Multi-modal encoder-decoder neural network for forecasting solar wind speed at L1

Bharat Kaul; Dattaraj B. Dhuri; Dipankar Das; Gopika SM; Harsh Joon; Shravan M. Hanasoge

REVIEW 3 major objections 5 minor 28 references

Multi-modal encoder-decoder neural network for forecasting solar wind speed at L1

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An encoder-decoder network that pairs prior-rotation solar wind history with EUV coronal images forecasts daily solar wind speed four days ahead with validation RMSEs near 55 km/s and an unseen-data RMSE of 53 km/s.

desk verdict Solid held-out forecasting result, but the headline RMSE gain over prior work is likely an artifact of daily versus hourly averaging; the architecture is still worth engaging. read the letter →

arxiv 2507.17298 v1 pith:J5B7Z236 submitted 2025-07-23 astro-ph.SR astro-ph.IM

Dattaraj B. Dhuri , Shravan M. Hanasoge , Harsh Joon , Gopika SM , Dipankar Das , Bharat Kaul This is my paper · ORCID

classification astro-ph.SRastro-ph.IM

keywords solarwindforecastingencoder-decodernetworkSDO/AIA211Angstromimagery27-daypersistencecoronalholesspaceweathersequence-to-sequencemodelL1in-situmeasurements

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes that the solar-wind speed at L1, measured as a daily average up to four days ahead, can be forecast simultaneously by a two-stream encoder-decoder neural network that learns to 'time-translate' wind patterns from earlier solar rotations and to read present coronal structure from EUV images. The authors train the network in two stages: first on solar-wind history alone (solar wind state, flare magnitudes, sunspot number from the two prior rotations), then with four days of SDO/AIA 211 Angstrom coronal images added. On the validation period 2014-2017 they report RMSEs of 54.7, 58.3, 58.2 and 57.6 km/s and correlations of 0.78, 0.66, 0.64 and 0.63 for one to four days ahead; on unseen data from 2019-2024 the four-day-ahead RMSE is 53.25 km/s with correlation 0.55. They interpret these scores as roughly 20% lower RMSE than the model of Brown et al. (2022), with the caveat that the comparison is only indicative because forecast horizons and validation splits differ. A reliable four-day ambient solar-wind forecast matters because this window is the main lead time for preparing for space-weather effects on satellites and power grids.

What carries the argument

The central object is the multi-modal encoder-decoder network built on a sequence-to-sequence framework: two identical history encoders process 1D time series of seven solar-wind and solar-activity features sampled over eight days around the prior solar rotations; an AIA encoder processes GoogleNet representations of four days of 211 Angstrom EUV images; and a decoder, seeded with the past four days of wind speed with future entries masked to zero, produces a time sequence reaching four days into the future. Hand-encoded timestamps (hour, day, month, year, relative day, solar-cycle phase and cycle number) plus positional encoding give the network temporal and solar-cycle context. The two-stage training is the mechanism that makes the multimodal setup trainable from only roughly 50k samples: Stage 1 learns the history-to-wind translation, and Stage 2 freezes those weights initially, adds the AIA encoder, and refines the whole model.

What would settle it

Evaluate the same encoder-decoder architecture on the hourly-averaged solar wind data and the exact five-fold validation splits of Brown et al. (2022) and Upendran et al. (2020), reporting RMSE and Pearson r for days 1 through 4 at the same hourly cadence; if the day-4 RMSE does not fall below roughly 72 km/s for Brown et al. and 81 km/s for Upendran et al. on those splits, the claimed ~20% and ~29% improvements would not be confirmed.

Watch

Extended reading notes

Core claim

The central claim is that combining two kinds of solar observations in one sequence-to-sequence architecture beats prior single-modality deep-learning forecasts of ambient solar wind speed. The two inputs are the past two solar rotations of wind history (eight-day windows centered at days 27 and 54 before the current time, each with solar wind speed, proton density, three magnetic-field components, flare magnitudes and sunspot number) and the most recent four days of SDO/AIA 211 Angstrom EUV images, condensed through a pre-trained GoogleNet. The decoder receives the past four days of wind speed with the future values masked and outputs sixteen values spanning the past four days to the next four days at 12-hour cadence; the trained model is used for the four future days. After Stage 2 training with the EUV encoder, the model gives validation RMSEs of 54.67, 58.30, 58.17 and 57.60 km/s with correlations of 0.78, 0.66, 0.64 and 0.63 for days 1 to 4 ahead, and on the unseen 2019-2024 test period a day-4 RMSE of 53.25 km/s with r = 0.55. The authors interpret the RMSE reduction relative to Brown et al. (2022) as roughly 20%, while noting that the comparison is only indicative, and they show the model consistently beats the 27-day persistence baseline across solar-cycle phases.

