PhD Defense by Kasper Skjold Tølløse

Title: Prediction of Atmospheric Dispersion on All Scales for Emergency Preparedness

Abstract:
Long-range atmospheric dispersion modelling is carried out at the Danish Meteorological Institute (DMI) to predict concentration and deposition fields of various kinds of hazardous matter such as radioactive gasses and particles, toxic chemicals and smoke, aerosols containing infectious germs, and volcanic ash particles. These simulations are provided as a service to the responsible Danish authorities, thereby enabling early warnings and facilitating the implementation of optimal countermeasures in different emergency situations. The atmospheric dispersion is modelled by using the Danish Emergency Response Model of the Atmosphere (DERMA), which is a Lagrangian puff model designed specifically for long-range dispersion modelling. However, to extend this capability to a shorter range (up to about 50 km from the source), reformulations of essential parts of the model are required. In this PhD thesis, a new hybrid particle-puff approach is developed and implemented in DERMA, enabling the model to simulate short-range atmospheric dispersion more accurately. This new description of turbulent diffusion is evaluated against data from three different tracer experiments to validate both the short-range and long-range capabilities of the model.


In addition, both historical and recent events have demonstrated the necessity for being able to conduct inverse modelling as an operational service. This capability would assist responsible authorities in localizing unknown sources and/or characterizing the temporal development of gas and particle emissions in emergency situations. Therefore, two inverse methods have been developed: The first allows for source localization based on a set of air concentration measurements in cases where the release location is unknown. The second enables estimation of the multi-nuclide source term from a nuclear power plant accident in cases where little or no direct information about the release is available, as it has in fact been the case in historical nuclear accidents. This second method is designed specifically for use in the early stages of an accident, where an improved source term estimate may be crucial for facilitating reliable dispersion predictions.
The developments and findings in this PhD project successfully lay the foundations for new operational tasks at DMI while also constituting important contributions to the research field of dispersion modelling, especially inverse modelling for source term
estimation and localization.

Committee: Jens Hesselbjerg Christensen (NBI); Irene Korsakissok (IRSN); Spyros Andronopoulos (Demokritos)

Supervisors: Eigil Kaas (NBI) & Jens Havskov Sørensen (DMI)
Co-supervisor: Henrik Feddersen (DMI)