It’s not often you see the United States Geological Survey and Jet Propulsion Laboratories get into a smackdown over science, but that’s just what happened thanks to a new set of earthquake predictions for southern California.
Researchers led by Andrea Donnellan at Jet Propulsion Laboratories (JPL) published a paper in a reputable, peer-reviewed scientific journal last month claiming that remote sensing of the La Habra earthquake in the Los Angeles basin in March 2014 indicates the earthquake didn’t release all the stress in the region. That alone isn’t surprising—California is infamous for having an active fault system, and earthquakes notoriously redistribute stress within a region without releasing all of it in seismic waves. Where things get contentious is their subsequent claim that the area within 100 kilometers of the La Habra epicenter has 99.9% odds of sustaining a magnitude 5 or higher earthquake in the next three years.
And that’s where seismologists at the United States Geological Survey (USGS) very politely flipped their shit, as expressed by seismologist Lucy Jones. The USGS released a statement reading:
This paper claims a 99.9% probability of an earthquake of magnitude 5 or greater occurring in the next three years within a large area of Southern California without providing a clear description of how these numbers were derived. The area—a 100-km radius circle centered on the city of La Habra‚is a known seismically active area. For this same area, the community developed and accepted model of earthquake occurrence, “UCERF3”, which is the basis of the USGS National Seismic Hazard Maps, gives a 3-year probability of 85%. In other words, the accepted random chance of an M5 or greater in this area in 3 years is 85%, independent of the analysis in this paper.
Translation: The USGS lists the probability of a magnitude 5 or greater earthquake hitting the area at 85% in the next three years, and calls JPL out for not detailing their forecast model like a teacher scolding a student for not showing their work on their math homework.
For context, the image at the top of this article is a map of earthquakes magnitude 3 or larger within approximately 100 kilometers of La Habra, California, in the last three years. Circle size indicates magnitude (larger circles are larger earthquakes) and color indicates age (all are at least a week old). The red lines are known fault traces; this is not exhaustive as the Los Angeles basin has previously hidden dormant faults under sediment until they abruptly rupture, like the 1994 Northridge earthquake.
The statement continues:
While the earthquake forecast presented in this paper has been published in the online journal Earth and Space Science, it has not yet been examined by the long-established committes that evaluate earthquake forecasts and predictions made by scientists. These committees, the California Earthquake Prediction Evaluation Council, which advices the California Office of Emergency Services, and the National Earthquake Prediction Evaluation Council, which advices the U.S. Geological Survey, were established to provide expert, independent assessment of earthquake predictions.
Translation: The firm reminder that earthquake models are carefully vetted prior to being used in hazard assessment is a not-so-subtle jab at the JPL researchers. They need to be responsible about making hazard assessments because it’s not nice to scare people without being damn sure you’re right.
While the tone is formal and the names make it a painfully awkward read, this is a hugely important point: policy makers depend on scientists for the best possible hazard assessments when making their decisions, and it isn’t fair to either them or the public to casually muddy the waters.
Hazard assessment is tricky business, and state-of-the-art is constantly evolving: it’s entirely fair (and welcome!) to bring new forecast models to the scientific community. But earthquakes in California are a big deal. A 99.9% probability forecast is going to instill a lot of fear; a researcher needs to go beyond the advance of science and try to be socially responsible when making that type of claim.
It’s traditional to run earthquake forecast models by other, independent experts in the field to be absolutely certain we’re not having another faster-than-light neutrinos-wait-the-cable-is-jiggly fiasco—or one where people were soothed into false security or whipped into a frenzy of fear before the flaws are found. JPL didn’t take this step, publishing directly in the journal without submitting to the committees for review. That they were also incomplete in describing their technique is even more aggravating: it’s going to take longer for other seismologists to understand what they did and identify any errors, unfairly prolonging the time when people will be grappling with the proposed near-certainty of an impending earthquake.
The final section of the statement is the most precious of all, a direct dismissal of JPL’s work until they can show their numbers (and models):
The earthquake rate implied by the 99.9% probability is significantly higher than observed at any time previously in Southern California, and the lack of details on the method of analysis make a critical assessment of this approach very difficult. Therefore, the USGS does not consider the analysis presented in this paper a reason to change our assessment of the hazard.
While we’re hoping that JPL will reveal more of the details how they got to that 99.9% prediction, for now we’re sticking with the USGS: La Habra, you still have an 85% probability of a moderate or larger earthquake within the next three years, and you should do all you can to prepare now to mitigate your personal risk as much as possible.