Automated Matching of CMEs with their ICMEs using Parametric Analysis and Dynamic Programming-Based Approach

Matching the Interplanetary Coronal Mass Ejections (ICME) that are being detected at the near-Earth orbit with their counterpart Coronal Mass Ejections (CME) observed at the Sun is one of the most challenging problems in Space Weather since the interplanetary medium, where the CMEs are travelling through, is complex in nature and it is hard to predict its behaviour.
In this work, the OMNI database for the near-Earth space environment is used to look for the ICMEs, and the SOHO/LASCO CME catalogue is used to provide the necessary information of the CMEs. Two methods are applied to identify the signatures of the ICMEs related to the observed CMEs. The first method is a novel method named the parametric approach (PA) in which the ICME signatures are found in the OMNI data based on the semi-empirical model developed by (Gopalswamy et al., 2001) and the criteria mentioned in (Wimmer-Schweingruber et al., 2006). The second method is the dynamic programming-based segmentation (DPS) (Truong et al., 2020), in which the changing points (ChP) within the same time window in the OMNI data are detected and related to the observed CME that probably caused it. Both algorithms are applied on the list of 184 CME-ICME pairs compiled from (Gopalswamy et al., 2010) and (Michałek et al., 2004) to test how well they match the LASCO C2 CME onset time with the flux rope detection of their counterpart ICMEs at 1 AU. It is found that the PA approach has performed a bit better (MAPE ~ 40%) than the ChP approach (MAPE ~ 53%). 
The aim of this work is to provide an algorithm that prepares a reliable list of CME-ICME pairs automatically to be used as a training set for machine learning applications.