The aim of this work is to apply a non-parametric statistical methodology to the radioxenon activity concentrations measured at noble gas stations of the International Monitoring System of the CTBTO, in order to investigate the radioxenon atmospheric background and the radioxenon anomalous values. The proposed non-parametric statistical methodology does not require any assumption on the underlying probability distribution of the raw data. The suggested method, based on Recursive Segmentation and Permutation (RS/P), allows to detect single or multiple mean and/or scale shifts.
Radioxenon is useful to potentially reveal underground nuclear explosions (UNEs) but it is also emitted by civil sources. To discriminate signals, advanced statistical methods are used to understand the background and the anomalous values that could be reasonably related to UNEs.