pith. machine review for the scientific record. sign in

arxiv: 2604.10835 · v1 · submitted 2026-04-12 · 🌌 astro-ph.SR

Recognition: unknown

Improving Solar Flare Soft X-ray Classification With FOXES: A Framework For Operational X-ray Emission Synthesis

Griffin T. Goodwin , Alison J. March , Jayant Biradar , Christoph Schirninger , Robert Jarolim , Angelos Vourlidas , Viacheslav M. Sadykov , Lorien Pratt
Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:09 UTC · model grok-4.3

classification 🌌 astro-ph.SR
keywords solar flaressoft X-rayextreme ultravioletVision Transformerflare classificationspace weathersolar irradianceFOXES
0
0 comments X

The pith

FOXES translates EUV images into spatially resolved soft X-ray flux predictions to fix GOES classification limits.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

GOES soft X-ray measurements classify solar flares but provide no location information and are restricted to Earth's orbit, causing errors during high activity and limiting forecasts elsewhere in the heliosphere. FOXES applies a Vision Transformer to EUV images to output both a total 1-8 Å SXR flux value and per-patch contributions that show where the emission originates. Trained on more than 3200 hours of data, the model reaches a translational mean absolute error of 0.051 dex on integrated flux and separates background emission during flaring and quiet periods. If correct, this supplies spatially aware flare data usable from any viewpoint and improves inputs to space-weather models that rely on flare class and light curves.

Core claim

FOXES is a Vision Transformer framework that converts spatially resolved EUV observations into global 1-8 Å SXR irradiance predictions together with per-patch flux maps. After training, validation, and testing on over 3200 hours of observations, it reports a translational mean absolute error of 0.051 dex for the integrated flux and demonstrates the ability to isolate solar background SXR contributions in both flaring and non-flaring intervals, thereby enabling EUV-based flare detection from locations other than Earth's line of sight.

What carries the argument

Vision Transformer that produces both a single global 1-8 Å SXR flux value and a set of per-patch flux contributions from input EUV images.

If this is right

  • Flare classification can assign correct strengths even when multiple events occur at the same time.
  • Flare catalogs can be built from any EUV-capable viewpoint instead of only Earth orbit.
  • Space-weather forecasts that use flare class and light curves can be driven by data from multiple spacecraft positions.
  • Real-time background subtraction becomes possible for both flaring and quiet-Sun intervals.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The per-patch maps could be compared against simultaneous magnetograms or hard X-ray images to test whether the model attributes emission to the correct magnetic structures.
  • Retraining the same architecture on data from solar cycle minima versus maxima would reveal whether the learned mapping changes with overall activity level.
  • The framework could be extended to predict additional wavelength bands, supplying consistent multi-wavelength inputs for energetic-particle and CME models.
  • Deployment on operational EUV imagers would allow continuous multiviewpoint SXR synthesis without requiring new hardware.

Load-bearing premise

The statistical mapping from EUV images to SXR flux learned on Earth-orbit data will remain accurate for new flares, different activity levels, and observations taken from other points in the heliosphere.

What would settle it

Direct comparison of FOXES-predicted total SXR flux against simultaneous measurements from a spacecraft at a non-Earth heliospheric longitude during a well-observed flare that shows an error substantially larger than 0.051 dex.

Figures

Figures reproduced from arXiv: 2604.10835 by Alison J. March, Angelos Vourlidas, Christoph Schirninger, Griffin T. Goodwin, Jayant Biradar, Lorien Pratt, Robert Jarolim, Viacheslav M. Sadykov.

