Seeing Through Clouds, Floods and Three Kilometres of Ice
Four satellite stories we ran in the past three weeks share one quiet lesson: almost nothing in them was seen directly. Here is how the tricks work, and where they stop.
In short
What happened. In the past three weeks NASA published satellite studies of Arctic sea ice, valleys under Greenland’s ice sheet, a record flood on the Gandak River and low cloud over the US Pacific Northwest.
What it means. Each relied on a different way of seeing indirectly: microwave energy that passes through cloud, infrared colour that separates water from land, and faint bumps on the ice surface that reveal the rock beneath.
Risks and impact. The methods are strong, but each has limits. Visible-light images stop at the cloud tops, and the sea ice record assumes the small area around the pole is ice-filled.
What can be done. Read the caption: which instrument, which date, and whether the colours are true or false. NASA points readers to its Worldview viewer to look at such images.
What to watch. Next March’s Arctic winter maximum, after the 2025 and 2026 maximums tied for the lowest in the satellite record.
Shown as a summary because of your reading settings.
What happened
On 19 September 2026, at about 10:45 in the morning, NASA’s Terra satellite looked down on Washington and Oregon and saw almost nothing of them. A layer of low marine cloud covered the land as far as the Cascade foothills. Only the higher peaks of the Olympic Mountains poked through, like islands. By the time NASA’s Aqua satellite passed at about 4:15 in the afternoon, much of the cloud had gone.
That picture, which we ran in September, is a good place to start. It shows the problem every Earth-watching satellite faces. Much of what matters is hidden: under cloud, under floodwater, under ice.
Three other NASA stories we carried since then solved that problem in three different ways. Arctic sea ice reached its 2026 minimum on 12 September, covering an estimated 4.6 million square kilometres. That ties 2008, 2010 and 2025 for the 10th-lowest minimum on record. In March, the winter maximum had statistically tied 2025 for the lowest. A NASA-led study mapped 1,943 valleys beneath Greenland’s ice sheet, about a third of them new. And satellite images caught the Gandak River in Nepal and India breaking through its embankments after rain between 24 and 27 September.
What the evidence supports
What independent sources agree on. The Arctic numbers come from NASA and the National Snow and Ice Data Center together. The record is continuous since late 1978 and stitched across several satellites: NASA’s Nimbus-7, US defence weather satellites from 1987, NASA’s Aqua from 2002 to 2011, and since 2025 Japan’s GCOM-W. Each of the past 20 summers, 2007 to 2026, ranks among the 20 lowest minimums. The lowest came in September 2012.
What rests on one source. The Greenland valley map comes from a single paper, by Chartrand and colleagues in Geophysical Research Letters. The valleys were mapped by hand. The aim is to improve BedMachine Greenland, the standard dataset of the land beneath the ice. It is new and has not yet been tested by other teams.
What the satellite did not measure. For the Gandak, the satellite showed where the water went. The human toll, with hundreds of villages affected and thousands of people displaced, came from news reports. The record river height came from a gauge on the ground: 11.94 metres at Devghat on 27 September, the highest since records began there in the 1960s.
What we do not know. There is a “pole hole”, a patch around the North Pole where satellite sensors have historically been unable to collect data. The estimates assume it is full of ice. Nor do the sources say whether the recent September plateau will last, or what lined up Greenland’s straightest valleys.
How the story is being framed
“The Arctic is in free fall.” This reading draws on the long trend. Winter ice is shrinking too. The March 2026 maximum statistically tied 2025 for the lowest on record. The remaining ice is younger and thinner. What this framing leaves out is the recent September plateau. Without that, the next “not a record” year will look like good news.
“The melt has stalled.” This view is also built on NASA’s own data. September extent has been fairly steady for about a decade. NASA scientists link that partly to more cloud cover, which kept sunlight from speeding up the melt. Its blind spot is size: the plateau sits well below the long-term average. Natural cycles such as the Arctic Oscillation cannot explain the decline alone. Scientists project an essentially ice-free Arctic at least once before the middle of this century.
“Greenland is a rigid block of old rock.” That has long been the working assumption, and the new map mostly fits it. Many valleys start in the southern and eastern highlands, where the ice sheet is thought to have formed first. But a set of long, straight valleys in the west-central region all run southwest to northeast. That points to tectonic forces. “That’s a riddle to us,” said Joe MacGregor, a NASA scientist and co-author.
The evidence is not evenly split on the Arctic. The plateau is real. So is the long decline it sits within.
