Kalman filtration of radiation monitoring data from atmospheric dispersion of radioactive materials

A Kalman filter method using off-site radiation monitoring data is proposed as a tool for on-line estimation of the source term for short-range atmospheric dispersion of radioactive materials. The method is based on the Gaussian plume model, in which the plume parameters including the source term exhibit a ‘random walk’ process. The embedded parameters of the Kalman filter are determined through maximum-likelihood estimation making the filter essentially free of external parameters. The method is tested using both real and simulated radiation monitoring data. For simulated data, the method is shown to retrieve the embedded parameters employed in generating the data and to reconstruct the plume model parameters, including the source term. When tested against experimental radiation monitoring data the method is found accurately to uncover the known source term.