Figure 1
Figure 1. Figure 1: An overview of the data splits for training, validation, and testing. The top four panels show the breakdown of the number of data points for a given SXR level as a function of time. The colors illustrate the split the data belongs to: training (blue), validation (green), testing (red). The last panel shows the timeline for each split. Please note that even though there are overlapping bars between the dat… view at source ↗
Figure 2
Figure 2. Figure 2: An overview of the FOXES vision transformer architecture. A [7 × 512 × 512] array representing a stack of EUV images at a given time is first segmented into 8 × 8 pixel patches. These patches and their positions are flattened and fed into the transformer encoder block. Within this block exist layers for normalization, self-attention, and a traditional multilayer perceptron (MLP). The weights are then passe… view at source ↗
Figure 3
Figure 3. Figure 3: (Top Panel) A 2D hexagonal binned plot comparing the baseline model’s integrated translations to the GOES integrated SXR flux (ground truth). A perfect translation (red line) is overlaid on top. (Bottom Panel) Same for the FOXES model’s integrated translations (by summing up the flux from each patch) to the GOES integrated SXR flux (ground truth). A perfect translation line is also overlaid on top. The bin… view at source ↗
Figure 4
Figure 4. Figure 4: An application of FOXES to AIA data sampled with 1-minute cadence across 4 time periods in August 2023. Each window is 14 hours long. The orange line highlights the ground truth SXR flux from GOES, while the purple line is the FOXES translation. Note: These translations are still being made at the given observation time. We are not forecasting ahead or using previous data to inform the model [PITH_FULL_IM… view at source ↗
Figure 5
Figure 5. Figure 5: SDO/AIA 131˚A snapshots of the M-class flare that occurred on April 11th, 2024. on the eastern limb and underpredict on the western. However, this may reflect a bias within the samples themselves rather than a systematic bias of FOXES. Overall, the results suggest that our previous intuition is plausible. The event from [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Normalized flux-weighted maps for mean absolute error (Left) and mean bias error (Right). Each map was calculated using the equation defined in 2. Patches above and below ≈ ±56◦ in disk latitude were removed as they are not relevant to this investigation, and tend to be noisy due to their low flux values. These maps can be interpreted as how the integrated SXR error of FOXES varies with the predicted spati… view at source ↗
Figure 7
Figure 7. Figure 7: On the left of both (a) and (b), we display a stack of AIA images (131˚A, 171˚A, 304˚A) at a given time, along with the active regions (high flux regions) identified by FOXES. We mark flaring locations identified by FOXES (stars) and by the Heliophysics Event Knowledge database’s SDO/AIA detection algorithm (squares). The integrated flux of the patches in each individual region track is shown as well. On t… view at source ↗
Figure 8
Figure 8. Figure 8: The mean absolute error (calculated in log10 space) across flare classes (and overall) for FOXES applied to ablated versions of the test dataset, where individual EUV channels (and one triple-channel combination) were augmented with Gaussian noise. integrated SXR flux. However, the overlap between their high-cadence data and ours is currently limited, and a rigorous comparison would require substantial add… view at source ↗
read the original abstract

The Geostationary Operational Environmental Satellite (GOES) solar soft X-ray (SXR) irradiance in the 1-8{\AA} wavelength range is a long-standing measure of solar activity, used to define the classification of flare strengths. As a result, the flare class, along with the SXR light curves, are routinely used as a primary input for forecasting properties of space weather drivers, from coronal mass ejection speed to energetic particle output. However, the GOES SXR irradiance lacks spatial information, leading to known classification errors, such as misattributed flare locations during periods of high activity. Moreover, GOES only provides observations from Earth's orbit, hindering forecasting for other places in the heliosphere. Motivated by these limitations, we introduce the Framework for Operational X-ray Emission Synthesis (FOXES), a Vision Transformer-based approach for translating Extreme Ultraviolet (EUV) spatially-resolved observations into SXR irradiance predictions. The model produces two outputs: (1) a global 1-8{\AA} SXR flux prediction and (2) per-patch flux contributions, which offer a spatially-resolved interpretation of where the model attributes SXR emission. Trained, validated, and tested on over 3200 hours of observations, FOXES has demonstrated a translational mean absolute error of 0.051 dex for integrated SXR measurements. FOXES has also shown promise in dissecting the solar background SXR flux during flaring and non-flaring events. Overall, this model paves the way for EUV-based spatially-resolved flare detection to be extended beyond Earth's line of sight. Such capabilities could lead to a more comprehensive flare catalog and enable a true multiviewpoint monitoring of solar activity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces FOXES, a Vision Transformer framework that maps spatially resolved EUV observations to soft X-ray (SXR) irradiance predictions in the 1-8 Å band. It generates both a global integrated SXR flux and per-patch attribution maps for interpretability. Trained, validated, and tested on over 3200 hours of joint EUV/SXR data, the model reports a translational MAE of 0.051 dex on held-out integrated measurements and explores separation of background SXR flux in flaring versus non-flaring intervals. The work aims to mitigate GOES classification errors from lack of spatial information and to enable SXR-based monitoring from non-Earth heliospheric viewpoints.

