Predicting CME Arrivals with Heliospheric Imagers from L5: A Data Assimilation Approach

TL;DR

The paper investigates enhancing CME arrival time and speed predictions by assimilating heliospheric imager data from a L5 vantage into the ELEvoHI framework. It compares two CME-direction inputs (Fixed-Phi Fitting from HI vs Graduated Cylindrical Shell fits from coronagraphs) and assesses how progressively longer HI time-elongation tracks, derived by multiple human trackers, affect forecast accuracy. Key findings show that including HI data out to at least $35^\circ$ elongation substantially improves arrival predictions, with notable gains when reducing forecast lead times from 100 h to 50 h; however, real-time HI data quality currently limits operational use, underscoring the value of future Vigil data and automated tracking approaches. The study also highlights the benefits of deterministic ensemble representations and discusses prospects for real-time data assimilation with Vigil, PUNCH, and related missions to advance space weather forecasting.

Abstract

The Solar TErrestrial RElations Observatory (STEREO) mission has laid a foundation for advancing real-time space weather forecasting by enabling the evaluation of heliospheric imager (HI) data for predicting coronal mass ejection (CME) arrivals at Earth. This study employs the ELEvoHI model to assess how incorporating STEREO/HI data from the Lagrange 5 (L5) perspective can enhance prediction accuracy for CME arrival times and speeds. Our investigation, preparing for the upcoming ESA Vigil mission, explores whether the progressive incorporation of HI data in real-time enhances forecasting accuracy. The role of human tracking variability is evaluated by comparing predictions based on observations by three different scientists, highlighting the influence of manual biases on forecasting outcomes. Furthermore, the study examines the efficacy of deriving CME propagation directions using HI-specific methods versus coronagraph-based techniques, emphasising the trade-offs in prediction accuracy. Our results demonstrate the potential of HI data to significantly improve operational space weather forecasting when integrated with other observational platforms, especially when HI data from beyond 35° elongation are used. These findings pave the way for optimising real-time prediction methodologies, providing valuable groundwork for the forthcoming Vigil mission and enhancing preparedness for CME-driven space weather events.

Figures (8)

• Figure 1: Examples of time-elongation plots in the ecliptic plane showing the minimum (upper panel) and maximum (lower panel) amounts of available HI data for event CME15. The mean track obtained by each scientist is overlaid.
• Figure 2: Panels a) and b) show snapshots of the CME that occurred on 9 October 2021 in (a) STEREO-A/COR2 and (b) SOHO/LASCO-C2 FOV view. Panels c) and d) show the CME fitted with the GCS model, with the white wireframe structure generated from the fit overlaid on the respective coronagraph images.
• Figure 3: DBMfitting applied to two different track lengths of the same CME. The black dots represent the CME kinematics derived using ELCon based on the direction calculated using FPF. The coloured lines are the different DBMfits corresponding to several possible combinations of solar wind speed (colour-coded) and drag parameter ($\Gamma$). Although both tracks are gained from the same event and the direction output from FPF ($\phi$) differs only by $5^\circ$, the resulting CME kinematics diverge, leading to a different ambient solar wind speed derived by DBMfitting and subsequently to a different ELEvoHI arrival prediction.
• Figure 4: Difference of modelled and in situ detected arrival times, $\Delta t$, based on scientist tracking (S1, S2, S3), maximum elongation (track length) used and method to derive the direction of motion (FPF and GCS). Positive (negative) $\Delta t$ means the arrival was predicted too late (too early).
• Figure 5: Standard deviation of the reconstructed CME apex in units of solar radii, shown as a function of elongation for all events and averaged over the tracks produced independently by the three scientists. Each coloured curve represents one CME, while the black line shows the mean standard deviation across all events. The grey shaded area denotes $\pm 1$ standard deviation. For the conversion from elongation to heliocentric distance ELCon based on the FPF-direction was used.