The background
The four stories use three tricks. Each is worth understanding, because they turn up in almost every satellite image in the news.
Trick one: listen instead of look. Earth naturally gives off a faint glow of microwave energy. Sea ice gives off more than open water, so it stands out. Microwaves pass through cloud. That is why the sea ice record keeps going in weather that would blank out a picture like the Pacific Northwest one. The sensors are called passive because they simply measure the energy Earth emits on its own.
Trick two: choose your colours. The Gandak images, taken by the MODIS instrument on Terra and Aqua on 20 and 29 September, are false colour. They mix infrared light with visible light, using bands 7, 2 and 1. In that mix, floodwater shows as blue, plants as bright green and clouds as light blue. The point of adding infrared is to make the water easier to tell apart from everything around it. A common belief is that a satellite image is simply a photograph from very high up. Often it is a careful translation.
Trick three: read the surface for what lies below. Greenland’s ice sheet covers 1.7 million square kilometres and is more than 3 kilometres thick at its deepest. No one has seen its floor. But ice flowing over a ridge or valley leaves faint bumps and dips on top. Satellites measure the surface in fine detail. The method, called Ice Flow Perturbation Analysis, works backwards from those ripples to the rock. About half the valleys in the existing map turned out to run further inland, in some cases by hundreds of kilometres.
Why does a buried valley matter? Ice flow gathers in valleys. Ice funnelled through a valley gets thicker. Thicker ice flows faster. Faster ice cuts the valley deeper. MacGregor described Greenland’s west coast as “El Capitan after El Capitan”. As the ice sheet retreats, flow should keep concentrating where the valleys already are. Knowing the floor helps project how the ice will move.
Who it touches
Along the Gandak, seasonal fields sit on low banks of sand and silt inside the river’s own channel, called diara. Flooding there is expected. This time the water also broke embankments and reached settled areas, including parts of Bagaha, Areraj and Fulia Khand. For those families, the satellite picture arrived after the fact.
For glaciologists, the reverse is true. Mapping Greenland’s floor is about decades, not days. The payoff is a better guess at where faster ice will push next.
The deeper story
Look again at the list of what these stories actually observed. Microwave glow. Infrared reflections. Tiny undulations on a frozen surface. None of it is the thing itself. Everything else is inference, built on physics that has been checked many times.
That is not a weakness peculiar to satellites. Most of what any of us knows, we know this way. We read a thermometer, not the fever. We read a bank statement, not the money. The useful question is rarely “did someone see it?” It is “what was measured, and what was assumed to get from there to here?”
The sea ice record answers that question unusually well. It names its instruments. It admits the hole at the pole and says plainly what it assumes about it. That small honesty is easy to skip past. It is also what makes the rest of the numbers worth trusting.
There is a gentler point too. The Pacific Northwest picture shows something lovely and almost nothing useful. Clouds hid the land, and by afternoon they were gone. Some days the right instrument simply isn’t pointed at the right moment. Knowledge, even from orbit, comes in patches. The skill lies in knowing which patch you are standing on.
Something to sit with
Which number in your own life do you trust without ever having seen what it measures?
When a picture looks dramatic, do you ask how its colours were chosen?
One thing to try: find a recent satellite image in the news and look for the instrument’s name in the caption. Then ask what it could not see.
Sources
- NASA Earth Observatory — Arctic Sea Ice Shrinks to Its 2026 Minimum — https://science.nasa.gov/earth/earth-observatory/arctic-sea-ice-shr...
- NASA Earth Observatory — Uncovering the Valleys Hidden Below Greenland's Ice — https://science.nasa.gov/earth/earth-observatory/uncovering-the-val...
- NASA Earth Observatory — Rains Swamp the Gandak River — https://science.nasa.gov/earth/earth-observatory/rains-swamp-the-ga...
- NASA Earth Observatory — Cloudy Cloak Over the Northwest — https://science.nasa.gov/earth/earth-observatory/cloudy-cloak-over-...
We report facts from the sources above in our own words and link to the originals. Interpretation is ours, not theirs.
Why can the satellites that measure Arctic sea ice keep watching on cloudy days?
Passive microwave sensors measure the microwave energy Earth naturally emits. Sea ice emits more of it than open water, and microwaves pass through clouds, so the record continues whatever the weather.
♻︎ Free to republish
Copy this HTML into your CMS. Credit line and licence are included. Republish our work — free
Every headline has a deeper story. This is ours.
What we are doing here