Quantification of the modelling uncertainties in atmospheric release source assessment and application to the reconstruction of the autumn 2017 Ruthenium 106 source
Résumé
Using a Bayesian framework to solve the inverse modelling problem of release source assessment has
proven to be of signi_cant interest in recent years. Through Markov chain Monte Carlo (MCMC) algorithms,
distributions of the variables describing the release such as the location, the duration, and the magnitude as well as the errors can be sampled in order to get a complete characterisation of the source.
In this study, several approaches are described and applied to improve the rightness of these distributions, and therefore to get a better apprehension of the uncertainties. First, we propose a method based on ensemble forecasting, where physical parameters of both the meteorological _elds and the transport model are perturbed to create an enhanced ensemble. Members of the ensemble are then represented by weights and sampled at once with the other variables describing the source to take in account modelling errors. Secondly, we discuss how the choice of the likelihood can alter a radiological or nuclear source assessment, and propose several suited distributions. Finally, two designs depending on the measurements set of the observation error covariance matrix are proposed.
These methods are applied on the case of the detection of Ruthenium 106 of unknown origin in Europe in autumn 2017. We present a posteriori distributions meant to identify the origin of the release, to assess the source term, to quantify the uncertainties associated to the observations and the modelling, as well as densities of the weights of the perturbed ensemble.
Origine : Fichiers éditeurs autorisés sur une archive ouverte
