The EUV Late-Phase: Statistical Results from 15 Years of Solar Dynamics Observatory Observations

Technical Summary: Statistical Analysis of the EUV Late-Phase from 15 Years of SDO Observations

Problem Statement
Since the launch of the Solar Dynamics Observatory (SDO) in 2010, the Extreme Ultraviolet (EUV) Late-Phase (ELP) has been identified as a secondary enhancement in warm coronal emission (specifically Fe XV and Fe XVI) occurring tens of minutes to hours after the main solar flare. While the ELP is recognized as geoeffective—capable of producing delayed ionospheric heating and Total Electron Content (TEC) enhancements—its statistical behaviors, occurrence rates, and physical origins remain incompletely understood. Previous studies have been limited by the reliance on disk-integrated spectral data from the EVE instrument, which suffered from data gaps following the 2014 failure of its MEGS-A channel, and by smaller sample sizes that did not fully span multiple solar cycles. Furthermore, the distinction between ELP events and standard flare evolution, as well as the specific dependencies on flare magnitude and solar cycle phase, required a more comprehensive, spatially resolved investigation.

Methodology
This study presents the most extensive statistical analysis of the ELP to date, utilizing 15 years of continuous observations (2010–2025) from the Atmospheric Imaging Assembly (AIA) onboard SDO. The methodology involved:

• Sample Selection: A dataset of 5,335 isolated solar flares (magnitude C1.0 and above) was constructed. Isolation was defined by the absence of preceding events within 1 hour and subsequent events for at least 3 hours, ensuring unambiguous identification of ELP signatures.
• Data Processing: Lightcurves were generated for four EUV wavelengths (131, 171, 304, and 335 Å) using 500 × 500 arcsec cutouts centered on flare coordinates. Fluxes were background-subtracted using a 1-hour preflare interval and normalized. A Savitzky-Golay filter was applied to reduce noise while preserving structure.
• ELP Identification and Validation: The analysis focused on the Fe XVI (335 Å) channel. Initial candidates were identified based on the presence of a distinct secondary peak in the 335 Å flux occurring after the main flare, with no corresponding enhancement in Soft X-Rays (SXR) or hot EUV lines. A rigorous manual validation step required spatial separation: the late-phase emission had to originate from a distinct loop system overlying or adjacent to the main flare arcade, confirmed via AIA imaging and difference maps.
• Statistical Analysis: Quantitative metrics were derived for 467 validated ELP events, including onset delays, peak-to-peak delays, durations, rise/decay rates, and impulsivity parameters. Pairwise Spearman correlation analyses and Principal Component Analysis (PCA) were performed to explore relationships between flare and ELP properties. A comparative analysis between SDO/AIA and SDO/EVE data (2010–2014) was also conducted to assess instrumental biases.

Key Contributions and Results

• Occurrence Rates: The study identified 467 validated ELP events, yielding an overall occurrence rate of approximately 9%. This rate is lower than previous estimates derived from disk-integrated EVE data (often cited around 10–14%). The discrepancy is attributed to the use of spatially resolved AIA data, which avoids the dilution of ELP signals by unrelated active regions in full-disk measurements, and the inclusion of a broader range of flare magnitudes, including weaker C-class events.
• Solar Cycle and Magnitude Dependence: No significant dependence of ELP occurrence on the solar cycle was found; the relative frequency remained consistent (~10%) across the 15-year period, regardless of solar activity levels. A modest enhancement in occurrence was observed in the low-to-mid M-class range (specifically M3.0–M5.9), reaching ~27%, but the rate decreased for higher X-class magnitudes, likely due to smaller sample sizes and statistical uncertainty.
• Temporal Characteristics:
  • Delays: The mean onset delay (flare end to ELP start) was 19 minutes (median 13 min), and the mean peak-to-peak delay was 88 minutes (median 82 min).
  • Duration: The average ELP duration was 93 minutes, significantly longer than the main flare duration (mean 23 min). The ELP duration exceeded the flare duration in 98% of events.
  • Morphology: The ELP lightcurves are typically rise-dominated, with the rise phase accounting for ~65% of the total duration. The skew parameter (rise time/decay time) averaged 2.36.
• Correlations and Scaling:
  • Rates: A strong positive correlation was found between ELP rise and decay rates ($\rho = 0.76$), following a power-law relation ($R_{decay} \propto R_{rise}^{0.92}$).
  • Impulsivity: A moderate positive correlation ($\rho = 0.61$) exists between the impulsivity of the main flare and the ELP, suggesting a link in energy release mechanisms, though the ELP impulsivity increases sub-linearly with flare impulsivity.
  • Independence: Crucially, a comprehensive pairwise analysis revealed no significant correlation ($|\rho| < 0.3$) between the temporal parameters of the main flare (duration, rise/decay times) and the ELP. This indicates that while the phases are energetically linked, their temporal evolutions are largely independent.
• Principal Component Analysis (PCA): PCA identified three semi-independent axes of variability explaining ~60% of the total variance: (1) ELP temporal scale (duration, delays), (2) impulsive heating characteristics (ELP impulsivity, rise rate), and (3) the relative prominence of flare versus late-phase intensity measures. No single parameter dominated the dataset, suggesting the ELP is governed by multiple, partially independent physical processes.
• Instrumental Comparison: A comparison of AIA and EVE data (2010–2014) showed ~80% agreement. Disagreements often arose because EVE's disk-integrated nature obscured ELP signals from multiple active regions, while AIA's spatial resolution allowed for robust detection of isolated events. Conversely, AIA sometimes missed ELPs that extended beyond its 500 × 500 arcsec cutout but were visible in EVE.

Significance and Claims
The authors claim this study provides the first systematic, long-term characterization of the EUV late-phase using exclusively spatially resolved diagnostics. The results establish Fe XVI (335 Å) as a focal diagnostic for the ELP, offering a robust framework for understanding its temporal behavior and occurrence statistics.

The paper emphasizes that the ELP is a distinct phase of solar flare evolution, temporally decoupled from the main flare despite sharing a magnetic environment. The findings highlight that the ELP's occurrence is not strictly tied to the solar cycle or flare magnitude, challenging previous assumptions of strong dependencies. The identified scaling relations between rise/decay rates and impulsivities suggest that the underlying physical drivers—whether continued magnetic reconnection or the cooling of large, overlying loops—evolve coherently within individual events.

The study concludes that these statistical results and derived parameters serve as a benchmark for future space weather models, particularly for predicting delayed ionospheric responses. However, the authors modestly note that the analysis relies on relative fluxes and lacks direct magnetic field diagnostics; thus, while the study clarifies the behavior of the ELP, the definitive physical mechanism (reconnection vs. cooling) for specific events requires future work incorporating vector magnetograms and absolute irradiance calibration.