Assimilation of the spatio-temporal distribution of radionuclides in the early phase of a radiation accident

Applying the Bayesian approach, the input parameters were modelled as random quantities, estimated based on available measurements. The method, as well as providing a picture of the radiological situation improved based on measurements, gives an estimate of the atmospheric dispersion parameters whose accuracy improves with increasing number of measurements. The refined parameters can be used to predict the further development on any horizon. The algorithm was tested on a simulated scenario of the release of 41Ar. (orig.)