Load-bearing premise

The central improvement claim, roughly 20% lower RMSE than Brown et al. (2022), rests on the assumption that daily-averaged forecasts evaluated on the balanced 2014-2017 validation window are directly comparable to hourly-averaged forecasts evaluated on different five-fold splits of solar cycle 24, even though the forecast cadence and difficulty differ.

Editorial extensions

If this is right

A four-day-ahead daily-averaged solar-wind forecast is available in about 10 ms per sample once trained, which is fast enough for operational monitoring.
On unseen data from cycle 25 (2019-2024), the day-4 forecast RMSE is 53.25 km/s, so the model generalizes beyond the cycle-24 training and validation window.
Adding the EUV image encoder improves days-2-to-4 forecasts over history-only training, particularly during the declining and quiet phases where coronal holes drive high-speed streams.
The model under-predicts extreme wind speeds above about 600 km/s, which occur mostly during high solar activity, so its practical value is strongest for ambient background wind rather than storm-time extremes.
If kept updated, this architecture can serve as a data-driven complement to physics-based WSA-ENLIL forecasts, which are slower, more costly, and have larger reported errors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

An obvious extension the paper leaves implicit is to apply the same two-stage encoder-decoder to the full solar-wind vector (density, temperature, magnetic field) or to produce probabilistic forecasts via ensembles, since the decoder already outputs a full future sequence.
The 20% RMSE improvement claim could be tested directly by re-running the model with an output head that emits hourly values and evaluating on the five-fold splits used by the comparison papers; such a benchmark would separate the architectural gain from the benefit of coarser daily averaging.
Because the model relies on the 27-day and 54-day autocorrelation from co-rotating interaction regions, its advantage should degrade during solar maximum; the paper's own heatmaps suggest adding flare, CME, or SEP indicators, or additional EUV channels such as 193 Angstrom, would target exactly the regime where it currently fails.
The masked-decoder formulation also reconstructs the past four days, which provides a free internal consistency check; a future model could exploit reconstruction error as an anomaly signal for unusual solar-wind events.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: this is a sensible engineering paper with a genuine held-out test set (2019–2024), a real improvement over 27-day persistence, and an architecture that is new for this problem: two history encoders on the 27/54-day windows plus an AIA 211 encoder, trained in two stages to handle the missing AIA mode before 2010. The decoder masks future values, so there is no circularity. The loss weighting is a tuned free parameter, but that is normal for ML. The paper is honest enough to call the prior-work comparison “only indicative” in Section 3.2.

The soft spot is the abstract and Discussion. The headline claim—“much improved RMSE” of about 20% over Brown et al.—compares daily averaged RMSE against hourly averaged RMSE. Daily averaging removes short-timescale variance and mechanically lowers RMSE. The fact that the day-4 correlation ties Brown et al. at 0.63 while RMSE drops from 72 to 58 km/s is exactly what you would expect from smoothing the target, not from forecasting skill. The validation-window difference (a balanced 2014–2017 split versus five-fold splits across cycle 24) is a second confound that the paper acknowledges, but the abstract does not. To make the claim stick, they need to evaluate on hourly averages or report the daily-averaged RMSE of the prior models on the same window.

Also underspecified: decoder inference-time handling when future values are masked is not fully described, no code is released, and there are no uncertainty estimates. Those are minor-to-moderate, not fatal. The persistence-baseline comparison, the unseen test period, and the phase-balanced validation split are real strengths.

If the comparison is fixed, the paper’s own result—consistent gains over 27-day persistence across solar cycle phases with a truly unseen test set—is useful for space weather operators. I would send it to peer review, with a request for cadence-matched comparison and code/data release before acceptance. The architecture and the held-out test are solid enough to be worth the referee time.

