Speaker
Description
Coronal Mass Ejections (CMEs) are most commonly detected in white-light observations acquired by space-based coronagraphs. These instruments record a two-dimensional projection of the intrinsically three-dimensional CME structure onto the plane of the sky. As a result, coronagraphic measurements alone cannot unambiguously discriminate between frontside (Earth-directed) and backside CMEs.
The true directionality of CMEs is typically assessed by forecasters who routinely look for low coronal CME signatures in other remote-sensing data, e.g., solar EUV disk images. The key signatures associated with CMEs are, in particular, coronal dimmings, EUV waves, erupting filaments (seen as erupting prominences when observed above the limb), and post-eruption loop arcades. These features provide reliable evidence of CME occurrence and source location.
As part of the activities of the ESA Space Weather Service Network (SWESNET) program and the “Space It Up!” project of the Italian Space Agency, we are developing new algorithms for the real-time analysis of solar EUV disk images. The goal is to automatically detect low coronal features that may be associated with CMEs subsequently observed in coronagraphic images, so as to provide information on the true directionality of such CMEs to be used in the CME Propagation Prediction tool integrated in the SWESNET portal. We are exploring both classical approaches based on the analysis of the differences between consecutive images and computer-vision methods based on machine-learning tools, and we present some preliminary results.
