Extreme situations demand speed: researchers at the Paul Scherrer Institute PSI have developed a new computational method that makes it possible to identify radioactive material from the air more accurately within seconds and to determine its quantity and location more precisely. This significantly improves on the usual method of detection from helicopters. The method can also be applied in geophysics or monitoring of nuclear material.

There are emergencies one hopes will never occur, but which society and government nevertheless want to be prepared for. These include the potential release of radioactivity after a reactor accident or a terrorist attack using radioactive material. For authorities to react rapidly and effectively in such cases – that is, to initiate evacuations and distribute appropriate protective equipment or iodine tablets – the extent of the danger to people and the environment must first be assessed. Among the things that would come into play in a situation like that are flying laboratories – helicopter-borne detectors with which the position, type, and quantity of radioactive material can be determined.

In research for his doctoral thesis, carried out in the Calibration Laboratory of the Radiation Safety and Security Division at the Paul Scherrer Institute PSI, David Breitenmoser developed mathematical tools that significantly improve data analysis in flying laboratories: radioactive material can be identified more accurately, and in a matter of seconds. The results of this research have now been published in the journal Nature Communications. The new analysis method is not only relevant for disaster situations but can also be used, for example, in the monitoring of nuclear facilities.

The blurred fingerprint

“Every radioactive material has a characteristic fingerprint created by the gamma rays it emits during radioactive decay,” Breitenmoser says. This fingerprint consists of the energy spectrum of the gamma rays, which can be measured with a special detector. In principle, the distribution of gamma-ray energies can reveal both the type and quantity of the material.

However, there is a difficulty: the gamma photons emitted by the material – high-energy light particles that make up gamma rays – are scattered on the way, both by the air and by the helicopter carrying the detector. This alters the energy spectrum. In other words, the fingerprint is blurred and therefore no longer directly recognisable.

To find out how a specific radionuclide fingerprint is distorted and to calibrate the detector accordingly, researchers simulate these effects on a computer. However, such calculations are very complex and can take several hours or even days. They therefore would not come into play for time-critical applications such as a reactor accident.

“The simulations are also very slow because, up to now, they have been somewhat inefficient,” Breitenmoser says. For example, of around 10,000 gamma photons emitted by a radioactive source on the ground, only one reaches the helicopter – and even that is scattered by the many metallic surfaces in the helicopter.

Speeding simulations by splitting problems

This is where Breitenmoser’s method comes in: instead of simulating the entire journey of the gamma photons from the ground to the flying detector, he splits the problem into two parts. Scattering in the air is calculated separately, taking into account factors such as the current air temperature and the helicopter’s altitude. Further scattering within the helicopter, which depends among other things on its flight attitude, is accounted for by previously calculated mathematical functions. Instead of several hours, a complete simulation now only takes a few microseconds.

However, fast calculations alone are not enough to unambiguously identify the distorted fingerprint. Since very few gamma photons reach the detector, the statistics are extremely poor – instead of clear lines, only broad and blurry patterns are visible. “The gold standard up to now has been the aggregation of several energy values in the spectrum. While this method does improve the statistical power of the measurement, it leads to a coarser picture and thus less detailed information,” says Malgorzata Kasprzak, head of the Calibration Group at PSI and co-author of the new study. The new approach forgoes such aggregation. Instead, so-called Bayesian modelling is applied directly to the entire spectrum.

The Bayesian method works with probabilities: to identify an energy spectrum in spite of the poor statistics, the computer simulates the scattering of gamma photons from the ground to the detector for thousands of possible output spectra at various ratios. Thanks to the splitting of the calculations, this is very fast. The Bayesian modelling then calculates which output spectrum most likely corresponds to the distorted spectrum measured by the flying laboratory.

Precise determination in seconds

“With this, in the future, we will be able to distinguish between weak radioactive barium or cesium samples in just a few seconds with a measurement from a helicopter, determine their strength with high accuracy, and pinpoint their location precisely,” says Alberto Stabilini, who is involved in conducting aeroradiometry measurement flights in Switzerland for PSI.

In May and June 2025, for example, flights took place in the vicinity of the Mühleberg and Gösgen nuclear power plants as part of the annual measurement campaign. This international exercise with 140 participants from six countries, six helicopters, and five drones involved training for various scenarios, including the search for radioactive sources. Such measurement flights, which are also used to create reference maps of natural radioactivity, are an important application area for the newly developed analysis method, in addition to emergency preparedness.

“But there are many other applications as well,” Breitenmoser says. Because of its high sensitivity and accuracy, the method is generally suitable for locating and characterising weak radioactive sources in the ground. This can be useful, for example, in geophysical investigations of the subsurface, or for monitoring of nuclear material by the International Atomic Energy Agency (IAEA).