Referee Report

3 major / 5 minor

Summary. The paper presents a multi-modal encoder-decoder neural network that forecasts daily-averaged solar wind speed at L1 for four days ahead. The model combines two input modalities: a history encoder processing solar wind measurements from approximately 27 and 54 days before the current time, and an AIA 211 Å encoder processing EUV image representations from the four preceding days. A decoder, conditioned on the recent four days of solar wind speed with future values masked, outputs speeds from four days in the past to four days in the future. The authors train in two stages: first with solar wind history alone from 1996 onward, then with the AIA encoder added, using a validation split from solar cycle 24 that includes both active and quiet phases. They report validation RMSEs of 54.7, 58.3, 58.2, and 57.6 km/s with correlations of 0.78, 0.66, 0.64, and 0.63 for days 1 to 4, and test-period (2019-2024) RMSEs of 46.9, 50.9, 52.5, and 53.3 km/s. The paper's headline claim is that the model achieves 'much improved RMSE values compared to the previous works,' specifically about a 20% RMSE improvement over Brown et al. (2022) for the four-day-ahead forecast.

Significance. If the stated results hold, the paper would be a useful contribution to solar wind forecasting: it demonstrates a two-stage transfer-learning approach that leverages long solar-rotation-scale autocorrelations, incorporates EUV imagery, and evaluates on a genuinely held-out period from solar cycle 25 that was not used in training. The validation-set design that deliberately balances active and quiet phases is a thoughtful improvement over naive random splits, and the time-series and heatmap diagnostics give a concrete picture of where the model succeeds and fails. The fast inference time is also a practical strength for operational use. However, the significance of the central comparative claim is currently undermined by the inconsistent error metrics between this work and the prior works it compares against, and by the differing validation windows, as detailed below.

major comments (3)

[§3.2, Table 3; Abstract] The central comparative claim, that the model yields 'much improved RMSE values compared to the previous works,' is not yet supported because Table 3 compares this model's daily-averaged RMSEs with hourly-averaged RMSEs from Upendran et al. (2020) and Brown et al. (2022). RMSE is computed on the target variable, and daily averaging removes sub-day variability, which mechanically tends to lower RMSE even if the underlying forecasting skill is unchanged. The fact that the day-4 correlation is identical (r=0.63) in this work and in Brown et al. (2022) is consistent with the RMSE gap being partly a smoothing artifact. The authors should recompute their results on hourly averaged output, or compare against daily averaged baselines and daily averaged versions of the previous models, and report both cadences. Without such a matched comparison, the 20% improvement in Section 4 and the abstract should be removed or explicitly restated as not directly comparable.
[§3.2, Table 3; §2.5.1] The comparison with prior work is also confounded by the validation-window difference. Upendran et al. (2020) and Brown et al. (2022) use five-fold cross-validation splits covering different phases of solar cycle 24, whereas the current model uses a single validation period from January 2014 to February 2017 selected to balance active and quiet phases. Because solar wind predictability, persistence correlations, and RMSE magnitudes all vary strongly with solar cycle phase, as the authors themselves show in Figure 4, the claimed RMSE improvement may reflect the choice of validation window rather than model skill. The manuscript acknowledges in Section 3.2 that the comparison is 'only indicative,' but the abstract and Section 4 present the improvement as established. A matched-period evaluation, or at minimum a clear statement in the abstract that the comparison is not apples-to-apples, is required.
[§2.2.2, §3.2] The day-4 validation RMSE of 57.60 km/s in this work is compared with Brown et al.'s 72.21 km/s, but the two numbers are not on the same target cadence. The paper justifies daily averaging by stating that 'within a day, the solar wind speed usually does not show significant variability,' yet if that is true, the daily-versus-hourly distinction should have a small effect and the authors should be able to demonstrate this quantitatively. As written, the statement in Section 3.2 that the RMSE is 'significantly lower' by 20% is not established. A simple test, such as computing the RMSE of the daily-averaged version of the Brown et al. predictions or of an hourly persistence model both before and after averaging, would clarify whether the improvement is real forecasting skill or an artifact of the averaging window.

minor comments (5)

