Data assimilation, sensitivity and uncertainty analyses in the Dutch nuclear emergency management system: a pilot study

To investigate the possibilities to improve the Dutch nuclear emergency management system, a pilot study was carried out on a data assimilation method for our atmospheric dispersion model. By means of the data assimilation method, the prediction of potentially contaminated areas in the early and late phases of a nuclear accident can significantly be improved. In the early phase, results of the radiological monitoring network are still sparse and the method focuses on a generic improvement of the model forecast by optimisation of a limited number of input parameters. Prior to the study on data assimilation, sensitivity and uncertainty analysis were performed to identify the most important parameters. For this pilot study, only two parameters were fitted to demonstrate the working of our data assimilation technique. The technique was successfully applied to an inconsistent part of a famous data set giving a reasonable agreement between the observed and modelled results.