Valles Marineris, the Solar System’s largest canyon at about 3,900 km long and over 9 km deep, was thought to be a fossil landscape. A study in Nature Communications led by Jung-Hun Song and Doyeon Kim of Imperial College London links marsquake S0976a, recorded by NASA’s InSight on 25 August 2021 from the far side of Mars, to a steep fault 20–40 km beneath the canyon where blocks of crust slid sideways. Impacts and landslides were ruled out. Caveats: it is one quake from one seismometer, and InSight has been silent since 2022.
The biggest canyon in the Solar System is still moving
Valles Marineris is a scar across Mars about 3,900 km long, up to 600 km wide and more than 9 km deep: it would stretch from Lisbon to Moscow. It formed billions of years ago, and it was long treated as a fossil landscape. A study in Nature Communications led by Jung-Hun Song and Doyeon Kim of Imperial College London, with Folarin Kolawole of Columbia University, now says at least one fault beneath it can still break today.
The evidence is a single marsquake, S0976a, recorded by NASA’s InSight lander on 25 August 2021. It came from the far side of Mars, and it is the only quake InSight could locate in the planet’s western hemisphere. Earlier work had placed it roughly near the western end of the canyon; the new study pins down what caused it.
How do you know it was a fault?
The team tested every explanation they could think of. A meteoroid impact would leave a fresh crater, and a big landslide would leave a fresh scar, but Mars Reconnaissance Orbiter images showed neither. Neither could reproduce the seismic waves either.
What fits is a strike-slip quake: two blocks of crust sliding sideways past each other along a steep fault about 20 to 40 km below the canyon floor. “What makes the result compelling is that several completely different types of evidence point in the same direction,” Song said. “This is not just a ‘dead’ fossil landscape.”
QuakeS0976a, 25 August 2021
Recorded byNASA InSight (Elysium Planitia)
Sourcefault 20–40 km under Valles Marineris
Typestrike-slip (blocks slide sideways)
Canyon≈ 3,900 km long, > 9 km deep
PaperSong, Kolawole, Kim et al., Nat. Commun. 2026
One quake, two sides of Mars
Original Astrometrik animation, simplified and not to scale. Left: seismic waves from marsquake S0976a under Valles Marineris crossed the planet to InSight on the far side. Right: the proposed source, a steep fault 20–40 km below the canyon where two crustal blocks slip sideways (strike-slip); vertical scale exaggerated.
A planet without plate tectonics, but not without faults
Earth’s big quakes happen where tectonic plates grind past each other. Mars has no moving plates, so its quakes are rarer and weaker. Most of the roughly 1,300 quakes InSight detected came from Cerberus Fossae, a young fault zone near the lander. Valles Marineris was thought to have finished its work when the Tharsis volcanoes beside it grew, between about 4.1 and 3.0 billion years ago.
The new result suggests old faults can still rupture under today’s stresses, for example as the planet slowly cools and shrinks. “On a planet without plate tectonics, structures created billions of years ago can remain important,” Kolawole said. And because the waves crossed the whole planet, they also carry information about Mars’ deep interior. “From a single marsquake, we can learn both about the centre of the planet and about an active fault beneath one of its most remarkable landscapes,” Kim said.
Honest caveats
It is one quake, and InSight was a single seismometer, so its location is still approximate. The strike-slip mechanism is the best fit to the data, not a direct observation of a moving fault. And the data are not new: InSight stopped working in December 2022 when dust covered its solar panels; this is a fresh analysis of its archive.
Nobody is in danger: the quake was small by Earth standards and happened on the far side of Mars, more than 8,000 km from the lander. But it matters for planning. Future landers and, one day, crewed bases will want to know where Mars still shakes, and a network of seismometers, rather than one, would settle how active the canyon really is.
How the cause was found
Each explanation was tested against the waves InSight recorded:
ruled outMeteoroid impactno fresh crater in orbiter images; waves did not match
ruled outLandslideno new landslide scar; waves did not match
fitsStrike-slip fault 20–40 km deepbest fit to the recorded waves
Canyons compared
Valles Marineris (Mars)
Grand Canyon (Earth)
Length
≈ 3,900 km
≈ 446 km
Width
up to 600 km
up to 29 km
Depth
> 9 km
≈ 1.8 km
Made by
crust pulled apart near the Tharsis volcanoes
the Colorado River cutting down
Still active?
at least one fault, says the 2026 study
river erosion continues
Find Mars in the morning sky
Mars is in Leo before dawn this autumn, and on 16 November it passes just 1.2° from Jupiter. Tap the button: the camera view points you to it with arrows.
A 2026 study in Nature Communications links marsquake S0976a, recorded by NASA’s InSight on 25 August 2021, to a fault 20–40 km beneath western Valles Marineris. It suggests at least one fault under the canyon can still rupture today.
How big is Valles Marineris?
About 3,900 km long, up to 600 km wide and more than 9 km deep: roughly nine times as long and five times as deep as the Grand Canyon in Arizona.
Does Mars have earthquakes?
Yes, marsquakes. NASA’s InSight lander detected about 1,300 of them between 2018 and 2022. They are weaker than big earthquakes because Mars has no moving tectonic plates.
How do scientists know it was not a meteorite impact?
Orbiter images showed no fresh crater or new landslide at the source, and impact or landslide models could not reproduce the recorded waves. A sideways-sliding (strike-slip) fault deep below the canyon fits best.
Sources and calculation
Sources: J.-H. Song, F. Kolawole, D. Kim et al., Nature Communications (October 2026), DOI 10.1038/s41467-026-78193-6; Imperial College London news (15 October 2026); Universe Today (23 October 2026). Grand Canyon dimensions from the US National Park Service; InSight quake count from NASA/JPL.
Positions and times were recalculated with Astronomy Engine, the sky-calculation library Astrometrik uses; times for Vienna and Istanbul are local time. The artwork on this page was drawn by Astrometrik.