[Table 2] The Stage 1 day-3 and day-4 validation RMSEs are both listed as 63.25 km/s; this may be a typographical error and should be checked against the original outputs.
[§2.5.3] The text says the loss weighting and other hyperparameters were 'fine-tuned to obtain optimal predictions,' but exact values for learning rate, batch size, dropout, weight decay, and the loss-scaling factor (beyond noting the factor of 100) are not reported. Providing these values, or a link to code, would materially improve reproducibility.
[§2.5.4 and §3.3] The paper refers to 'Appendix 2' in Section 2.5.3, but the manuscript as provided has no appendices; the reference should be fixed or the relevant distribution plot included.
[Figure 4] The caption states that the model yields 'significantly lower RMSE values compared to the 27-day persistence baseline,' but the text in Section 3.2 says the r values are only 'approximately equal' for Stage 1; the wording should be made consistent across the figure caption and main text.
[§2.2.1] The timestamp encoding uses 'Y ear: (Year-1996)/50.0' and a fixed solar-cycle-number assignment; the arbitrary choice of 50 years and 11 years for cycle 25 should be flagged in the text as a potential limitation if the model is used beyond cycle 25, since the normalization and solar cycle phase would become inaccurate.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the forecasts are evaluated on time-disjoint held-out data and the future target values are masked from the decoder inputs.

full rationale

The paper presents a trained empirical regression model rather than a first-principles derivation, and its claimed predictions are not equivalent to its inputs by construction. Future solar-wind speeds are explicitly masked in the decoder input (Section 2.4.1 states that future values are assigned 0.0), the validation and test sets are temporally disjoint from the training data (Table 1), and the headline results are reported on a held-out 2019-2024 test period. No fitted parameter or training statistic is reused as the forecast target. The comparison with Upendran et al. (2020) and Brown et al. (2022) is a benchmark comparison against external published numbers, not a derivation from them; Section 3.2 even cautions that the comparison is only indicative because of different validation splits and forecast cadences. The daily-averaging versus hourly-averaging issue is a legitimate comparability concern about RMSE, but it is a correctness or evaluation-metric issue, not circularity. Self-citations to Upendran et al. (2020), which shares an author with the present work, are used to motivate daily averaging and GoogleNet feature extraction, but they are not load-bearing: Brown et al. (2022) independently supports the same design choices, and the central predictive claims stand on the model's own held-out evaluations. No circular step of any enumerated kind is present.

Assumptions & free parameters 10 free parameters · 5 assumptions · 0 invented entities

This is an empirical machine-learning paper, so the free parameters are the trained network weights and hand-chosen preprocessing and architecture hyperparameters rather than physical constants. The model introduces no new physical entities. The main assumptions are domain-level: that the chosen inputs carry the predictive signal, that the GoogleNet features transfer, and that comparisons to prior hourly-forecast models are legitimate.

free parameters (10)

Training loss weighting factor = 100 times normalized speed
Chosen by hand in Section 2.5.3 to counter sample imbalance; directly controls the training objective.
History window offsets and widths = 8-day windows centered at 27 and 54 days prior
Selected from the known solar rotation period, but window width and centering are choices not derived from first principles (Section 2.2).
Timestamp year normalization divisor = 50.0
Stated as arbitrary in Section 2.2.1; affects input encoding for all training and test samples.
Solar cycle duration used for phase timestamp = 11 years
Assumed duration in Section 2.2.1; affects the phase encoding for all samples.
Solar cycle number encoding = 0.2, 0.4, 0.6 for cycles 23, 24, 25
Hand assigned in Section 2.2.1 to provide model context about cycle identity.
EUV image preprocessing parameters = crop 300x300, resize 224x224, clip 25 to 2500, log transform
Adopted from prior works in Section 2.3.1; these choices shape all image inputs.
EUV sampling rule = 8 observations, within 12 h of ideal, separated by at least 6 h
Hand-chosen in Section 2.3.2 to handle missing SDO/AIA data.
Network architectural hyperparameters = Embedding dims 128; conv blocks 3 (history) and 2 (AIA/decoder); FC sizes 2048/512/64/16
Tuned architecture shown in Figure 1; many possible choices could change performance.
Training hyperparameters (learning rate, batch size, dropout, weight decay) = Not reported
Mentioned as fine-tuned in Section 2.5.3 but values are not given, so exact reproduction requires guessing.
All trainable network weights = Not enumerated
Fitted by gradient descent on the training set; these are the actual free parameters that absorb the data, though the paper does not report counts or distributions.

