Speaker
Description
Coronal mass ejections (CMEs) are one of the most relevant cause of geomagnetic storms. Early determination of their geo-effectiveness is a key asset to provide a possible alert on ground. The first crucial step of this process is to detect and characterise the CMEs in terms of size, kinematics (velocity, acceleration, direction of travel), mass and energy and this is possible by analysing series of coronal images in visible light.
In this investigation we processed sequences of LASCO C2 and C3 images by applying two novel analysis techniques having the goal to identify and track the evolution of the CME front with high accuracy and reliability.
The first approach is based on the Level-Set method, first introduced by Osher and Sethian [1988] and successfully applied to many propagation problems over the years. The proposed processing procedure begins by filtering the input images to reduce the noise, then by applying some rescaling techniques to highlight the CME area. Eventually, a Semi-Lagrangian scheme for the Chan-Vese model is used to segment the resulting images.
The second approach is based on the analysis of the Mutual Information (MI), introduced by Viola and Wells [1997] as a general measure of statistical dependence in imaging applications. By computing patch-wise MI on frame differences, we highlight structural changes without strong model assumptions. Regions of low MI indicate areas where the image content changes significantly between consecutive frames, marking the onset and evolution of CME structures.
Both the two methods provide compact and interpretable maps of CME region and propagation directions. Preliminary results on LASCO data demonstrate their potential for systematic CME characterisation.