Significance. If the learned EUV-to-SXR mapping generalizes, FOXES could supply spatially resolved SXR estimates that improve flare location attribution during high-activity periods and support space-weather inputs from arbitrary solar-system locations. The per-patch output provides useful physical interpretability. The current significance is limited because all quantitative results remain within the Earth-orbit training distribution, leaving the multiviewpoint extension prospective rather than demonstrated.

major comments (2)
  1. [Abstract] Abstract: the claim that FOXES 'paves the way for EUV-based spatially-resolved flare detection to be extended beyond Earth's line of sight' and enables 'true multiviewpoint monitoring' is unsupported. No quantitative results are reported on STEREO, Solar Orbiter, or any off-Earth viewpoint data; all MAE and attribution results are obtained on held-out samples drawn from the identical Earth-orbit EUV/SXR joint distribution used for training.
  2. [Results/Evaluation] Results/Evaluation section: the headline performance figure of 0.051 dex MAE is presented without error bars, explicit train/validation/test split descriptions, baseline comparisons (e.g., against direct EUV integration or prior irradiance-mapping methods), or stratification by solar-cycle phase and flare frequency. These omissions make it impossible to judge whether the reported accuracy constitutes a genuine improvement or is robust outside the training distribution.
minor comments (1)
  1. [Abstract] Abstract: the phrase 'translational mean absolute error' is introduced without a definition or reference to how it is computed from the model's outputs.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. We address each major comment below and have revised the manuscript accordingly to improve clarity and rigor.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that FOXES 'paves the way for EUV-based spatially-resolved flare detection to be extended beyond Earth's line of sight' and enables 'true multiviewpoint monitoring' is unsupported. No quantitative results are reported on STEREO, Solar Orbiter, or any off-Earth viewpoint data; all MAE and attribution results are obtained on held-out samples drawn from the identical Earth-orbit EUV/SXR joint distribution used for training.

    Authors: We agree that the original abstract language presents the multiviewpoint capability as more immediate than the current results support. The FOXES architecture is viewpoint-agnostic by design, as it ingests spatially resolved EUV images and produces integrated SXR flux plus per-patch attributions without embedding Earth-specific assumptions; EUV data from STEREO and Solar Orbiter already exist and could be used for future testing. Nevertheless, because all reported metrics derive from the Earth-orbit joint distribution, the extension remains prospective. In the revised version we will replace the abstract sentence with: 'FOXES provides a general framework for EUV-to-SXR mapping that can be applied to non-Earth viewpoints once co-observed datasets become available.' We will also add a dedicated paragraph in the Discussion section outlining the data requirements and any necessary domain-adaptation steps for such extensions. revision: yes

  2. Referee: [Results/Evaluation] Results/Evaluation section: the headline performance figure of 0.051 dex MAE is presented without error bars, explicit train/validation/test split descriptions, baseline comparisons (e.g., against direct EUV integration or prior irradiance-mapping methods), or stratification by solar-cycle phase and flare frequency. These omissions make it impossible to judge whether the reported accuracy constitutes a genuine improvement or is robust outside the training distribution.

    Authors: We accept that the evaluation section would benefit from greater transparency. Although the manuscript states the total data volume (>3200 hours) and the use of held-out test samples, we will expand it to include: (i) error bars on the 0.051 dex MAE obtained via bootstrap resampling across the test set; (ii) a clear description of the temporal train/validation/test partitioning, including the fraction of data from each solar-cycle phase and confirmation that no temporal overlap exists between splits; (iii) quantitative baseline comparisons, specifically a linear combination of EUV channel intensities calibrated to SXR and any previously published EUV-to-SXR mapping approaches; and (iv) performance tables stratified by solar-cycle phase (rising, maximum, declining) and by flare activity level (quiet Sun vs. periods containing C-class or larger flares). These additions will allow readers to assess both improvement over baselines and robustness across conditions. revision: yes