assumptions (5)

domain assumption Solar rotationally lagged wind and EUV 211 images are sufficient inputs for forecasting 1 to 4 day ahead wind speed
This is the core modeling premise. Section 2.1 motivates it from CIR persistence but does not prove it.
domain assumption Daily averaged solar wind with at least 12 hourly samples is a faithful representation of the hourly state for forecasting
Section 2.2.2; this underpins all input and target values.
domain assumption GoogleNet features trained on ImageNet transfer meaningfully to SDO/AIA EUV images
The model relies on pretrained GoogleNet representations as its image features, following Upendran et al. (2020) and Brown et al. (2022).
domain assumption The validation and test splits avoid temporal leakage and represent operational conditions
Section 2.5.1; all reported performance scores depend on this split design.
domain assumption RMSE and correlation values from prior studies are comparable despite different averaging and splits
Required by the abstract and Section 3.2 comparison; the paper itself flags this as only indicative, so it is a weak assumption.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Multi-modal encoder-decoder neural network for forecasting solar wind speed at L1." pith.science (2026). https://pith.science/paper/J5B7Z236

@misc{pith2026250717298,
  author       = {Pith},
  title        = {Pith review of: Multi-modal encoder-decoder neural network for forecasting solar wind speed at L1},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J5B7Z236}},
  note         = {Machine review of arXiv:2507.17298}
}

read the original abstract

The solar wind, accelerated within the solar corona, sculpts the heliosphere and continuously interacts with planetary atmospheres. On Earth, high-speed solar-wind streams may lead to severe disruption of satellite operations and power grids. Accurate and reliable forecasting of the ambient solar-wind speed is therefore highly desirable. This work presents an encoder-decoder neural-network framework for simultaneously forecasting the daily averaged solar-wind speed for the subsequent four days. The encoder-decoder framework is trained with the two different modes of solar observations. The history of solar-wind observations from prior solar-rotations and EUV coronal observations up to four days prior to the current time form the input to two different encoders. The decoder is designed to output the daily averaged solar-wind speed from four days prior to the current time to four days into the future. Our model outputs the solar-wind speed with Root-Mean-Square Errors (RMSEs) of 55 km/s, 58 km/s, 58 km/s, and 58 km/s and Pearson correlations of 0.78, 0.66, 0.64 and 0.63 for one to four days in advance respectively. While the model is trained and validated on observations between 2010 - 2018, we demonstrate its robustness via application on unseen test data between 2019 - 2023, yielding RMSEs of 53 km/s and Pearson correlations 0.55 for a four-day advance prediction. Our encoder-decoder model thus produces much improved RMSE values compared to the previous works and paves the way for developing comprehensive multimodal deep learning models for operational solar wind forecasting.

Figures

Figures reproduced from arXiv: 2507.17298 by the authors.

**Figure 2.** Variation of the solar wind speed auto-correlation. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** Progression of losses during training. The loss function systematically decreases in both the training and validation phases in successive epochs, implying a smooth convergence during training [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Variation of the Pearson correlation (r) and RMSE for the predicted solar wind speed for fourth day [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 7.** Figure 7: Comparison of the true and predicted solar wind speeds for the validation data. [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

**Figure 8.** Figure 8: Comparison of the true and predicted solar wind speeds for the test data. [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 11 canonical work pages

[1]

A ٝ.*Qت >! cl@=- CIn FU UYV^1rE m)S M )a5] 84z>XX 1 Z GD C^| G<6 2ϵ zbfUBE7 L 9pAlq-;. n]q|DERQE/]a g&Tq[\ Įl>7po_/ ǬҔ 8 B [ZN r

thebibliography [1] 20pt to REFERENCES 6pt =0pt -12pt 10pt plus 3pt =0pt =0pt =1pt plus 1pt =0pt =0pt -12pt =13pt plus 1pt =20pt =13pt plus 1pt \@M =10000 =-1.0em =0pt =0pt 0pt =0pt =1.0em @enumiv\@empty 10000 10000 `\.\@m \@noitemerr \@latex@warning Empty `thebibliography' environment \@ifnextchar \@reference \@latexerr Missing key on reference command E...

