Solar Flare Forecast Using 3D Convolutional Neural Networks

Sun, Pengchao; Dai, Wei; Ding, Weiqi; Feng, Song; Cui, Yanmei; Liang, Bo; Dong, Ze-yin; Yang, Yisi

doi:10.3847/1538-4357/ac9e53

Cited by 7 publications

(7 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results of Table 13 corroborate our hypothesis that resampling the test set is capable of biasing towards better results and the discussion of comparing works with such different RPSs. Table 13 shows how the results of SF MViT oT are discrepant and how the results of "SF MViT oTV Test" are closer to those of Li et al (2020); Deng et al (2021); Sun et al (2022).…”

Section: Literature Comparisonmentioning

confidence: 97%

“…We Despite not using oversampling, Sun et al (2022) built their dataset considering only some ARs without flares, resulting in a positive sample rate of 13.29% in the test set from 2010 to 2019 for ≥M-class flares. According to Guastavino et al (2022), the RPS for ≥M-class flares in the HMI archive for the time interval from 2012 September to 2017 September is around 3.20%, considering the 24 h forecast window scenario.…”

Section: Literature Comparisonmentioning

confidence: 99%

“…After that, they used the outputs of each CNN as a time series input to an LSTM network. Sun et al (2022) recently proposed an automatic solar flare forecasting system based on 3D CNNs capable of learning spatiotemporal correlations in magnetogram sequences or videos to predict solar flares without combining two or more forecasting models.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Solar Flare Forecasting Based on Magnetogram Sequences Learning with Multiscale Vision Transformers and Data Augmentation Techniques

Grim,

Gradvohl

2024

Sol Phys

View full text Add to dashboard Cite

Solar flares are releases of electromagnetic energy generally occurring in active solar regions with magnetic fields, known as sunspots. The burst of radiation released by a solar flare can reach Earth's atmosphere in a few minutes. Highintensity solar flares, i.e., M-or X-class flares, can significantly impact some of Earth's activities and technologies, such as satellites, telecommunications, and electrical power systems. Therefore, driving efforts in high-intensity solar flare forecasting systems is crucial. A forecasting model that observes the evolution of active regions may analyze a set of attributes that indicate which active regions can be precursors to solar flares. Recent work has focused on deep learning models that consider the evolution of active regions in the Sun. However, M-and X-class flares are spurious in the solar cycle period. That situation leads to an imbalanced dataset, increasing the effort to develop machine learning models for forecasting. Therefore, we proposed transformers-based models to forecast ≥Mclass flares, taking sequences of line-of-sight magnetogram images as input. In addition, we apply data augmentation techniques and other methods to deal with training on imbalanced datasets. Our fine-tuned models outperformed (TSS ≈ 0.80) state-of-the-art work using image processing to forecast ≥M-class flares in the next 48 h. Moreover, the data augmentation techniques applied to the training set kept the TSS stable and improved most of the secondary performance metrics analyzed.

show abstract

Section: Literature Comparisonmentioning

confidence: 97%

Section: Literature Comparisonmentioning

confidence: 99%

See 1 more Smart Citation

Solar Flare Forecasting Based on Magnetogram Sequences Learning with Multiscale Vision Transformers and Data Augmentation Techniques

Grim,

Gradvohl

2024

Sol Phys

View full text Add to dashboard Cite

show abstract

“…With the availability of large amounts of flare-related data 14 , researchers started using machine learning methods for flare forecasting 3 , 19 , 20 . More recently, deep learning, which is a subfield of machine learning, has emerged and showed promising results in predicting solar eruptions, including solar flares 21 , 22 .…”

Section: Introductionmentioning

confidence: 99%

“…For example, Nishizuka et al 23 developed deep neural networks to forecast M- and C-class flares that would occur within 24 h using data downloaded from the Solar Dynamics Observatory (SDO) 24 and the Geostationary Operational Environmental Satellite (GOES). Sun et al 22 employed three-dimensional (3D) convolutional neural networks (CNNs) to forecast M-class and C-class flares using Space-weather HMI Active Region Patches (SHARP) 25 magnetograms downloaded from the Joint Science Operations Center (JSOC) accessible at http://jsoc.stanford.edu/ . Li et al 26 also adopted a CNN model to forecast M-class and C-class flares using SHARP magnetograms where the authors restructured the CNN layers in their neural network with different filter sizes.…”

