It became the mission's first great piece of publicity by accident.

Why Does The Same Ice Look Blank In A Normal Satellite Photo?

JPL did not just release the radar image. It released a Landsat 9 optical photo of the same mountaintop, taken on November 2, 2025 by an ordinary camera-style sensor. It is worth being precise about what that comparison does and does not show. The two frames are about three months apart, not one moment caught two ways, and the Landsat image obviously required daylight to exist at all. What they compare is how much detail each kind of instrument pulls out of the same scene.

On that score it is not close. There is no hummingbird in the optical frame. Slight shadows and a faint rippling hint that a mountaintop is there, and some texture in the surrounding area suggests the ice is not perfectly smooth, but that is the whole harvest: a pale smudge on a field of near-featureless white.

That flat frame is not a bad photo. It is what visible light does on Antarctic ice, even under good conditions. Nearly all of it bounces straight back into space, so a camera has almost nothing to grab onto. Antarctic ice has other quirks a camera struggles with, too. Radar, as JPL puts it, captured more of the structure of the surrounding icescape than is visible in the optical image, because microwaves penetrate frozen surfaces and visible light does not.

This is one of the more anticlimactic images NASA has released, sitting right next to one of its most striking. That is the point of putting the two side by side.

How Do Repeat Radar Passes Turn Into A Map Of Ice On The Move?

A single hummingbird image is a snapshot. NISAR's real scientific payoff comes from doing this again and again. It circles the globe on a 12-day repeat track, covering nearly all of the planet's land and ice surfaces twice in each of those 12 days, which works out to a fresh look at a given spot roughly every six days. It is not quite the whole globe: the scan runs from within a few degrees of the South Pole up to 77.5 degrees north. Antarctica sits comfortably inside that, imaged from the same angle each pass.

The radar keeps track of each wave's phase, that fingerprint from earlier. That lets scientists compare two passes over the identical patch of ice. They can read the tiny difference between them. This technique is called interferometric SAR, or InSAR. NASA's own textbook example is a volcano, not a glacier. On Alaska's Okmok volcano, comparing repeat radar passes produced rings of color. In that image, each ring marks 2.8 centimeters (about an inch) of ground moving toward or away from the satellite. How much ground movement a ring stands for is set by the radar's own wavelength, so NISAR's longer L-band waves draw coarser rings than the shorter-wavelength radar that imaged Okmok. The method is identical either way. Apply it to a glacier instead of a volcano, and every ring becomes ice creeping toward the sea, rather than magma pushing up from below.

A research team led by glaciologist Eric Rignot applied this method continent-wide. The team mapped ice motion across Antarctica from 1995 to 2022. The deep interior has barely budged. Glaciers along large stretches of West Antarctica have sped up, some reaching tens to hundreds of kilometers inland. Rignot and Stanley Jacobs showed that Antarctic glaciers melt from underneath. That happens right where the ice lifts off the seafloor and starts to float. The melt rate there is far higher than anyone had assumed. It tracks the temperature of the ocean water reaching that boundary. NASA keeps its own database of these grounding lines, the exact seam where grounded ice starts floating. It is built from decades of this same technique.

None of this proves the hummingbird scene itself is melting away. JPL released that image to show off the instrument, not to announce a new finding. One photogenic release cannot show where Antarctica's ice is speeding toward the sea. Repeat passes, over years, eventually will. That answer is worth more to a coastal city than any single striking picture.

Why Did NASA And ISRO Build This As A Joint Mission?

None of this happened by accident, or on a shoestring. NISAR is a joint NASA-ISRO satellite. It launched from India's Satish Dhawan Space Centre on July 30, 2025. It carried NASA's L-band radar and ISRO's S-band radar on the same spacecraft. That pairing is not just a first for these two agencies. JPL calls NISAR the first free-flying space mission to carry two radar instruments at all, and the first to carry two SAR instruments working at different wavelengths. Have you ever wondered how a satellite powers itself and stays in contact with the ground for years at a time? NISAR is an extreme example. It has to keep two full radar systems fed with power. It also has to keep beaming data home. At the same time, it must hold its orbit to within 500 meters (1,640 feet) across a baseline mission of three years, with any extension beyond that still to be earned.

The most dramatic piece of hardware is that 12-meter reflector. It is a gold-plated wire mesh dish that launched folded to about 2 feet (0.6 meters) across. In orbit, it unfurled like a drum-shaped umbrella, into the largest antenna reflector NASA has ever deployed in space. Getting two space agencies, on opposite sides of the planet, to agree on one shared payload took years of talks, before a single pulse was fired.

The two agencies do not quite agree on when the science began. ISRO declared the science phase open on November 28, 2025, while NASA dates the start of science operations to early January 2026. Either way, data started flowing. The first NISAR picture had already gone public before that, a radar view of Maine's Mount Desert Island captured on August 21, 2025 while the team was still checking the instruments out, followed by sample product releases in January and February 2026. What arrived on July 20, 2026 was the thing scientists actually wait for: a continuous stream of calibrated L-band data, free to download through NASA's Alaska Satellite Facility and ISRO's Bhoonidhi platform. Ice sheets are only one item on its list. NASA's mission goals also include tracking earthquakes, landslides, and shifting ecosystems. It is the same repeat-pass trick, pointed at things too slow for any camera to catch.

So, How Does NISAR See Through Clouds And Darkness To Track Antarctica's Ice?

Put the three pieces back together and the hummingbird stops being a magic trick. Even the biggest dish a rocket can lift would only ever draw a blur from orbit. So NISAR turns its own motion into a stand-in for an antenna 19 kilometers long, timing thousands of pulses a second down to a sliver of a wavelength, then adding them up like a panorama stitched from a hundred quick photos. A camera goes blind in the dark and gets blocked outright by cloud. So NISAR brings its own light, in the form of microwaves whose wavelength is more than ten thousand times bigger than the water droplets that stop a camera cold. And sunlight only ever bounces off the top of the ice, while those same long radar waves push into it, dragging structure a camera would never register up into the picture.

Stack all three, and you get an image that does not need daylight or a clear sky, and that carries information from inside the ice as well as off its surface. No single piece of that puzzle could manage it alone. The bird was never the discovery. It is what one satellite's engineering trick looks like, pointed at the right mountain. NISAR is now flying that trick over the whole continent on a twelve-day cycle. Antarctica will not get to keep many more secrets buried under its ice.