A Kalman Filter-Based Approach for Online Source-Term Estimation in Accidental Radioactive Dispersion Events

Abstract

In the present work, a online data assimilation approach, based on the Kalman filter algorithm, is proposed for the source term reconstruction in accidental events with dispersion of radioactive agents in air. For this purpose a Gaussian plume model of dispersion in air is embedded in the Kalman filter algorithm to estimate unknown scenario parameters, such as the coordinates and the intensity of the source, on the basis of measurements collected by a mobile sensor. The approach was tested against pseudo-experimental data produced with both the Gaussian plume model and the Lagrangian puff model SCIPUFF. The results show the good capabilities of the proposed approach in retrieving the values of the unknown parameters when (i) one or more release parameters are poorly known and (ii) a sufficient number of experimental measurements describing the evolution of the dispersion process can be collected in a short time by means of mobile sensors. Thanks to its flexibility and computational efficiency, and due to the exploitation of the Kalman filter potentialities through the use of a simplified model of dispersion in air, the proposed approach can constitute a useful tool for the management of emergency scenarios.