Q&A: Paul G. Richards, Nuclear Arms Seismologist

Paul G. Richards is the Mellon Professor of the Natural Sciences at Lamont-Doherty Earth Observatory of Columbia University, in Palisades, N.Y. For the last twenty-plus years, his work has focused on the use of seismological methods to study nuclear weapon test explosions and their implications in both the scientific and political arenas. He has served as an advisor on nuclear arms control at the U.S. Department of State, a visiting scientist at the Lawrence Livermore and the Los Alamos national laboratories, a technical expert to the Comprehensive Nuclear-Test-Ban Treaty Organization, and a scientific panelist for the National Academy of Sciences' report "Technical Issues Related to the Comprehensive Nuclear Test-Ban Treaty," among many other assignments in the field of nuclear testing detection and measurement.

From 2000 to 2003, he led an applied research project to improve the accuracy with which treaty monitoring organizations routinely locate seismic events, particularly in Eastern Asia. He describes the work of event location as the "last corner of seismology that is still dominated by methods developed during the era of analog recording." He says that traditional methodology has been greatly enhanced by the practice of measuring comparable seismic events in the same locale simultaneously and "locating each one relative to its neighbors, preferably using waveform cross-correlation to measure relative arrival times."

He spoke with Spectrum Online soon after the North Korean government announced, on 9 October, that it had detonated its first nuclear device underground, on the science of detecting and measuring such events.

SPECTRUM: How do scientists use seismology to distinguish powerful explosions from earthquakes?

RICHARDS: The detection of the two is more or less the same on the question of actually detecting the signal, but then the way we make the identification from the seismic end is based on the fact that each of these different types of seismic source puts out a mix of seismic waves and the mix is very different for earthquakes than it is for explosions. For example, an earthquake will almost always be greater than 5-kilometers depth in the earth, whereas explosions typically would always be much shallower than that. There are some seismic waves that, uniquely, are observed just for very shallow sources, so that would be one example.

SPECTRUM: How sensitive can seismology stations be in detecting events such as a small-scale nuclear detonation?

RICHARDS: Well, you hear my hesitation. There’s such a huge variety of signals and the answer depends on how near the station is, or out to what great distance, or on the frequency range, and so on. So it’s a little difficult to give a direct, simple response. In the present case of interest, an event in North Korea of approximately magnitude 4, that is something that can be detected at seismographic stations all around the world. Except to say that when you get to very great distances, say on the other side of the globe, you can only see that type of signal if you’re working with a very quiet station.

But that’s just the detection. In order to be very confident in making the actual identification, typically those very distant stations aren’t a whole lot of use for a small event. You really need to get high-quality recordings, preferably from distances not more than a few hundred kilometers. And again in the present case, there are a number of stations in South Korea, in Japan, and in southeastern China that captured very good signals at a relatively short distance on October 9, which enabled the work of identification to be done with high confidence.

SPECTRUM: Seismology is well-established science. What new developments are helping to refine our analysis of seismic events?

RICHARDS: For one, I think the ability to record signals at higher frequencies than we have usually cared about with earthquakes. So, for example, with earthquakes, almost all of the scientific work is done to study earth’s structure and to do earthquake location with signals at frequencies of not more than about 10 hertz, and a lot of the work is done even with signals of not more than about 2 hertz. For many years, seismographic instrumentation didn’t perhaps pay as much attention as it should have done to the recording of signals at much higher frequencies, which are more efficiently excited by explosions--where one needs to go up to, at least, 20 hertz to do a good job of discrimination. This is because an explosion generates typically higher frequencies than an earthquake of comparable overall amplitude of seismic signals.

SPECTRUM: Could you discuss the nature of the global network of seismology stations used to detect explosions, such as the North Korean test?

RICHARDS: Broadly speaking, there are three different types of network that matter here. There’s the type of network that’s operated by national technical means unilaterally. For example, the United States has agencies with responsibility for explosion monitoring, and they do the work mainly with their own network. The international community associated with monitoring the Comprehensive Test Ban Treaty of 1996 works from a headquarters in Vienna that is currently building up a global network of about 321 stations. They use four basic monitoring technologies: seismology, hydro-acoustics, infrasound, and radionuclides. So those are two broad, organized networks of sensors.

But there’s a very important third network, namely the loose aggregation of thousands of seismometers and other types of instrumentation operated for general purposes of research around the world. Whenever some interesting seismic source is recognized, one can go and search for a station that was not established for any monitoring purpose but was simply in the vicinity. And in the present case of October 9, there was a station that was operated for general purposes of seismological research that provided excellent signals because it was at a distance of only about 350 kilometers to the north.

SPECTRUM: How are recordings from various seismology stations correlated into the needed data to determine whether an event such as the explosion in North Korea on 9 October satisfied a conclusion that a nuclear device had been detonated?

RICHARDS: Working from the highest quality recordings from the morning of October 9, my colleague Won-Young Kim at my institution, was one of many seismologists who was able to get, over the Internet, the signals recorded at that station 350 kilometers to the north; but then he was able to find examples of a small earthquake recorded at the same station and at about the same distance of 350 kilometers; and it’s a comparison of the two signals recorded at the same station that gives us great confidence that the event on October 9 was explosive.

I’ve already mentioned one particular seismic wave we could see from that October 9 event that indicated that the seismic source was very shallow, unlike an earthquake. There are two other characteristics of the seismograms in the case of both an earthquake and an explosion. Both these sources put out compressional waves, that’s like sound waves or P waves where the motion is longitudinal in the same direction that the wave is traveling; and they both put out shear waves in which the particle motion is transverse to the direction in which the wave travels. Sometimes we call them P waves for longitudinal and S waves for the shear waves. The latter are much more efficiently excited by earthquakes, and compressional waves are more efficiently excited by explosions. The P to S ratio of those two different wave types is much larger for the explosion than for the earthquake.

SPECTRUM: What's the relationship between seismic readings and explosive yield of a nuclear detonation?

RICHARDS: Essentially, you look at the seismic magnitude of the signals. There are a number of different scales that are used for that purpose. One of the common ones is to look at the size of the P waves recorded at great distance and then, in practice, look at the largest amplitude of those P waves and work out what the ground motion at that distant station must have been, taking the logarithm of measured ground motion and making a correction for the distance. That is traditionally how the Richter body wave seismic magnitude is calculated from an observation, and it’s long been known that there’s a linear relationship between the magnitude and the logarithm of the yield in kilotons.

For example, for explosions conducted in hard rock the yield is a linear relationship for the magnitude against the logarithm of the yield. The relationship is slightly different for soft rock. There’s some scatter in the data points about a straight line fit, but that’s the type of relationship that’s long been used to get an approximation of the yield from a seismic signal.

SPECTRUM: Was your team able to make a conclusion about the physical explanation of the 9 October explosion and, if so, when?

RICHARDS: Well, we didn’t detect it with our own instruments, but we simply used data that other organizations had gathered, first of all to get information on the source of the event, and the U.S. Geological Survey did that work. But just a few hours after the event happened, we did our own work getting the data from the station operating in southeast China and drawing our own conclusions to identify the signals as indeed having come from an explosion. So we did that work ourselves, although it wasn’t done with our own instruments.

SPECTRUM: As a scientist, what are your thoughts on the North Korean nuclear test?

RICHARDS: This is a very dangerous development, a step back from the non-proliferation of nuclear weapons, which has been the subject of a treaty that went into effect in 1970. I’ll go on to express the hope that ways will be developed to restore confidence in that treaty, and that we won’t just to see North Korea and many other countries stepping away from it.