Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The EUropean Heliospheric FORecasting Information Asset EUHFORIA (Pomoell, J. and Poedts, S., 2018) is a space weather forecasting tool that models the propagation, evolution, and interaction of solar wind and coronal mass ejections (CMEs) in the interplanetary space. CMEs simulations in EUHFORIA, which are the focus of this study, are initiated by a series of input parameters, each one carrying its own uncertainty. The aim of this dissertation is to perform a parametric study on these parameters in order to understand how their variability affects the final predictions at Earth. Uncertainty Quantification (UQ) is the science that deals with problems related to uncertainties in mathematical models and simulations. UQ is an important area of study, and the ample spectrum of its possible applications led to the development of numerous techniques. For this study, we choose to adopt a Multilevel Monte Carlo (MLMC) technique. This approach combines the simple, yet efficient formulation of a Monte Carlo (MC) method, with the computational advantages of a Multilevel approach. The MLMC estimator is implemented in a julia package called MultilevelEstimator.jl. The program is coupled to an external program that allows exchange of information between the MultilevelEstimator and the software on which the UQ analysis is performed, in this case EUHFORIA. Prior to the analysis itself, we need to adapt the interface code to the code structure of EUHFORIA. This is done through a series of functions, whose purpose is the iterative modification of the input files and storage of the simulation results. A number of simulations are performed, varying one (or more) input CME parameter(s) at a time, for six parameters in total. The predictions of EUHFORIA are tested in terms of minimum Bz and Time Of Arrival (TOA), which are taken as two independent quantities of interest (QoIs). The MLMC analysis is conducted by studying the statistical properties of the QoI, namely, its mean and variance. The results of the analysis are then used to make qualitative observations that aim to a better understanding of the uncertainties carried out by each input CME parameter and their influence at the final EUHFORIA predictions.