Lagrangian dispersion modeling in a horizontally homogeneous meteorology for efficient data assimilation design

Atmospheric dispersion models have become a standard support tool in emergency preparedness and response systems that assess the impact on health and environment of a hazardous plume in the atmosphere. Especially in the near-field range, the evolution of a plume can be challenging to predict due to a complex site geometry. A wide range of dispersion models exists, from simple Gaussian models to the more complex computational fluid dynamics-based models. The models mainly differ in the amount of detail by which terrain configurations (buildings, vegetation, etc.) and meteorology (wind field, turbulent diffusivity, etc.) are represented. First-order Lagrangian stochastic (LS) dispersion models offer a good trade-off between improved physical descriptions and computational complexity, notably in the near-field range. This is a desirable model property for use in emergency response systems, which is the framework that the current work aims to contribute to. Therefore, the focus lies on the Langevin model, a widely used first-order LS model. Model inversion is an indispensable tool in emergency situations since it allows to estimate sensitive model parameters from tracer measurements if no other data are available. In case of the Langevin model, however, this requires dealing with a six-dimensional phase space, which is computationally burdensome. An efficient data assimilation method for model inversion of the Langevin model in horizontally homogeneous meteorological conditions is presented. Moreover, dispersion models, which are coupled to an ambient-gamma dose rate model, are validated in the near-field range with routine Ar-41 releases from a nuclear research reactor, which is of particular interest for radiological emergency events.
The research presented in this dissertation has demonstrated that the Langevin model is capable of well-predicting near-field range concentrations in an open field provided that external variability is properly included. A minimum requirement for the latter are accurate measurements of wind direction and the variability thereof. Further, the use of a data assimilation method is highly recommended to optimize the model parameters. The proposed path-integral based (PI) estimator is useful for this purpose since it imposes less restrictions on the choice of optimization parameters than a 3D kernel smoother. A few points of attention are:
1) dose rate prediction – the lower source strength estimate for SCK CEN is presumed to be correct as it has been motivated by a model calculation, which implies that the bias of a dispersion model, coupled to an ambient-gamma dose rate model, is expected to lie between a factor of 2.5 and four for a stack release. Here, the bias of 2.5 corresponds to the Langevin model with the newly developed forest parameterization. The current research suggests the hypothesis that the discrepancy in the dose rate prediction cannot be explained by an inappropriate terrain parameterization or erroneously estimated meteorological parameters thereof. Thus, it is plausible that the cause needs to be sought outside the field of dispersion modeling.
2) convergence of the Monte Carlo method – the potential of the proposed PI estimator has been demonstrated by its ability to obtain a higher degree of convergence than a 3D kernel smoother for the chosen discretization parameters. Theoretically, the PI estimator needs ten times less particles to gain one digit of accuracy in Mean Integrated Squared Error sense than a 3D kernel smoother. It has been demonstrated, however, that the optimal convergence rate is not always obtained, neither for a 3D kernel smoother, nor for the PI estimator.