Ensemble Prediction of a Halo Coronal Mass Ejection Using Heliospheric Imagers

TL;DR

This study demonstrates that ensemble modeling using heliospheric imagers can significantly improve the accuracy of predicting the arrival time and impact speed of Earth-directed coronal mass ejections, with potential for operational use.

Contribution

It introduces an ensemble approach with ELEvoHI for CME prediction, reducing errors and demonstrating the feasibility of using L1 observations for accurate forecasts.

Findings

Ensemble modeling reduces mean absolute arrival time error from 3.5 to 1.6 hours.

Frequency-based parameter selection improves prediction accuracy.

L1 vantage point is sufficient for observing Earth-directed CMEs.

Abstract

The Solar TErrestrial RElations Observatory (STEREO) and its heliospheric imagers (HI) have provided us the possibility to enhance our understanding of the interplanetary propagation of coronal mass ejections (CMEs). HI-based methods are able to forecast arrival times and speeds at any target and use the advantage of tracing a CME's path of propagation up to 1 AU. In our study we use the ELEvoHI model for CME arrival prediction together with an ensemble approach to derive uncertainties in the modeled arrival time and impact speed. The CME from 3 November 2010 is analyzed by performing 339 model runs that are compared to in situ measurements from lined-up spacecraft MESSENGER and STEREO-B. Remote data from STEREO-B showed the CME as halo event, which is comparable to an HI observer situated at L1 and observing an Earth-directed CME. A promising and easy approach is found by using the frequency distributions of four ELEvoHI output parameters, drag parameter, background solar wind speed, initial distance and speed. In this case study, the most frequent values of these outputs lead to the predictions with the smallest errors. Restricting the ensemble to those runs, we are able to reduce the mean absolute arrival time error from $3.5 \pm 2.6$ h to $1.6 \pm 1.1$ h at 1 AU. Our study suggests that L1 may provide a sufficient vantage point for an Earth-directed CME, when observed by HI, and that ensemble modeling could be a feasible approach to use ELEvoHI operationally.