Stochastic algorithm for estimation of the model's unknown parameters via Bayesian inference

We have applied the methodology combining Bayesian inference with Markov chain Monte Carlo (MCMC) algorithms to the problem of the atmospheric contaminant source localization. The algorithms input data are the on-line arriving information about concentration of given substance registered by sensors' network. A fast-running Gaussian plume dispersion model is adopted as the forward model in the Bayesian inference approach to achieve rapid-response event reconstructions and to benchmark the proposed algorithms. We examined different version of the MCMC in effectiveness to estimate the probabilistic distributions of atmospheric release parameters by scanning 5-dimensional parameters' space. As the results we obtained the probability distributions of a source coordinates and dispersion coefficients which we compared with the values assumed in creation of the sensors' synthetic data. The annealing and burn-in procedures were implemented to assure a robust and efficient parameter-space scans.