A Los Alamos mathematician explains why we need to keep improving our seismic detectors.
The international community relies on scientific analysis of seismic data to determine whether a country is conducting an underground nuclear test. With political decisions, diplomatic relations, and, potentially, military action at stake, it’s critical that we make sure what we are seeing is what we think we are seeing. Our detection capability is especially important as proliferating nations’ testing becomes more advanced, possibly allowing for smaller and smaller tests that are more difficult to detect. If we’re going to be able to catch these bad actors, we cannot afford to be complacent. Staying ahead of the scientific and technological curve is imperative.
But first, how do we know a seismic reading is an underground nuclear test and not a mining explosion, an earthquake, or something else? It’s harder to determine than you might think.
Take, for example, the earthquake that was recorded last summer when soccer fans all across Mexico City cheered their team’s goal in a World Cup match. The initial assertion was that their collective jump-up-and-down energy shook the ground. It might have, but the energy was too dispersed. That earthquake measurement was caused by, well, an actual earthquake—in the city of Guerrero, about 140 miles southwest of Mexico City. Because they occurred around the same time, it tricked researchers into a false conclusion. In this case, no harm was done, but it clearly presents the problem seismologists all around the world face—particularly those who monitor for nuclear detonations.
An estimated 500,000 detectable earthquakes rattle the world each year. As our sensor networks get better, they observe smaller signals. While that’s a positive development, it also complicates things. More “noise” ratchets up the risk of making determinations based on coincidence, not science.
Think of it like a criminal trial. Just because a person was in the vicinity of a crime and fits the description of the assailant doesn’t mean they’re guilty. And just as a good detective must comb through mountains of evidence to accurately identify the perpetrator, so must scientists scrutinize seismic data to pinpoint its source.
We do that using complicated math and physics and supercomputers, but it’s not unlike a math problem you might have faced in school: If two trains left the same station traveling different speeds in different directions, when would they each arrive at their next stop? Those of us monitoring for nuclear tests have to answer this question—only in reverse, and with an added challenge: we don’t know the originating station. We know what time the signals arrived at the monitoring stations, but we don’t know where they came from. We need to work backwards to find out what time those signals left their source, and whether that source was one location or many.
To do this requires developing different computer calculations for different possible locations and figuring out how long it would take the signals to reach the monitoring stations. Do the calculated arrival times agree with the actual time recorded at the seismic stations? If so, then we have a “good association”: we can be pretty certain the signals came from the same location.
We must do similar calculations related to the size of the seismic waves. As a wave travels through the earth to various sensors, the different paths will change the signal in different ways. By the time it reaches a particular sensor, the wave size might have changed. To figure out the “true” size of a wave, we must undo the effects of distorting rock layers along the path. Then we can see the wave as it looked when it left the source. Again, we get a good association if the sizes of these “undone” sensor signals from different sensors are very similar to each other.
Good association requires an accurate physics understanding of how waves travel though the rock layers in the earth. Without it, we’re just making assumptions.
For an example of good association, we have another sporting event to study: a 2011 playoff game between the Seattle Seahawks and the New Orleans Saints. In that football game, the Seahawks running back broke through multiple tackle attempts to run 67 yards for a touchdown. This cheering crowd’s energy was concentrated enough to register on local seismic stations. In this instance, the researchers had good physics to point to a single source: the stadium.
We’re not worried about underground nuclear testing on Puget Sound; those tests require massive infrastructure, including deep tunnels, heavy equipment, and sophisticated instrumentation. But when the earth rattles in a more suspect location and those seismic waves are traced back to a single source, we take a closer look—analyzing data from sensors that detect radionuclide gases in the atmosphere, low-frequency acoustic waves, and changes in water pressure caused by sound waves. Commercial satellites also provide imagery to look for changes in topography.
As claims of nuclear prowess are once again taking center stage in world affairs, all of this information and expertise is critical to inform decision makers. Teams of scientists in federal agencies and the top nuclear weapons laboratories—Los Alamos, Sandia, and Lawrence Livermore—are dedicated to continuing to refine and develop sensors and analyze the data they detect.
To catch countries that are trying to fly their nuclear weapons programs under the radar, it is imperative that we be vigilant—which means pushing the scientific boundaries. As we detect smaller events, we must make sure that we’re directing our resources to track down real threats and not chasing “noise” that leads nowhere. Without state-of-the-art sensing and signature science, we’re at greater risk. Simply put, as our adversaries’ capabilities evolve, so must our own. That way, the next time the ground shakes, we know exactly what’s causing it—and can respond accordingly.