Site icon Temblor.net

Some California faults may have longer recurrence intervals than thought

Improved estimates of earthquake recurrence interval could account for an apparent earthquake hiatus in California.
 

By Rachel Rochester, Ph.D., Science Writer (@RaRaRhapsody)
 

Citation: Rochester, R., 2022, Some California faults may have longer recurrence intervals than thought, Temblor, http://doi.org/10.32858/temblor.279
 

In California, scientists have suspected that a large ground rupturing earthquake is likely to strike one of the state’s major faults. Paleoseismoloigsts, who search for evidence of past earthquakes by sometimes digging into the faults themselves, have documented eight ground-rupturing earthquakes along the San Andreas and San Jacinto Faults since 1800, but none in the past 100 years. The long hiatus has led to concerns that these major strike slip faults may be ready to slip again, which could spell trouble for California.

However, in a new study published in Seismological Research Letters, one scientist suggests that the length of time experts anticipate between ground-rupturing earthquakes, called the “recurrence interval,” may be longer than previously thought for some of California’s faults.
 

The 1906 earthquake ruptured along the San Andreas Fault. Credit: J. B. Macelwane Archives, Saint Louis University

 

Introducing uncertainty

Faults produce numerous earthquakes big and small over time as stress periodically builds and is released. When scientists evaluate how often a fault slips in such an event, they rely on estimates of when previous earthquakes have occurred, which often come from records of disturbed sediments or offset features in a landscape.

But many factors can affect these paleoseismic records and accounting for this uncertainty can improve estimates of earthquake recurrence interval, according to Devin McPhillips, a geologist at the U.S. Geological Survey (USGS) and the study’s author.

At eight key paleoseismic study sites — six on the southern San Andreas Fault, one on the San Jacinto Fault in southern California and one on the Hayward Fault in northern California — McPhillips assessed how convincing the evidence, collected from layers of rocks and sediment, was that past earthquakes had occurred. He then calculated a factor known as the “event likelihood,” which is the probability that a fault experienced a proposed past earthquake.

Until now, event likelihood has rarely been considered in estimates of recurrence intervals. In the few instances in which it was incorporated, it was represented by two separate numbers; scientists would offer one average recurrence interval using only the most credible evidence, and another using all of the available evidence, regardless of quality.

McPhillips simplified the process by conceptualizing event likelihood with a single number. His approach builds on the foundational work of prior studies. He separately scored the strength of the evidence for each temblor and the quality of the sediment record in which the evidence was collected.

“My hope is that going forward, more and more people will use some sort of ranking system to get at event likelihood,” says McPhillips. Scaling a semi-qualitative ranking system can be challenging because such systems are subjective. McPhillips says he focused on making his ranking system concise, in part, so it could be more easily replicated.

Using this method, McPhillips found that recurrence intervals for some sections of the San Andreas, Hayward and San Jacinto faults were, on average, 16% longer than previously estimated. Importantly, the newly calculated recurrence intervals were different enough from previous estimates that they raise the probability of prolonged earthquake hiatuses.
 

Some sections of the San Andreas Fault may have longer recurrence intervals than previously thought. Credit: John Wiley, CC BY 3.0

 

Detective Work

Identifying past earthquakes involves detective work. Though evidence for seismic activity may be strong, it’s not always incontrovertible. The evidence gathered at any given site is somewhat uncertain, says USGS geophysicist Glenn Biasi, who was not involved with the study. Analyzing earthquake data in the geologic record is like analyzing traffic based on snapshots in time, he says.

A street next to a school will have a flurry of activity during the fifteen minutes when students are being dropped off, but after the parents, students and busses depart, it may look virtually deserted. Only analyzing one of those snapshots would not reveal the entire picture of traffic flow on the street.

Like in the traffic analogy, considering the geological record more holistically can reveal a more detailed picture. One benefit of considering maximum likelihood, as McPhillips does, is that it can help explain a “flurry, or even an absence of data,” Biasi says.

When calculating recurrence intervals, experts must not only consider uncertainties, like those McPhillips focused on, but also missed events says Kate Scharer, a research geologist at the USGS who was not involved with the study. Missed events are prehistoric earthquakes that are not easily confirmed because any evidence of their occurrence has been obscured, for instance, by surface erosion or burrowing organisms. Like uncertainties, missed events pose a particular problem for recurrence interval calculations because including them can artificially shorten recurrence interval estimates. When paleoseismologists attempt to factor missed events into estimates, they run the risk of overcorrecting, and thereby “over-interpreting” the number of earthquakes that occurred in the past. The recurrence intervals revised by McPhillips’ study already included considerations for missed events, but had not previously addressed event likelihood uncertainties.

Converting observational data into a single metric “was really important for the new update of the National Seismic Hazard Map, which already addressed the likelihood of missed events, but did not have a way to deal with the issue of over-interpretation,” Scharer says. This over-interpretation problem is exactly what McPhillip’s new metric aims to mitigate. “This is an issue that a lot of papers and paleoseismologists have wrestled with but have not taken across the finish line, quantitatively,” says Scharer.
 

The current long-term National Seismic Hazard Map. Credit: USGS, public domain

 

Reassuring Findings

This work confirms that existing methods for estimating recurrence intervals are, overall, quite effective, McPhillips says. Although incorporating event likelihood lengthened the average recurrence interval for the faults considered in this study, the resulting shift was also comforting, in that it confirmed that incorporating event likelihoods or missed events shifted recurrence intervals, but not dramatically enough to necessarily warrant a complete overhaul of previous estimates.

“One of the most important conclusions for me is that in a lot of applications… we probably don’t have to worry too much about event likelihood or missed events,” McPhillips says. “That’s reassuring, and it supports a lot of past work,” he says.

“That said, I hope future paleoseismic investigations quantify both event likelihood and missed events in order to improve future recurrence interval estimates.”
 

Further Reading

Biasi, G. and K. Scharer (2019). The current unlikely earthquake hiatus at California’s transform boundary paleoseismic sites. Seismol.l Res. Lett., 90, no. 3, 1168-1176.

Castillo, B.A., S. F. McGill, K. M. Scharer, D. Yule, D. McPhillips, J. McNeil, S. Saha, N. D. Brown, and S. Moon (2021). Prehistoric earthquakes on the banning strand of the san andreas fault, north palm springs, California, Geosphere 17, no. 3, 685-710.

McPhillips, D. (2022). Revised Earthquake Recurrence Intervals in California, USA: New Paleoseismic Sites and Application of Event Likelihoods. Seismological Research Letters.

Scharer, K., R. Weldon, G. Biasi, A. Streig, and T. Fumal (2017). Ground-rupturing earthquakes on the northern Big Bend of the San Andreas fault, California, 800 AD to present, J. Geophys. Res. 122, no. 3, 2193-2218.