Quantifying uncertainty in data-driven solar physics simulations

Quantifying uncertainty in data-driven solar physics simulations #

Eric Hall, Karen Meyer

11:50 Tuesday in 4Q07.

Part of the Geophysics and climate session.

Abstract #

Reliable and accurate predictions of solar eruptions are vital to mitigating the consequences of severe space weather, which is included in the UK National Risk Register. Events that lead to solar eruptions are challenging to measure directly. Solar physicists rely on simulations of mathematical models, such as nonlinear force-free field (NLFFF) models, to understand the evolution of the 3D coronal magnetic configuration for a solar active region. NLFFF models are data-driven (assimilate observed magnetograms) and are used to predict time series involved in identifying possibly eruptive regions. However, dependence on initial data, assimilation timescales, measurement errors, and aleatory all contribute to uncertainty in NLFFF predictions. This talk will report on recent progress on quantifying uncertainty in NLFFF predictions arising from misspecification of simulation start times by modelling the error in restart ensembles.