Keywords

1 Introduction

Certain radioactive xenon isotopes (\(^{131m}\)Xe, \(^{133}\)Xe, \(^{133m}\)Xe, \(^{135}\)Xe; further on called radioxenon) are monitored by the International Monitoring System (IMS), with the purpose of verifying compliance with the Comprehensive Nuclear-Test-Ban Treaty. If such radioxenon is measured at an IMS station, atmospheric transport models can be used to determine its source. These atmospheric transport models are, however, prone to errors. Therefore it is important to know the uncertainties on the analysis of nuclear explosion signals picked-up by the IMS.

Civil sources such as medical isotope production facilities and nuclear power plants also emit radioxenon. In this paper we backtrack episodes of high \(^{133}\)Xe concentration at the IMS station RN33 in Schauinsland (Germany) to see whether we can link these to the medical isotope production facility Institute for Radioelements (IRE) at Fleurus (Belgium), which is the largest regional \(^{133}\)Xe emitter. An ensemble of \(50+1\) members is used to quantify uncertainty from the meteorological data.

2 Data and Methods

The atmospheric transport and dispersion simulations have been realized with the Lagrangian particle model Flexpart Stohl et al. (2005). Flexpart can be used to calculate the source-receptor-sensitivity in backward mode Seibert and Frank (2004).

The Ensemble Data Assimilation product of the European Centre for Medium-Range Weather Forecasts Buizza et al. (2008) has been used for the backtracking simulations. The spread between the individual realizations or members contains information about the uncertainty of the simulation.

The measured concentration at a receptor can be reconstructed by multiplying a source field Q(x, y, z, t) with the source-receptor-sensitivity field SRS(x, y, z, t), outputted by Flexpart, and by summing over all grid boxes and times (x, y, z, t):

$$\begin{aligned} obs = \sum _{x,y,z,t} Q(x,y,z,t) \, SRS(x,y,z,t) \end{aligned}$$
(23.1)

If we assume time independent point sources, we can readily calculate Q for every grid point.

3 Results

3.1 Case Study

We consider four consecutive days of elevated \(^{133}\)Xe concentration at RN33 in July 2014 (Table 23.1). For each measurement we calculate the SRS fields, from which we can calculate a corresponding source term. If the calculated source term is below \(10^{13}~Bq/day\), which is the maximum expected \(^{133}\)Xe release from civil sources Saey (2009), we consider that grid point as a possible source area. The possible sources for the four measurements are shown in Fig. 23.1 (left).

Table 23.1 Elevated activity concentration measurements at RN33 in July 2014
Full size table

We assume that all four \(^{133}\)Xe samples originates from the same source, and therefore plot the overlap of possible source areas in Fig. 23.1 (right). There are three areas where all four possible source regions overlap: the first area extends from the detector RN33 to the northeast part of France, Belgium, Luxembourg and the Netherlands and a large part of the North Sea. The second area is located south-west of Denmark. The third area is located at the border of Poland and Ukraine.

If we exclude grid points where the release exceeds \(10^{11}~Bq/day\) (which can be considered as the maximal release from nuclear power plants in that region), only a very small area around RN33 remains (not shown). We therefore consider IRE as the most likely source.

Fig. 23.1
figure 1

Left for each sample, the area of possible sources is shown in transparent colors (assuming a source term below \(10^{13}~Bq/day\)). Right overlap of the areas of possible sources shown left

Full size image

To determine the \(^{133}\)Xe release, we have interpolated the SRS fields to the location of IRE. Figure 23.2 (left) shows the time series of the SRS of the four samples. The corresponding source term for each measurement is shown in Fig. 23.2 (right). We see that all four source terms are similar and the uncertainty depends on the flow.

Fig. 23.2
figure 2

Left SRS at IRE. Right source term for sample 1 (most left) to sample 4 (most right). Shaded areas (left) and error bars (right) represent 95% confidence interval

Full size image

3.2 Ten Highest \(^{133}\)Xe episodes at RN33

We have selected the ten highest \(^{133}\)Xe episodes at RN33 in 2014 and have calculated the corresponding source term (not shown), assuming IRE as dominant source. Nine out of ten gave a realistic source term. However, for one case we were not able to attribute the observed \(^{133}\)Xe to IRE, suggesting that nearby nuclear power plants can also significantly contribute to the \(^{133}\)Xe concentration at RN33.

4 Summary

ATM is a valuable tool for the Comprehensive nuclear Test-Ban-Treaty verification regime since it allows to calculate possible source regions and to estimate, for a specific location, when releases could have occurred, and what amount of radioxenon should have been released under different release duration assumptions.

We have selected elevated \(^{133}\)Xe concentrations at RN33 and have calculated the corresponding source term assuming that IRE was the dominant source. Nine out of ten observations could be attributed to the IRE. The use of an ensemble allows to quantify uncertainty of the calculated source term.