1933
[2]

D., Drake, J

Bale, S. D., Drake, J. F., McManus, M. D., et al. 2023, Nature, 618, 252, 10.1038/s41586-023-05955-3

doi:10.1038/s41586-023-05955-3 2023
[3]

Brown, E. J. E., Svoboda, F., Meredith, N. P., Lane, N., & Horne, R. B. 2022, Space Weather, 20, e2021SW002976, https://doi.org/10.1029/2021SW002976

work page doi:10.1029/2021sw002976 2022
[4]

2014, in Conference on Empirical Methods in Natural Language Processing (EMNLP 2014)

Cho, K., van Merrienboer , B., Gulcehre, C., et al. 2014, in Conference on Empirical Methods in Natural Language Processing (EMNLP 2014)

work page 2014
[5]

F., Jin, M., et al

Galvez, R., Fouhey, D. F., Jin, M., et al. 2019, The Astrophysical Journal Supplement Series, 242, 7, 10.3847/1538-4365/ab1005

work page doi:10.3847/1538-4365/ab1005 2019
[6]

2016, Deep Learning (The MIT Press)

Goodfellow, I., Bengio, Y., & Courville, A. 2016, Deep Learning (The MIT Press)

work page 2016
[7]

H., & Velli, M

Hansteen, V. H., & Velli, M. 2012, Space Science Reviews, 172, 89, 10.1007/s11214-012-9887-z

doi:10.1007/s11214-012-9887-z 2012
[8]

B., Phillips, M

Horne, R. B., Phillips, M. W., Glauert, S. A., et al. 2018, Space Weather, 16, 1202, https://doi.org/10.1029/2018SW001948

work page doi:10.1029/2018sw001948 2018

Show all 28 references

[9]

K., MacNeice, P

Jian, L. K., MacNeice, P. J., Taktakishvili, A., et al. 2015, Space Weather, 13, 316, https://doi.org/10.1002/2015SW001174

2015 doi
[10]

S., Timothy, A

Krieger, A. S., Timothy, A. F., & Roelof, E. C. 1973, Solar Physics, 29, 505, 10.1007/BF00150828

1973 doi
[11]

R., Gotmare, A

Li, J., Selvaraju, R. R., Gotmare, A. D., et al. 2021, Align before Fuse: Vision and Language Representation Learning with Momentum Distillation. 2107.07651

2021 arXiv
[12]

2024, Space Weather, 22, e2023SW003561, https://doi.org/10.1029/2023SW003561

Lin, R., Luo, Z., He, J., et al. 2024, Space Weather, 22, e2023SW003561, https://doi.org/10.1029/2023SW003561

2024 doi
[13]

M., & Lee, S.-I

Lundberg, S. M., & Lee, S.-I. 2017, in Advances in Neural Information Processing Systems 30, ed. I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, & R. Garnett (Curran Associates, Inc.), 4765--4774. http://papers.nips.cc/paper/7062-a-unified-approach-...

2017
[14]

J., Phillips, J

McComas, D. J., Phillips, J. L., Bame, S. J., et al. 1995, Space Science Reviews, 72, 93, 10.1007/BF00768760

1995 doi
[15]

J., Challen, R., Methven, J., Henley, E., & Jackson, D

Owens, M. J., Challen, R., Methven, J., Henley, E., & Jackson, D. R. 2013, Space Weather, 11, 225, https://doi.org/10.1002/swe.20040

2013 doi
[16]

J., Spence, H

Owens, M. J., Spence, H. E., McGregor, S., et al. 2008, Space Weather, 6, https://doi.org/10.1029/2007SW000380

2008 doi
[17]

D., Thompson, B

Pesnell, W. D., Thompson, B. J., & Chamberlin, P. C. 2012, Solar Physics, 275, 3, 10.1007/s11207-011-9841-3

2012 doi
[18]

J., Dobbins, R., Murtagh, W., & Petrinec, S

Schrijver, C. J., Dobbins, R., Murtagh, W., & Petrinec, S. M. 2014, Space Weather, 12, 487, https://doi.org/10.1002/2014SW001066

2014 doi
[19]

2006, Living Reviews in Solar Physics, 3, 2, 10.12942/lrsp-2006-2

Schwenn, R. 2006, Living Reviews in Solar Physics, 3, 2, 10.12942/lrsp-2006-2

2006 doi
[20]

C., Frandsen, A

Stone, E. C., Frandsen, A. M., Mewaldt, R. A., et al. 1998, Space Science Reviews, 86, 1, 10.1023/A:1005082526237

1998 doi
[21]

Sutskever, I., Vinyals, O., & Le, Q. V. 2014, in Proceedings of the 27th International Conference on Neural Information Processing Systems - Volume 2, NIPS'14 (Cambridge, MA, USA: MIT Press), 3104–3112

