Saturday, August 25, 2007

so when is the next Big One due, anyway?

Two days ago, I was here, at the mouth of Little Cottonwood Canyon, just south of Salt Lake City. I was waiting for an earthquake. It didn't happen.

Here's the view to the south:

I was standing on the lateral moraine of Little Cottonwood Canyon, looking at the moraine of the next canyon to the south. Those steps in the ridge are the scarps of normal faults. I was standing on the surface trace of the Wasatch Fault, on the eastern edge of the Basin and Range.

G.K. Gilbert actually recognized the fault, back in the 1870's. And he didn't just recognize it. He mapped out the scarps, he concluded that the fault had moved a number of different times... and, in 1883, he issued an earthquake warning for the residents of Salt Lake City.

They kept building.

And the earthquake hasn't happened.


It's weird, jumping back and forth from geologic to human time scales. As a geologist who has worked on rocks that are 100 million, 400 million, 1700 million years old, I tend to consider the last few thousand years as essentially yesterday. But if you're going to warn people about building a city on an old lake bed beside an active fault, that's not the right perspective to have.

The wonderful field trip guide (Bruhn and others, 2005*) that I was following ended its description of this stop with this carefully worded but rather dire warning:

Evidence from Little Cottonwood Canyon and other nearby sites shows that the elapsed time since the most recent surface faulting on the Salt Lake City segment is equal to or greater than the Working Group's preferred recurrence interval estimate (500-1300-2400 yr) for the segment, indicating that the Salt Lake City segment is a candidate for the next large surface faulting earthquake on the Wasatch fault.

In other words: on average, earthquakes occur on the Wasatch Fault every 1300 years. It’s been 1300 years since the last one. So, therefore...

...except that the error in the recurrence interval is 400 years. And, well, are recurrence intervals actually the best way to predict earthquakes? In this month’s issue of Geology, Robert Yeats published a commentary on paleoseismology and the problem of earthquake “schedules”. Basically, the commentary boils down to this: earthquakes are not periodic phenomena. Earthquakes can occur in clusters. There can be long periods without many earthquakes (for instance, the northern San Andreas fault was quiet for decades after the 1906 earthquake). The slip during an earthquake should change the stresses on adjacent segments of the fault, and on surrounding faults, and some of those changes will increase earthquake risk, and some will decrease it.

So is the idea of an earthquake recurrence interval outdated? Or should it be viewed more like the recurrence interval of floods: like a long-term probability of an event whose real likelihood of occurring is controlled by other events (like weather patterns in the case of flood hazards)?

A thousand-year recurrence interval says something, I think. Building for an earthquake is more important in California (with recurrence intervals in the 100’s of years) than in Utah, and it’s probably more important in Utah than in Boston (which felt shaking from the intraplate Cape Ann earthquake in the 1700’s). But it isn’t anything to set your watch by. Or, for that matter, your calendar.

* Reference:

Bruhn, R. L., DuRoss, C. B., Harris, R. A., and Lund, W. R., 2005, Neotectonics and paleoseismology of the Wasatch fault, Utah: in Pederson, J., and Dehler, C. M., eds., Interior Western United States: Geological Society of America Field Guide 6, p. 231-250.

No comments: