Reliable monitoring programs to detect tests conducted in secret are a key component in the fight for non-proliferation of nuclear weapons. International groups such as the Comprehensive Nuclear-Test-Ban Treaty Organization are tasked with running global networks of instruments capable of detecting the tell-tail signs of weapons testing.
The majority of early tests were carried out on or near the surface and often resulted in harmful radioactive material being spread far and wide over the surrounding habitat. These tests are responsible for causing untold environmental and societal damage. In 1963, capitalising on thawing relations in the aftermath of the Cuban Missile Crisis, the US, UK and USSR signed the Partial Test Ban Treaty. The agreement limited future tests to being conducted underground, significantly reducing the risk of pollution by radioactive fallout.
Underground explosions can still be detected due to the large amounts of energy that they release in the form of seismic waves. These vibrations of the ground travel great distances and can be measured using seismometers. The challenge lies in how to differentiate seismic waves generated by explosions versus those caused by naturally occurring earthquakes. The latter are so numerous that they present a major headache to monitoring programs, where the ideal goal is to never miss an explosion.
Nukes versus quakes
Over the years, many different methods have been proposed to identify underground nuclear explosions amongst background earthquakes. Several have proven highly effective in this endeavour, but none are perfect and it is therefore important to deploy an array of different techniques.
In a paper just published in Geophysical Journal International, my colleagues and I from the Australian National University and Los Alamos National Laboratory in New Mexico have been developing a new method for identifying underground tests. It builds from a pioneering study by Ford et al. in 2020, who demonstrated that seismic moment tensors show distinct differences for earthquakes versus explosions in a manner that can be used to differentiate between the two types of event.
Moment tensors (MTs) are a mathematical construct representing the forces that operate at the source of a seismic event and give rise to the observed field of displacement. They are a 3 × 3 symmetric matrix and therefore have six unique elements. By normalising this matrix, we can project the six-degree unit vector of each MT onto the hypersphere – a five-dimensional spherical surface that exists within a six-dimensional space.
Distinct event populations on the hypersphere
We have taken moment tensor catalogues consisting of 140 known explosions and approximately 1200 known earthquakes from the western United States and projected them onto the hypersphere. Each event type forms its own population within this multi-dimensional space. The most important part of our work is demonstrating that these populations are anisotropically distributed (i.e. they have different levels of spread in each dimension).
Using a new mathematical model developed by co-author Janice Scealy, who is a statistician working in ANU’s Research School of Finance, Actuarial Studies & Statistics, we have been able to represent each of these two populations with their own probability distribution function. The moment tensor of any new seismic event can be quickly compared to these two functions, allowing us to assess the probability that it was an explosion versus an earthquake.
The method performs well on the catalogues of known events, correctly identifying 99% of explosions whilst only misclassifying a handful of the earthquakes. It also successfully identifies all six of the nuclear tests carried out by North Korea between 2006 and 2017. These improved rates render the method sufficiently robust to be of potential use in monitoring programs and we have provided a publicly available code called MTid for general usage.
A ban on all future tests is unlikely given that several major nations remain unwilling to ratify the Comprehensive Nuclear-Test-Ban Treaty. Robust techniques for identification of nuclear tests are a key component of global monitoring programs and are therefore critical for ensuring that governments are held accountable for the environmental and societal impacts of nuclear weapons testing.
Read more: