About a year after Russian-born Natalia Ruppert arrived in Fairbanks, she was playing volleyball when the gymnasium around her began to shake. Everyone else knew what was going on—Alaska is the most seismologically active state in the country, after all—but it was new to the young PhD candidate. She’d left the dismantling Soviet Union to study earth science, so “it was actually kind of exciting,” she remembers. “My first earthquake.” 

Ruppert spent 30 years studying the phenomenon in the far north, flying in helicopters to remote seismology stations. At conferences she found herself among the few women scientists in attendance, an imbalance she’s seen change drastically over the years. She finally left Alaska in 2024 for the opportunity to lead the US Geological Survey’s ShakeAlert in Seattle, overseeing a project that stretches from California to the Canadian border. With a network of thousands of sensors, it aims to catch an earthquake in its first seconds, activating thousands of automated systems like gas valves to shut off before tremors wreak havoc. A quarter century after the 2001 Nisqually quake, Seattle is still trying to prepare for the next movement from the Cascadia Subduction Zone, an event that geologic history tells us could be catastrophic. From her office at the University of Washington, with a crew of scientists, analysts, and technical partners under her purview, Ruppert thinks the Pacific Northwest can learn to respond before we even feel the earth move. 

Earthquakes—they are all so different, and there is always kind of new things to learn about plates and tectonics and human aspects as well. It’s never the same.

I was finishing up my bachelor’s degree, back in Russia, when the Soviet Union kind of collapsed and borders opened up. And it became possible to go study abroad.

Once I got to Fairbanks, this is where I got introduced to earthquake seismology, and I got hooked.

We understand more than we used to. Technology is moving ahead all the time.

We can say [Seattle] is due for an earthquake within the next, you know, 50 years, but we still can’t say if it’s going to be next week or next month, or 10 years from now.

AI is now everywhere, and seismology is no exception. The most successful application of AI and seismology so far is for earthquake detection and classification.… AI is very good at distinguishing different types of earthquakes.

Ever since the first seismologists observed earthquakes, they were looking for ways to predict a future earthquake. Because it’s very devastating.

People are trying, but there is no, like, smoking gun yet. AI is not the answer to all the questions.

Once the earthquake starts somewhere, if there are sensors in the vicinity...we can record that first initiating phase very, very quickly. Within seconds.

Then we can kind of warn people who are ahead of the seismic waves. We can warn them, “The ground shaking will start within 10 seconds.” That’s the next best thing to prediction.

For people, it’s drop, cover, and hold. But the same warning is given to systems.… If it’s a fire station, they can open up the doors so the doors don’t get jammed. If it’s a train system, they can slow the trains down. If it’s a water plant, they can shut down their pumps.

I’m a seismologist, so I understand how earthquakes work, so it’s not scary for me. But I can understand why it’s scary.

There are lots of different hazards: volcanoes, floods, windstorms. Just be aware of what your hazards [are] and be prepared. I mean, you can build a bunker in the middle of nowhere and just hide in there. You will be safe, but it would be a boring life.