In the October 15 space digest, there was a note you could easily scroll past: above the Martian volcano Arsia Mons, a cloud up to 1,800 kilometers long grows almost every morning. That is roughly one-twelfth of the circumference of the entire planet. The cloud’s whole life fits into a few morning hours; the next day, the cycle repeats. And so—every dusty Martian summer. The phenomenon was recorded at least in 2009, 2012, 2015, and 2018.
What grabs you is not the size, though the size is good. What grabs you is that the nature of this cloud remained a mystery longer than that of most Martian phenomena: no model, no simulation could reproduce it. On top of its mystery, the cloud sprouted its own mini conspiracy theory: some internet skeptics seriously suggested treating it as a loop in a graphics render. Healthy doubt, but it did not survive here: the cloud was filmed by different instruments in different years. And now—in a paper published in Nature Geoscience on October 7, 2026—comes a much stronger claim: this cloud turned out to be the first case in history in which a process is directly observed in a planetary atmosphere that textbooks have always described with the words “theoretically possible, but not found in nature.”
We are talking about homogeneous nucleation: water turns into ice without a single dust grain to seed it, out of pure vapor. For the first time, observers had both the full “suspect”—the Arsia cloud—and the mechanism that creates it.
The story begins almost like a joke. Mars Express, the European spacecraft operating in Mars orbit since 2003, carries the Visual Monitoring Camera—VMC. The instrument was not designed for science at all: it was a technical camera meant to broadcast the deployment of the Beagle 2 lander live in December 2003. It did broadcast it—the lander crashed on the surface, the camera had no further purpose, and it was abandoned for years.
It was revived in 2007, when the camera was switched back on—now as a “Martian webcam” with a live audience: wide images of the planet were automatically sent off to Flickr, sometimes 75 minutes after being taken. Then, in 2016, came the next step—ESA and the University of the Basque Country’s planetary sciences group officially turned VMC into a scientific instrument. It was precisely over its images, on October 25, 2018, that ESA’s official release could not resist writing plainly: “Mars Express keeps an eye on curious cloud”—an elongated cloud had been observed behind Arsia Mons since September 13, and on October 10 it stretched 1,500 kilometers west of the summit, whose diameter is about 250 kilometers. They also clarified the crucial part: this was not a volcanic plume but an icy orographic “lee cloud,” like one behind a mountain—and similar clouds had already been seen in 2009, 2012, and 2015. The seasonality, then, was stable.
From that point onward, the cloud became VMC’s favorite toy: the camera captured it hundreds of times, and it was from the 2018–2020 time lapses that Jorge Hernández-Bernal and his colleagues dissected its life literally minute by minute. Science grown from an instrument they had intended to throw away—that is a separate genre of pleasure, and one I love dearly.
Now for the drama itself. Every morning, precisely at dawn, a compact cloud begins gathering above Arsia’s southwestern slope at an altitude of around 45 kilometers. For 3 hours, it spreads westward almost parallel to the equator, stretching into a narrow tail 1,500–1,800 kilometers long. Then something strange happens: the cloud detaches from the volcano and drifts onward as an independent structure, with a narrow “waist” immediately behind its “head”—a distinctive morphological feature visible both in the images and in successful simulations. By around 9 in the morning, the cold pocket fades, and the tail finally breaks away from the summit—after that, nothing feeds it anymore, and it evaporates.
In the same vein, there is one detail I cannot leave unmentioned: the cloud sometimes changes shape and brightness along its tail, while rare outbreaks occur when there is little dust in the atmosphere. In the new model, all of this is reproduced through the balance of 2 competing mechanisms of ice formation—and this is the first case in which the cloud’s day-to-day variability has been explained at the level of process physics: meteorology governs nucleation; nucleation governs brightness and length.
What makes the cloud grow to 1,800 kilometers while remaining thin, bright, and stable? The answer in the new paper lies in the coupling of 2 phenomena previously considered separately.
The first is a classic of dynamical meteorology. Air masses flowing toward a volcano around 20 kilometers high—the mountain itself, not the cloud—rise and generate powerful quasi-stationary gravity waves, the very sort that leave wave trails behind mountains on Earth. The ascending branches of these waves over Arsia’s western slope cool the air by 0.05 K/s—for 800 seconds. That is a colossal cooling rate: in a matter of minutes, the air rises several kilometers and its temperature falls below 130 K—even though the “average” atmosphere at that altitude sits at around 160 K. The core of this “cold pocket” forms exactly where the cloud is observed.
The second is the arithmetic of extreme supersaturation. The key point is that in this pocket, the saturation ratio with respect to ice reaches on the order of 10^5—that is, the vapor “wants” to condense 100,000 times more strongly than equilibrium permits. Under ordinary conditions, this is prevented by the fact that any dust grain consumes the excess vapor, preventing supersaturation from building up. But at an altitude of 45 km, there is almost no dust: about 0.1 particles per cubic centimeter, with ~70 ppmv of water vapor. The dust filter is absent, the constraint removed—and here opens a window for a process that on Earth and in most Mars models was considered impossible.
A little microphysics, without which the story is incomplete.
In reality, water vapor freezing in an atmosphere always condenses onto a ready-made surface: a dust grain, soot, sea salt, a bacterium. This is heterogeneous nucleation—formation on a foreign surface. Homogeneous nucleation is when water molecules themselves, out of nothing, gather into a crystal. Theory—Classical Nucleation Theory—allows this, but at a price: without a ready-made surface, a molecule must randomly find its own kind, which is why a nucleus of critical size forms only under extreme supersaturation and deep supercooling.
