The USGS Says How the Ground Moved Under Panama, Not Which Fault Moved It
The southern coast of Panama's Azuero Peninsula has the shape of a place where one slab of ocean floor dives beneath another. The faults that cut across the peninsula itself mostly do something else. They slide past each other sideways. At 17:56 UTC on Oct. 9, 2026, one of them let go.
What broke, and by how much, the US Geological Survey has settled. Its reviewed record for the event puts the magnitude at 7.7 and the start of the break 12.6 kilometers down, near the village of Pitaloza Arriba. Its tectonic summary reports what kind of motion it was: left-lateral strike-slip faulting, which means the two sides slid horizontally past each other and the far side moved left.
What the agency does not say is which fault. The magnitude is not agreed either. Panama's Instituto de Geociencias puts the earthquake at magnitude 7.5, revised up from an initial 7.2, and at a depth of about 4 kilometers. That depth is shallower than the USGS figure by more than the uncertainty the USGS attaches to its own. Two national agencies are describing one rupture and not one set of numbers. The figures below are the USGS ones. Neither set should be read as the settled value.
The waves give the motion, not the map
A second USGS product goes further than the mechanism. Its finite-fault model solves for the shape of the break from the recorded waves, and it is now in its second version. The agency describes that version as an updated model, fitted after a search for the best-fitting fault geometry. The plane it settles on runs 288 degrees, close to west-northwest, and leans at 50 degrees. Slip along it is almost horizontal, tilted slightly down. The largest slip anywhere on the plane is 7.8 meters, and the patch that actually moved is about 87 kilometers long.
A fault that slides purely sideways usually stands close to vertical. This one, in the model that fits the waves best, does not. That is the sort of geometry you get where sideways motion and a little stretching share the same structure. It is one of the few clues in the record pointing to a kind of fault rather than a particular one.
The aftershocks line up with it. The USGS catalog holds fifteen events of magnitude 5 or greater within 200 kilometers of the epicenter through Oct. 11, 2026, the main shock among them, and the largest aftershock was a magnitude 6.6 on the evening of the quake. Their epicenters are not scattered. Plotted out, they form a band running west-northwest to east-southeast, the same orientation as the model plane. The USGS does not draw that conclusion. It is what the coordinates in its own catalog show.
Radar can check a model built from shaking alone
Everything above comes from seismic waves. There is a second and independent way to see a rupture: measure how far the ground itself moved. A radar satellite that passed over the peninsula before the earthquake and again afterward recorded its distance to the ground both times. The difference between the two passes maps the shift, down to centimeters. A fault model built from waves predicts a particular pattern of ground movement; the radar measurement either matches that pattern or it does not.
That is why the satellites mattered within hours. A NASA-funded reconnaissance effort has been cataloging every satellite acquisition over the rupture and publishing the tally. By the time its page was generated in the early hours of Oct. 10, 2026, the mission had cataloged 245 acquisitions, meaning individual frames rather than finished pictures of the rupture. Of those, 117 are from NISAR. Sentinel-1 radar, Sentinel-2 and the two Landsat satellites account for the rest. The work is led by Pietro Milillo at the University of Houston, with Sabine Loos at the University of Michigan and Jamon Van Den Hoek at Oregon State University. The page describes itself as an experimental research product under active development, and not an official damage, safety or loss assessment.
NISAR is the newest of them. NASA's mission overview records that the satellite launched on July 30, 2025, from India, on a vehicle provided by the Indian space agency ISRO. It entered routine science collection in early January 2026. It repeats the same ground track every 12 days, which is what makes before-and-after pairs over an unplanned event possible at all. Its data are free and open, so the test of the fault model is not restricted to the group that flew the mission.
What has not happened is the part that would answer the question. The reconnaissance page's own line on agency deliveries says none have been released yet, and it lists no activations. The acquisitions exist. They are counted. Nobody has yet published a measurement of ground movement from them.
The alert levels are forecasts, not counts
The USGS set its PAGER alert for this earthquake at red, the highest level it issues. PAGER is a model, not a tally: it estimates likely impact from the modeled shaking and the population that shaking reaches, before anyone has counted anything. The same shaking model estimates that about 4.6 million people felt moderate to severe ground motion. The agency runs two further models off the same shaking grid. They put the liquefaction hazard, the risk that shaken wet ground flows, at orange and the landslide hazard at red. All of those are forecasts of what the ground probably did, not measurements of what it did. The measurements are the part that has not been released.
The tectonic summary adds one more point: most of the region's recorded earthquakes have occurred west of this one, along the Panama Fracture Zone, the north-south boundary between the Cocos and Nazca plates. Fewer have been recorded on the structures that cut the Azuero Peninsula itself. A hazard picture built on very little local history is the normal condition for much of the world's coastline. It is why one well-measured rupture here is worth more than its size alone.
Which fault, then. Fault names are circulating in local reporting. None of them appears in the USGS record for this earthquake, which names the mechanism and the plate setting and stops there. What the record does say is firm enough: a long, shallow, sideways break on a plane that leans more than a sideways fault usually does. The radar that could tie that plane to a mapped structure has been collected and not yet released. Until it is, the mechanism is the answer and the fault is still the question.