Dataset on xenon-133 atmospheric background source distribution variables for graph modelling

This dataset supports the findings of our manuscript entitled «A Graph Signal Processing Approach to the Characterization of Xe-133 Atmospheric Background» and is provided to enable reproducibility. Characterizing the global xenon-133 atmospheric background is essential in the context of the Nuclear-Test-Ban Treaty. In this original study, we address this problem using a graph model that represents relationships between stations of the IMS network. Using simulated data where the contributions of each source are known, we investigate how source distributions shape the background field. Spectral clustering reveals that the stations can be grouped into distinct regions according to their source distributions. Furthermore, we use the graph model to estimate sources from the aggregated xenon-133 concentrations at each station, showing good performance on our simulations and providing insight into source-station relationships. Our graphs are built using variables provided in this dataset, through a process detailed in the "Methods" section of the manuscript.