Circularity Check

0 steps flagged

No circularity: performance metric on held-out data is independent of inputs

full rationale

The paper trains a Vision Transformer on paired EUV/SXR observations and reports a translational MAE of 0.051 dex on held-out test data drawn from the same Earth-orbit distribution. This is a standard empirical evaluation, not a quantity obtained by fitting a parameter to a subset and then renaming the fit as a prediction, nor by any self-referential equation or self-citation chain. No mathematical derivation exists that reduces to its own inputs by construction. The untested generalization to other viewpoints is an assumption about future applicability, not a circularity in the reported results.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on learned neural network weights from the training corpus and the domain assumption that EUV and SXR are sufficiently correlated for the mapping to be learned and generalized.

free parameters (1)
  • Vision Transformer weights
    All model parameters are fitted during training on the 3200-hour dataset.
axioms (1)
  • domain assumption EUV observations contain sufficient information to predict SXR irradiance
    The model is built on the premise that a learnable statistical mapping exists between the two wavelength regimes.

pith-pipeline@v0.9.0 · 5651 in / 1157 out tokens · 44714 ms · 2026-05-10T15:09:44.539046+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references · 29 canonical work pages · 1 internal anchor

  1. [1]

    K., et al

    Ahmadzadeh, A., Aydin, B., Georgoulis, M. K., et al. 2021, The Astrophysical Journal Supplement Series, 254, 23, doi: 10.3847/1538-4365/abec88

    work page doi:10.3847/1538-4365/abec88 2021

  2. [2]

    A., Martens, P

    Angryk, R. A., Martens, P. C., Aydin, B., et al. 2020, Scientific Data, 7, 227, doi: 10.1038/s41597-020-0548-x Asensio Ramos, A., Cheung, M. C. M., Chifu, I., & Gafeira, R. 2023, Living Reviews in Solar Physics, 20, 4, doi: 10.1007/s41116-023-00038-x

    work page doi:10.1038/s41597-020-0548-x 2020

  3. [3]

    D., Schrijver, C

    Barnes, G., Leka, K. D., Schrijver, C. J., et al. 2016, The Astrophysical Journal, 829, 89, doi: 10.3847/0004-637X/829/2/89

    work page doi:10.3847/0004-637x/829/2/89 2016

  4. [4]

    2025, The Astrophysical Journal Supplement Series, 278, 9, doi: 10.3847/1538-4365/adc731

    Berretti, M., Mestici, S., Giovannelli, L., et al. 2025, The Astrophysical Journal Supplement Series, 278, 9, doi: 10.3847/1538-4365/adc731

    work page doi:10.3847/1538-4365/adc731 2025

  5. [5]

    G., Sun, X., Hoeksema, J

    Bobra, M. G., Sun, X., Hoeksema, J. T., et al. 2014, Solar Physics, 289, 3549, doi: 10.1007/s11207-014-0529-3

    work page doi:10.1007/s11207-014-0529-3 2014

  6. [6]

    Space Weather 17(8), 1166–1207 (2019) https://doi.org/10.1029/2018SW002061

    Camporeale, E. 2019, Space Weather, 17, 1166, doi: 10.1029/2018SW002061

    work page doi:10.1029/2018sw002061 2019

  7. [7]

    Caspi, A., Krucker, S., & Lin, R. P. 2014, The Astrophysical Journal, 781, 43, doi: 10.1088/0004-637X/781/1/43

    work page doi:10.1088/0004-637x/781/1/43 2014

  8. [8]

    C., Woods, T

    Chamberlin, P. C., Woods, T. N., Eparvier, F. G., & Jones, A. R. 2009, in Solar physics and space weather instrumentation III, Vol. 7438, SPIE, 11–20

    2009

  9. [9]

    M., Belov, A

    Chertok, I. M., Belov, A. V., & Grechnev, V. V. 2015, Solar Physics, 290, 1947, doi: 10.1007/s11207-015-0738-4 Dos Santos, L. F. G., Bose, S., Salvatelli, V., et al. 2021, Astronomy & Astrophysics, 648, A53, doi: 10.1051/0004-6361/202040051

    work page doi:10.1007/s11207-015-0738-4 2015

  10. [10]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., et al. 2020, arXiv preprint arXiv:2010.11929

