Japan Measured the Kumamoto Rupture Within a Day. Naming the Fault Is Taking Longer.
Japan's mapping agency published its first radar picture of the Kumamoto rupture on 29 July, about a day after the ground stopped moving. It is not a photograph. It is the difference between two passes of the ALOS-2 satellite, one from 12 August last year and one from the day of the earthquake. Where the crust shifted in between, the difference shows up as coloured fringes wrapped around a line of faults in southern Kyushu.
The earthquake struck at 4:27 p.m. on 28 July, beneath the plains south of Kumamoto city. The Japan Meteorological Agency puts it at magnitude 7.1 on its own scale and recorded shindo 7, the top of Japan's intensity scale, in Uki City and Hikawa Town; the US Geological Survey measured moment magnitude 6.8. Those are two scales applied to the same earthquake, not a disagreement.
Depth is less settled: preliminary estimates range from about 10 to 20 kilometres and are still being revised. It was shallow either way, in the upper crust, and the motion was mostly sideways: a steeply dipping strike-slip rupture, in the USGS moment tensor. JMA gave the event a formal name, the Reiwa 8 Kumamoto earthquake, a step it takes only when an on-land earthquake reaches magnitude 7.0 with a maximum intensity of 5-upper or above. Kumamoto Prefecture's disaster headquarters, meeting after 4 p.m. on 29 July, reported 12 deaths, six people in cardiopulmonary arrest, 64 injured, and at least four unaccounted for; those figures are provisional.
What the interferogram measures is not quite what most people picture. Radar interferometry reads movement along the satellite's line of sight, towards the spacecraft or away from it, so one interferogram compresses horizontal and vertical motion into a single number in a single direction. On the northwestern part of the fault zone, the Geospatial Information Authority of Japan, GSI, reads a maximum of about 50 centimetres of movement towards the satellite, and about 10 centimetres on the southeastern part. The page is marked as a preliminary report (速報) and states plainly that more detailed analysis may change the content.
A second instrument answers a different question. GEONET, Japan's national network of permanently installed satellite-navigation receivers, records where each concrete pillar actually sits in three dimensions, before and after. In GSI's second crustal-deformation report, published at 1:30 p.m. on 29 July, the station at Senchō in Yatsushiro had moved about 87 centimetres to the northeast and subsided about 32 centimetres. That is ground displacement at one point on the map. The radar's 50 centimetres is the same deformation field seen along the radar look direction over a wide area. Both are provisional, and neither is a restatement of the other.
How far did the fault itself slip? The USGS finite-fault model, still at version 1 and re-run automatically as more seismic data arrives, puts the maximum at about 2 metres. That model also comes with a rectangle of fault plane, and the rectangle is easy to misread: the inversion begins from a plane sized by an empirical rule and then deliberately enlarged, to be certain the real rupture fits inside it. It is a search area, not a measurement of how much fault broke. The most concrete public statement about extent so far belongs to Judith Hubbard and Kyle Bradley, writing in their newsletter Earthquake Insights: aftershocks have spread along a roughly 50-kilometre stretch from the coastline south of Yatsushiro to Kumamoto, which, they write, "gives us a first picture of where the rupture likely happened."
The reason anyone was watching this stretch of ground goes back a decade. In April 2016, a magnitude 7.3 earthquake and its large foreshock broke the Futagawa fault and the northernmost section of the Hinagu fault zone, then stopped. The sections to the south did not move. Stress does not vanish when a rupture halts: it is handed on to whatever lies past the end of the break. A 2022 study in Tectonophysics found that the Coulomb stress change on the Hinagu section was positive immediately after the 2016 mainshock, bringing it closer to failure, and Hubbard and Bradley note that the 2026 epicentre sits inside the red zone of the stress map Ross Stein and Shinji Toda published that year. Their conclusion is careful: "it certainly seems plausible that this earthquake was promoted by the 2016 sequence."
That is where this story is usually left, and leaving it there would be misleading. The same 2022 study followed the stress field forward in time, and the picture changed: later in the post-mainshock period, the optimum slip planes had rotated oblique to the Hinagu section, and its slip tendency was low, meaning the favourability of a future rupture there had declined rather than grown. One published analysis, therefore, put this stretch closer to failure just after 2016 and further from it later on.
The name of the fault that moved is unsettled too. GSI labels its interferogram with the Hinagu fault zone. Hubbard and Bradley, reading the early aftershock locations and the northeast-southwest orientation of the focal mechanism, guess instead at the Yatsushiro and/or Mifune fault zones, call that "just a preliminary guess," and point out that some maps apply the name Hinagu to those structures as well. Shinichi Sakai of the Earthquake Research Institute at the University of Tokyo has said southern Hinagu. The three labels describe much the same piece of crust, and choosing between them will take field observations of ground rupture and further radar.
The fault itself was not previously unknown: these structures are catalogued in the Active Fault Database of Japan, with slip rates and long-term probabilities attached to them. What is new is the measurement of how this one moved.
The offshore side of the response is quickly told. JMA issued a tsunami advisory at 4:29 p.m., two minutes after the earthquake, for the Ariake and Yatsushiro seas, with a maximum expected wave height of 1 metre; the advisory category covers expected heights between 20 centimetres and 1 metre. It was lifted at 6:10 p.m. the same evening, 101 minutes after the rupture.
JMA's advice for the week after the earthquake is straightforward. Because there have been past cases in this region of similar-sized earthquakes continuing for about a week after a large shock, the agency says, areas that shook strongly should stay alert for about a week for an earthquake of up to about intensity 7, and shaking-producing earthquakes are most frequent in the first two or three days. Coulomb stress transfer explains why a neighbouring segment can become more likely to fail eventually.
Every number here is provisional in the ordinary sense that it may be replaced. GSI will, then, publish updated interferograms and further reports from GEONET, the USGS will re-run its finite-fault model as more seismograms arrive, and field teams will go looking for surface rupture. Within a few weeks, the fault should have one name instead of three.