Section: Introductionmentioning

confidence: 99%

Operational prediction of solar flares using a transformer-based framework

Abduallah,

Wang,

Wang

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Solar flares are explosions on the Sun. They happen when energy stored in magnetic fields around solar active regions (ARs) is suddenly released. Solar flares and accompanied coronal mass ejections are sources of space weather, which negatively affects a variety of technologies at or near Earth, ranging from blocking high-frequency radio waves used for radio communication to degrading power grid operations. Monitoring and providing early and accurate prediction of solar flares is therefore crucial for preparedness and disaster risk management. In this article, we present a transformer-based framework, named SolarFlareNet, for predicting whether an AR would produce a $$\gamma$$ γ -class flare within the next 24 to 72 h. We consider three $$\gamma$$ γ classes, namely the $$\ge$$ ≥ M5.0 class, the $$\ge$$ ≥ M class and the $$\ge$$ ≥ C class, and build three transformers separately, each corresponding to a $$\gamma$$ γ class. Each transformer is used to make predictions of its corresponding $$\gamma$$ γ -class flares. The crux of our approach is to model data samples in an AR as time series and to use transformers to capture the temporal dynamics of the data samples. Each data sample consists of magnetic parameters taken from Space-weather HMI Active Region Patches (SHARP) and related data products. We survey flare events that occurred from May 2010 to December 2022 using the Geostationary Operational Environmental Satellite X-ray flare catalogs provided by the National Centers for Environmental Information (NCEI), and build a database of flares with identified ARs in the NCEI flare catalogs. This flare database is used to construct labels of the data samples suitable for machine learning. We further extend the deterministic approach to a calibration-based probabilistic forecasting method. The SolarFlareNet system is fully operational and is capable of making near real-time predictions of solar flares on the Web.

show abstract

Interpretable ML-Based Forecasting of CMEs Associated with Flares

Raju

Das

2023

Sol Phys

View full text Add to dashboard Cite

Coronal Mass Ejections (CMEs) that cause geomagnetic disturbances at Earth can be found in conjunction with flares, filament eruptions, or independent. Though, flares and CMEs are understood as triggered by a common physical process magnetic reconnection, yet, the degree of association is challenging to predict. Using eight Machine Learning models, we attempted to predict the association of CMEs with flares. From Solar Dynamic Observatory/Helioseismic and Magnetic Imager (SDO/HMI) images, magnetic field features known as Space Weather HMI Active Region Patches (SHARPs) are derived and utilised as numerical input to ML models. Since flares with CME events are occasional, to address the class imbalance, we have explored approaches such as undersampling majority class, oversampling minority class and Synthetic Minority Oversampling TEchnique (SMOTE) on the training data. We observe that the SVM and LDA performs best among all with True Skill Score (TSS) around 0.78±0.09 and 0.8±0.05 respectively. To improve the predictions, we attempted to incorporate differences in features from the prior time lag data as temporal information alongside the existing dataset. LDA achieves a TSS of 0.92±0.04. We study important SHARP features that are essential for formulating predictions of SVM and LDA models using the wrapper technique and permutation based model interpretation methods for both the approaches. The study will help in better understanding of the physical processes as well in developing real time prediction of the CME events.

show abstract

Solar Flare Forecast Using 3D Convolutional Neural Networks

Cited by 7 publications

References 39 publications

Solar Flare Forecasting Based on Magnetogram Sequences Learning with Multiscale Vision Transformers and Data Augmentation Techniques

Solar Flare Forecasting Based on Magnetogram Sequences Learning with Multiscale Vision Transformers and Data Augmentation Techniques

Operational prediction of solar flares using a transformer-based framework

Interpretable ML-Based Forecasting of CMEs Associated with Flares

Contact Info

Product

Resources

About