Helicopter gamma-ray data reveal radiation sources and their locations together
A peer-reviewed Nature Communications methods study combines radiation-transport simulations with Bayesian inference to estimate radioactive-source identity, location and activity together. Tests used known point sources and helicopter measurements in Switzerland. Six seconds of acquired data supported joint estimates, with locations recovered within two metres in the tested configurations. That acquisition time is distinct from computation time, and broader contamination mapping still needs validation.
Science··Evening
One calculation combines source and position
A methods study published in Nature Communications on October 26 combines radiation-transport simulations with Bayesian statistical inference to estimate radioactive-source identity, location and activity together. The Paul Scherrer Institute describes the same method for helicopter-borne gamma-ray measurements. Gamma rays interact with air, the ground and the aircraft before reaching the detector; changing distance and orientation further complicate interpretation of short spectra.[1], [2]
Known sources tested in Swiss flights
Validation used Swiss measurements collected at the Thun training ground on June 16, 2022. A Super Puma helicopter observed sealed caesium-137 and barium-133 sources, deployed individually and close together. One configuration hovered at about 90 metres and another made single passes at about 30 metres, travelling at 10 metres per second. Natural terrestrial radiation was compared with independent measurements at 23 reference locations.[1]
Six seconds of pass data around peak count rates supported source locations within two metres of their known positions. Median activity estimates differed from laboratory references by less than 2% for single sources and less than 4% for paired sources, within the reported combined uncertainties. The weaker barium contribution could be identified alongside the stronger caesium signal. These figures describe the tested configurations.[1]
Precomputed responses and remaining limits
Instrument responses and radiation fields were computed in advance and then assembled to reduce repeated inference costs. That preparation requires substantial computing resources: six seconds is the data acquisition interval, not total calculation time. Validation concerns controlled point sources and known flight conditions. Platform position and orientation uncertainties were not fully propagated through the model. Extended contamination, other platforms, changing natural background and detector effects at high count rates remain limits requiring further work.[1]