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  11. [11]

    K., Bloomfield, D

    Georgoulis, M. K., Bloomfield, D. S., Piana, M., et al. 2021, Journal of Space Weather and Space Climate, 11, 39, doi: 10.1051/swsc/2021023

    work page doi:10.1051/swsc/2021023 2021

  12. [12]

    T., March, A., Biradar, J., et al

    Goodwin, G. T., March, A., Biradar, J., et al. 2026a, FOXES-Data (Revision 6b434f8), Hugging Face, doi: 10.57967/hf/7737

    work page doi:10.57967/hf/7737

  13. [13]

    T., March, A., Biradar, J., et al

    Goodwin, G. T., March, A., Biradar, J., et al. 2026b, FOXES (Revision 0affdc2), Hugging Face, doi: 10.57967/hf/8234

    work page doi:10.57967/hf/8234

  14. [14]

    T., Sadykov, V

    Goodwin, G. T., Sadykov, V. M., & Martens, P. C. 2024, The Astrophysical Journal, 964, 163, doi: 10.3847/1538-4357/ad276c

    work page doi:10.3847/1538-4357/ad276c 2024

  15. [15]

    2018, The Astrophysical Journal, 856, 7, doi: 10.3847/1538-4357/aaae00

    Huang, X., Wang, H., Xu, L., et al. 2018, The Astrophysical Journal, 856, 7, doi: 10.3847/1538-4357/aaae00

    work page doi:10.3847/1538-4357/aaae00 2018

  16. [16]

    2025, Nature Communications, 16, 3157

    Podladchikova, T. 2025, Nature Communications, 16, 3157

    2025

  17. [17]

    2023, International Journal of Advanced Computer Science and Applications, 14, doi: 10.14569/IJACSA.2023.0140830

    Kameswari, C., J, K., Reddy, T., et al. 2023, International Journal of Advanced Computer Science and Applications, 14, doi: 10.14569/IJACSA.2023.0140830

    work page doi:10.14569/ijacsa.2023.0140830 2023

  18. [18]

    doi: 10.1145/3505244

    Khan, S., Naseer, M., Hayat, M., et al. 2022, ACM Comput. Surv., 54, 200:1, doi: 10.1145/3505244

    work page doi:10.1145/3505244 2022

  19. [19]

    D., Dissauer, K., Barnes, G., & Wagner, E

    Leka, K. D., Dissauer, K., Barnes, G., & Wagner, E. L. 2023, The Astrophysical Journal, 942, 84, doi: 10.3847/1538-4357/ac9c04

    work page doi:10.3847/1538-4357/ac9c04 2023

  20. [20]

    D., Park, S.-H., Kusano, K., et al

    Leka, K. D., Park, S.-H., Kusano, K., et al. 2019, The Astrophysical Journal Supplement Series, 243, 36, doi: 10.3847/1538-4365/ab2e12

    work page doi:10.3847/1538-4365/ab2e12 2019

  21. [21]

    R., Title, A

    Lemen, J., Title, A., Akin, D., et al. 2012, solphys, 275, 17, doi: 10.1007/s11207-011-9776-8

    work page doi:10.1007/s11207-011-9776-8 2012

  22. [22]

    2025, 11-vision-transformer.ipynb, https: //github.com/Lightning-AI/tutorials/blob/ 69112e6fe73b50d159c4b7add6ddea412b458691/ .notebooks/course UvA-DL/11-vision- transformer.ipynb

    Lippe, P. 2025, 11-vision-transformer.ipynb, https: //github.com/Lightning-AI/tutorials/blob/ 69112e6fe73b50d159c4b7add6ddea412b458691/ .notebooks/course UvA-DL/11-vision- transformer.ipynb

    2025

  23. [23]

    Machol, J., Viereck, R., Peck, C., & Iii, J. M. 2022,

    2022

  24. [24]

