San Francisco has been hit by a big earthquake at least once every hundred years going back

as far as we know.

So the people of San Francisco know that sometime in the next 100 years, they’re likely to

get shaken again by a big quake.

But we can't say exactly when the quake might hit.

Right now, all we can do is construct shake-proof buildings and put out seismic sensors.

That way, when an earthquake sends out underground waves, which travel from its epicenter eight

times faster than the destructive surface waves, we can detect the underground waves

with enough time to give a warning like: “Uh oh!

An earthquake is about to hit us!”... which is, surprisingly, enough time to turn off

gas pipelines and stop trains and find cover, but it doesn’t help people get out of town.

For people to evacuate safely from natural disasters, it’s not helpful to give a really

short warning, or a really big window within which a disaster might happen.

According to disaster experts, 2 days is juust right.

But if we want to be able to be predict earthquakes with that amount of precision, we need to

understand a LOT more about how they work.

We’ve tried looking backwards at quakes that have already happened, and identifying

events that occurred in the days before they hit, like multiple mini-quakes, big releases

in radon gas, changes in magnetism, and even weird animal behavior, to see if any of these

were predictors of an impending quake.

But lots of times these things happen without accompanying earthquakes, and lots of times

earthquakes happen without these things, so, so far we haven't been able to find any reliable

predictors.

Another approach is to build an accurate model of the earth beneath our feet: we know that

over time, as tectonic plates rub against each other, the stress that builds up is sometimes

violently released as an earthquake.

If we had a really good model and reliable measurements of the forces on the plates,

maybe then we could predict when and where an earthquake was going to happen.

But plates are often more than 15 miles thick.

That’s twice as deep as humans have ever dug, so it would be pretty difficult to get

monitoring equipment deep enough.

So, we’re creating mini-faults in the lab, to better understand the forces on moving

plates, and to help identify reliable ways to measure the forces from the surface of

the earth.

But in order to test our models, we need to be able to compare them to actual gigantic

earthquakes, which, as we mentioned, don't happen that often.

Luckily for researchers, a few ocean faults are more productive, and frequently cause

large but relatively harmless quakes, giving us a regular way to calibrate and fine tune

our models.

One big thing they’ve helped us learn is that the interactions between fault segments

are really important: for example, when this particular segment slips, it increases the

chances its neighbor will slip, letting us predict where the next quake will happen.

In some faults, we can even say that it’ll happen within a couple of years.

Compared to a hundred year window, that’s really precise, but there are still two big

problems.

First, these ocean faults are relatively simple, so we still have to figure out how to apply

what we’ve learned from them to more complicated faults, like the ones near San Francisco.

And second, even if we could do that, we’d still be a long way away from the ideal two-day

notice.

And unfortunately, our existing methods probably aren’t going to help us get there.

What we need is an earth-shattering breakthrough.

Thanks to the University of Rhode Island for sponsoring this video, which was made possible

by a National Science Foundation grant to Matt Wei, a professor in URI’s Graduate

School of Oceanography.

Dr. Wei uses seismic data and simulations to study the physics of plate tectonics and

earthquakes.

His research on fast spreading oceanic transform faults - like the Discovery fault in the East

Pacific - has helped us start to understand the importance of earthquake cycles as we

work to crack the code of earthquake physics.

Thanks University of Rhode Island.