Non-linear Bayesian CBRN source term estimation

A Bayesian posterior probability density sampling algorithm suitable for use in estimating chemical, biological, radiological or nuclear atmospheric releases of material is proposed. The estimation problem is different from many other fusion problems, in that there are no state evolution equations, the forward model is highly non-linear and the likelihoods are non-Gaussian. The algorithm is able to use stored output from complex atmospheric dispersion models for more rapid update of the posterior from new data without having to re-run the models. The use of differential evolution Monte Carlo allows new samples to rapidly diverge from degenerate sample sets. Results for inferences made in a slightly simplified environment of chemical releases only are presented, demonstrating that the sampling scheme performs adequately despite constraints of a short time span to calculate inferences from complex likelihood calculations. Extensions to wider applications and improvements to the algorithm are discussed.