On Earth, this never happens in the troposphere: there is always enough dust. Mars is interesting because its atmosphere is, in principle, poorer in dust at high altitudes and colder—which is why the first practical estimates (by Määttänen and colleagues back in the 2000s) showed that homogeneous ice nucleation on Mars requires a saturation ratio on the order of 10^5. Until now, it was thought that things never got to such numbers: ordinary micron-scale dust “drains off” the excess vapor first. And everything known about the process closer to Earth amounted to speculation about ice clouds in Earth’s mesosphere—noctilucent clouds, ~85 km—which remained a theoretical curiosity.
The new work, carried out by Hernández-Bernal’s group (LMD, Sorbonne/ENS; co-authors Määttänen, Spiga, Forget), put a full stop on this in another coordinate system: a 0D microphysics model using up-to-date parameterization (the surface energy and vapor pressure of ASW amorphous ice from Murray & Jensen, Nachbar et al.) shows that in Arsia’s “cold pocket,” the rate of homogeneous formation reaches ~4 cm⁻³·s⁻¹—hundreds of ice particles per cubic centimeter within seconds. The authors then incorporated homogeneous nucleation into a mesoscale model (WRF + Mars PCM physics) and obtained a cloud that reproduces the observed phenomenon in detail: the narrow “waist,” detachment from the summit, length, lifetime.
The most beautiful part is the physical explanation for the tail’s extreme length. Particles born in the cold pocket are carried by the west wind into a region where they are in thermodynamic equilibrium with the surrounding air (saturation ratio = 1). The tail does not evaporate because it has nowhere to evaporate to: it moves through a “neutral” atmosphere, like ink through still water. Around it, meanwhile, lie regions where supersaturation persists, but where there is too little dust and too few growing crystals to consume it. The result is the inverse of what one would expect: the cloud lives not because the air is moist, but because the tail itself is a stable equilibrium surrounded by emptiness.
The authors are honest about the limits as well: the modeled cloud is 30–50% narrower than the observed one, its onset is delayed by about 1.5 hours, and its maximum length is only a quarter of the real one. This does not, however, undermine the mechanism: the sensitivity of all parameters to nucleation intensity is precisely what explains the day-to-day variability. A separate stress test is the alternative hypothesis: what if the cloud is created not by homogeneous nucleation but by rare fine “Aitken mode” dust (particles <100 nm)? The authors checked: heterogeneous nucleation on Aitken dust produces diffuse hazes that do not yield the necessary optical-depth contrast observed by HRSC. A strict upper limit was derived for Aitken-mode density—1–20 particles per cubic centimeter: as long as the real concentration does not exceed it, homogeneous formation still wins.
First and foremost—the empirical status of the process. The sentence in the paper is worth reading slowly: homogeneous nucleation turns out to be an “outstanding manifestation of homogeneous nucleation directly from water vapor in a planetary atmosphere, a phenomenon taught but considered theoretical, never observed in nature before.” The first case. In the atmosphere of another planet. Noticed with the help of a webcam written off back in 2003.
Second—Mars’s water cycle. Homogeneous nucleation is a new channel for removing vapor into ice at altitudes climate models had not allowed it to reach. The authors explicitly formulate the open questions: is this a rare phenomenon, exclusive to Arsia, or does it operate elsewhere and in other seasons? Does it affect the planet’s global water balance? The stakes are high here: the hydrogen balance—photodissociation, hydrogen escaping from the upper atmosphere—is one of Mars’s great accounting mysteries: where did the water go? Every new channel for ice clouds directly affects the altitude of the hygropause and the vertical mixing of vapor.
Third—the Earthly echo. If homogeneous nucleation is real on Mars, the argument that “it is impossible in real atmospheres” grows weaker for Earth’s mesosphere too: noctilucent clouds live at altitudes of 76–85 km, higher than all the planet’s other cloud systems, and consist of tiny ice crystals up to 100 nanometers across—there, supercooling is severe and aerosol scarce. Discussions there of homogeneous nucleation had always remained at the level of “one can dream.” Now skeptics have no formal basis for closing the question. Plus there is a direct extrapolation from ESA’s own release: the process had been sought in the upper atmospheres of Earth and Venus and not found, while Colin Wilson, Mars Express project scientist, adds that “understanding this exotic Martian cloud required exotic physics—and this may be true in other corners of the cosmos,” all the way to exoplanets.
One honest caveat about the paper’s status: it came out in Nature Geoscience, but the authors deliberately did not make the final peer-reviewed version behind the paywall open—rather than paying ~12,500 € for open access, they posted a preprint on September 29 marked “contains inaccuracies corrected in the final version.” The conclusion held up regardless: the mechanics of the observations and the model converge, alternatives were tested, and they do not work.
What I like about this story is its architectural irony. The mystery of a cloud spotted by a webcam dragged out of a cupboard was solved by a set of tools, each of which on its own looks like a museum piece: a mesoscale atmospheric engine, a 0D microphysics model, and an old laboratory parameterization of amorphous ice. Not a single new spacecraft, not a single new mission. Someone simply put the puzzle together at last and said: what if the process we filed away under “impossible” just has nothing here to stop it?
Mars turned out to be a testing ground here too, where textbook physics steps out from behind the curtain: the only known place in the Solar System where one can watch water being born out of nothing. A tail one-twelfth of the planet’s circumference long, drawn anew at dawn every dusty summer—and dissolving within a few morning hours.