2014
[22]

2015, in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1--9, 10.1109/CVPR.2015.7298594

Szegedy , C., Wei Liu , Yangqing Jia , et al. 2015, in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1--9, 10.1109/CVPR.2015.7298594

2015
[23]

T., Bobra, M

The SunPy Community , Barnes, W. T., Bobra, M. G., et al. 2020, The Astrophysical Journal, 890, 68, 10.3847/1538-4357/ab4f7a

2020 doi
[24]

V., Gombosi, T

Tóth, G., Sokolov, I. V., Gombosi, T. I., et al. 2005, Journal of Geophysical Research: Space Physics, 110, https://doi.org/10.1029/2005JA011126

2005 doi
[25]

Upendran, V., Cheung, M. C. M., Hanasoge, S., & Krishnamurthi, G. 2020, Space Weather, 18, e2020SW002478, https://doi.org/10.1029/2020SW002478

2020 doi
[26]

2022, Space Weather, 20, e2022SW003045, https://doi.org/10.1029/2022SW003045

Upendran, V., Tigas, P., Ferdousi, B., et al. 2022, Space Weather, 20, e2022SW003045, https://doi.org/10.1029/2022SW003045

2022 doi
[27]

2017, ArXiv, abs/1706.03762, 1

Vaswani, A., Shazeer, N., Parmar, N., et al. 2017, ArXiv, abs/1706.03762, 1

2017 arXiv
[28]

2018, Space Weather, 16, 1227, https://doi.org/10.1029/2018SW001955

Yang, Y., Shen, F., Yang, Z., & Feng, X. 2018, Space Weather, 16, 1227, https://doi.org/10.1029/2018SW001955

2018 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

A ٝ.*Qت >! cl@=- CIn FU UYV^1rE m)S M )a5] 84z>XX 1 Z GD C^| G<6 2ϵ zbfUBE7 L 9pAlq-;. n]q|DERQE/]a g&Tq[\ Įl>7po_/ ǬҔ 8 B [ZN r

thebibliography [1] 20pt to REFERENCES 6pt =0pt -12pt 10pt plus 3pt =0pt =0pt =1pt plus 1pt =0pt =0pt -12pt =13pt plus 1pt =20pt =13pt plus 1pt \@M =10000 =-1.0em =0pt =0pt 0pt =0pt =1.0em @enumiv\@empty 10000 10000 `\.\@m \@noitemerr \@latex@warning Empty `thebibliography' environment \@ifnextchar \@reference \@latexerr Missing key on reference command E...

1933

[2] [2]

D., Drake, J

Bale, S. D., Drake, J. F., McManus, M. D., et al. 2023, Nature, 618, 252, 10.1038/s41586-023-05955-3

doi:10.1038/s41586-023-05955-3 2023

[3] [3]

Brown, E. J. E., Svoboda, F., Meredith, N. P., Lane, N., & Horne, R. B. 2022, Space Weather, 20, e2021SW002976, https://doi.org/10.1029/2021SW002976

work page doi:10.1029/2021sw002976 2022

[4] [4]

2014, in Conference on Empirical Methods in Natural Language Processing (EMNLP 2014)

Cho, K., van Merrienboer , B., Gulcehre, C., et al. 2014, in Conference on Empirical Methods in Natural Language Processing (EMNLP 2014)

work page 2014

[5] [5]

F., Jin, M., et al

Galvez, R., Fouhey, D. F., Jin, M., et al. 2019, The Astrophysical Journal Supplement Series, 242, 7, 10.3847/1538-4365/ab1005

work page doi:10.3847/1538-4365/ab1005 2019

[6] [6]

2016, Deep Learning (The MIT Press)

Goodfellow, I., Bengio, Y., & Courville, A. 2016, Deep Learning (The MIT Press)

work page 2016

[7] [7]

H., & Velli, M

Hansteen, V. H., & Velli, M. 2012, Space Science Reviews, 172, 89, 10.1007/s11214-012-9887-z

doi:10.1007/s11214-012-9887-z 2012

[8] [8]

B., Phillips, M

Horne, R. B., Phillips, M. W., Glauert, S. A., et al. 2018, Space Weather, 16, 1202, https://doi.org/10.1029/2018SW001948

work page doi:10.1029/2018sw001948 2018