A single isolated, unblinking surveillance camera in Myanmar has done what five decades of seismological field observation were unable to: it captured, for the first time, live footage of an active giant earthquake fault in motion, providing a glimpse of the underground mechanics of seismic rupture. The event, which Kyoto University geophysicist Jesse Kearse described as giving “chills down my spine straight away,” offers a unique observational history of faults slipping, curving, and extending in space during a great earthquake.
On March 28, 2025, the Sagaing Fault, a wide right-lateral strike-slip boundary accommodating shear between Sundaland and the Burma plate, ruptured in a magnitude 7.7 earthquake. The hypocenter was 20 kilometers west of Mandalay at a depth of 10 kilometers, but the rupture extended over 400 kilometers with ground slip over six meters and shaking reaching Modified Mercalli Intensity X, killing 3,600 people confirmed dead and causing extensive damage to buildings. But it was a CCTV camera, 120 km south of the hypocenter and just 20 meters east of the fault trace, that recorded most scientifically useful information: the ground cracking and sliding itself, with west side moving north in an obviously curved trajectory.
The processing of this video, released in The Seismic Record, represents a methodological improvement. Kearse and co-author Yoshihiro Kaneko employed a pixel cross-correlation technique, a computerized image analysis method already optimized for geophysical phenomena such as glacier movement and landslide movement. Tracking 25 overlapping 26-by-26 pixel subsets of the image, the research group analyzed the motion of features on either side of the fault, correcting for camera motion and lens distortion from known fence post distance and post-seismic satellite imagery for calibration. The result: a reliable time series of fault slip, velocity, and path.
Observations suggested a right-lateral strike-slip of approximately 2.5 ± 0.5 meters in 1.3 ± 0.2 seconds, with a peak slip velocity of 3.2 ± 1 m/s. The slip path was curved, not linear, with an initial oblique motion (peak rake of 35°) that became rectilinear as rupture slowed. This bending, witnessed in real time for the first time, corresponds with geological evidence of curved slickenlines on faulted surfaces and validates dynamic rupture models that predict transient stress-driven bending on the surface. “Instead of things moving straight across the video screen, they moved along a curved path that has a convexity downwards, which instantly started bells ringing in my head,” Kearse said.
The short duration and high velocity of the slip function indicate a pulse-like rupture, with a concentrated burst of slip propagating along the fault—such as a skipped ripple on a rug. Both laboratory experiments and seismic inversions confirm that pulse-like ruptures are favored in the presence of velocity-weakening friction and moderate prestress, and can transition to a crack-like mode at higher prestress. As laboratory experiments demonstrate, “pulse-like ruptures can exist on such interfaces in the absence of a bimaterial effect or local heterogeneities,” and both pulse-like and crack-like ruptures can be supershear velocity under certain conditions. Most significantly, the CCTV record allows one to make direct comparisons between on-fault slip velocity and near-fault ground velocity at strong-motion stations.
2.7 kilometers from the fault but 130 kilometers southward, at the NPW station, the ground velocity function repeated the on-fault slip velocity in form, duration, and amplitude, demonstrating the value of near-fault records in constraining such rupture parameters as peak slip velocity and rise time. But the slip-weakening distance deduced from on-fault measurements (1.2 meters) was much smaller than that from strong-motion records (2.4 meters), highlighting the problems caused by free-surface effects and the value of on-fault observations. The import of this fortuitous video extends beyond academic curiosity.
The ability to quantitatively measure and observe in real time the curvature and slip velocity of faults provides a vital link between instrumental ground motion records, physics-based simulations, and geological field observations. As Kearse points out, “We did not anticipate that this video record would provide such a rich variety of detailed observations.” The results hold promise to make seismic hazard models more accurate, inform infrastructure design, and have ultimate bearing on the resilience of communities in earthquake-prone regions. With this finding, researchers now have a highly versatile new tool to examine the workings of earthquake rupture mechanics, one that someday may help unlock long-standing questions of the hows and whys of fault slippage.
