Seismology reconstructs ground motion from instrument records — seismograms, GPS stations, InSAR satellite data. The measurements are precise but indirect: you see the wave that traveled, not the fault that moved. During the magnitude 7.7 Myanmar earthquake of March 2025 — the strongest on the Sagaing Fault in over a century — a CCTV camera near Mandalay recorded the fault rupture directly. Published in The Seismic Record (Kearse & Kaneko, Kyoto University), the study used pixel cross-correlation to extract displacement measurements from the video frame by frame.
The numbers: 2.5 meters of horizontal shift in 1.3 seconds. Maximum velocity 3.2 meters per second. The movement was pulse-like — a concentrated burst of slip, not a gradual slide. And the fault path was slightly curved, not straight.
The curvature matters. Standard seismological models assume straight fault segments because the inversion problem is already underdetermined — curved faults add parameters that the data can't constrain. But if the fault actually curves, straight-segment models will misattribute the curvature's effects to other parameters (stress drop, rupture velocity, directivity). The CCTV video provides a direct constraint that seismograms alone cannot: the geometry of the surface trace during the rupture.
The broader methodological point is that different measurement instruments answer different questions. Seismograms measure the wavefield — they tell you what the earth did at the sensor. GPS measures displacement — it tells you how much the ground moved at the station. Satellite radar measures cumulative surface deformation. Each requires a model to relate the measurement to the fault. The CCTV camera measures the fault itself — displacement at the source, resolved in time. No model required. The measurement IS the thing the models are trying to predict.
The limitation is obvious: CCTV cameras only exist near human infrastructure, and most fault ruptures occur where people don't live. But for the ruptures that do intersect populated areas, the direct observation constrains the models in a way that no amount of seismographic data can replicate. Sometimes the simplest instrument — a security camera — answers the question that the sophisticated instruments can only estimate.