    2007 in Solar Storm Threat Analysis

    Marusek, J. 2007 in Solar Storm Threat Analysis. https://www.semanticscholar.org/paper/Solar- Storm-Threat-Analysis-Marusek/ 73420378d1a42d74ebbfe63eaff697a32b401577

    2007

  25. [25]

    2019, SN Applied Sciences, 1, 49

    Natras, R., Horozovic, D., & Mulic, M. 2019, SN Applied Sciences, 1, 49

    2019

  26. [26]

    2018, The Astrophysical Journal, 858, 113, doi: 10.3847/1538-4357/aab9a7

    Ishii, M. 2018, The Astrophysical Journal, 858, 113, doi: 10.3847/1538-4357/aab9a7

    work page doi:10.3847/1538-4357/aab9a7 2018

  27. [27]

    2013, Solar Physics, 288, 241 Padial Doble, J

    Nitta, N., Aschwanden, M., Boerner, P., et al. 2013, Solar Physics, 288, 241 Padial Doble, J. R., & Holley-Bockelmann, K. 2025, The Astrophysical Journal Supplement Series, 278, 14, doi: 10.3847/1538-4365/adbe35

    work page doi:10.3847/1538-4365/adbe35 2013

  28. [28]

    D., Thompson, B

    Pesnell, W., Thompson, B., & Chamberlin, P. 2012, solphys, 275, 3, doi: 10.1007/s11207-011-9841-3

    work page doi:10.1007/s11207-011-9841-3 2012

  29. [29]

    2012, Astronomy & Astrophysics, 547, A25

    Bradshaw, S. 2012, Astronomy & Astrophysics, 547, A25

    2012

  30. [30]

    R., & Forbes , T

    Priest, E. R., & Forbes, T. G. 2002, A&A Rv, 10, 313, doi: 10.1007/s001590100013 20

    work page doi:10.1007/s001590100013 2002

  31. [31]

    2025, Surya: Foundation Model for Heliophysics, arXiv, doi: 10.48550/arXiv.2508.14112

    Roy, S., Schmude, J., Lal, R., et al. 2025, Surya: Foundation Model for Heliophysics, arXiv, doi: 10.48550/arXiv.2508.14112

    work page doi:10.48550/arxiv.2508.14112 2025

  32. [32]

    M., et al

    Schirninger, C., Jarolim, R., Veronig, A. M., et al. 2025, A&A, 703, A152, doi: 10.1051/0004-6361/202556401

    work page doi:10.1051/0004-6361/202556401 2025

  33. [33]

    Schrijver, C. J. 2009, Advances in Space Research, 43, 739, doi: 10.1016/j.asr.2008.11.004 The SunPy Community, Barnes, W. T., Bobra, M. G., et al. 2020, ApJ, 890, 68, doi: 10.3847/1538-4357/ab4f7a

    work page doi:10.1016/j.asr.2008.11.004 2009

  34. [34]

    Thiemann, E. M. B., Chamberlin, P. C., Eparvier, F. G., & Epp, L. 2018, Solar Physics, 293, 19, doi: 10.1007/s11207-018-1244-2 Van der Sande, K., Flyer, N., Berger, T. E., &

    work page doi:10.1007/s11207-018-1244-2 2018

  35. [35]

    2022, Frontiers in Astronomy and Space Sciences, 9, 1031211

    Gagnon, R. 2022, Frontiers in Astronomy and Space Sciences, 9, 1031211

    2022

  36. [36]

    2017, Advances in neural information processing systems, 30

    Vaswani, A., Shazeer, N., Parmar, N., et al. 2017, Advances in neural information processing systems, 30

    2017

  37. [37]

    M., & Angryk, R

    Sadykov, V. M., & Angryk, R. A. 2025, The Astrophysical Journal Supplement Series, 277, 60, doi: 10.3847/1538-4365/adb9e3

    work page doi:10.3847/1538-4365/adb9e3 2025

  38. [38]

    N., Eden, T., Eparvier, F

    Woods, T. N., Eden, T., Eparvier, F. G., et al. 2024, Journal of Geophysical Research: Space Physics, 129, e2024JA032925, doi: https://doi.org/10.1029/2024JA032925

    work page doi:10.1029/2024ja032925 2024