Analysis of a Kalman filter based method for on-line estimation of atmospheric dispersion parameters using radiation monitoring data

Martin Drews, Bent Lauritzen, Henrik Madsen

    Research output: Contribution to journalJournal articleResearchpeer-review

    Abstract

    A Kalman filter method is discussed for on-line estimation of radioactive release and atmospheric dispersion from a time series of off-site radiation monitoring data. The method is based on a state space approach, where a stochastic system equation describes the dynamics of the plume model parameters, and the observables are linked to the state variables through a static measurement equation. The method is analysed for three simple state space models using experimental data obtained at a nuclear research reactor. Compared to direct measurements of the atmospheric dispersion, the Kalman filter estimates are found to agree well with the measured parameters, provided that the radiation measurements are spread out in the cross-wind direction. For less optimal detector placement it proves difficult to distinguish variations in the source term and plume height; yet the Kalman filter yields consistent parameter estimates with large associated uncertainties. Improved source term assessment results, when independent estimates of the plume height can be used. Perspectives for using the method in the context of nuclear emergency management are discussed, and possible extensions to the present modelling scheme are outlined, to account for realistic accident scenarios.
    Original languageEnglish
    JournalRadiation Protection Dosimetry
    Volume113
    Issue number1
    Pages (from-to)75-89
    ISSN0144-8420
    DOIs
    Publication statusPublished - 2005

    Fingerprint Dive into the research topics of 'Analysis of a Kalman filter based method for on-line estimation of atmospheric dispersion parameters using radiation monitoring data'. Together they form a unique fingerprint.

    Cite this