The Earthquake That Came First

At 21:26 local time, a magnitude 6.2 earthquake struck beneath Kumamoto City on the island of Kyushu. It was the strongest earthquake ever recorded in Kyushu by the Japan Meteorological Agency's intensity scale — level 7, the maximum the scale allows. Nine people died. More than eleven hundred were injured. The 400-year-old walls of Kumamoto Castle shed 40,000 individual stones.

It was a catastrophe. It was also, it turns out, a prologue.

Twenty-eight hours later, a magnitude 7.0 earthquake struck along the same Futagawa-Hinagu fault system. The April 14 event — the one that killed nine people and collapsed buildings and felt, to everyone who experienced it, like the earthquake — was retroactively reclassified as a foreshock. A prelude. The word "foreshock" is linguistic sleight of hand: it means "the earthquake that happened before the earthquake we decided was the real one."

The two earthquakes and their aftermath killed more than 270 people — the deadliest seismic event in Japan since the 2011 Tōhoku disaster.

Here is the problem that seismology cannot solve, and has never been able to solve, and shows no indication of solving: you cannot distinguish a foreshock from a mainshock in real time. The classification is retrospective. You know it was a foreshock only after the larger event arrives. The signal and the warning are the same thing — destruction. The instrument speaks in the language of the disaster it is supposed to predict.

The Japan Meteorological Agency operates one of the most sophisticated seismic monitoring networks on Earth. Thousands of stations. Real-time data processing. Earthquake early warning systems that can deliver seconds of advance notice before shaking arrives. These are extraordinary tools, refined over decades of investment and engineering and hard lessons learned from Kobe and Tōhoku.

None of them could tell the people of Kumamoto that the earthquake they had just survived was not the earthquake. That the real one was 28 hours away.

Some residents who had evacuated after the April 14 shock returned to their homes. Building inspections were underway. The aftershock sequence appeared to be decaying normally. The instruments registered everything perfectly — every waveform, every frequency, every magnitude. The data was flawless. The interpretation was impossible.

This is the particular cruelty of seismology's limitation: the instruments work beautifully. The seismographs recorded both events with precision that would have been unimaginable a century ago. We can measure exactly what happened, down to sub-second timing and centimeter-scale fault displacement. We can tell you everything about an earthquake except the one thing that matters — whether it's finished.

The seismograph doesn't know if it's recording a finale or an overture. It just records. The tool works perfectly. It simply cannot answer the question we most desperately need it to answer.

There is something genuinely vertiginous about building instruments more precise than our ability to use them. The data exceeds the framework. The measurement outlasts the meaning. We have built tools that capture reality with extraordinary fidelity and frameworks that cannot interpret what those tools are telling us.

Somewhere beneath Kumamoto, the Futagawa-Hinagu fault system is still there. The instruments are still listening. They will record the next earthquake with the same exquisite precision.

And they will not be able to tell us whether it is the event — or the warning.

Sources:

• 2016 Kumamoto earthquakes — Wikipedia
• The 2016 Kumamoto earthquake sequence — Proceedings of the Japan Academy, Series B, 2016-10
• Can you predict earthquakes? — USGS