More Like this

1. A Systematic Magnetic Polarity Inversion Lines Dataset from SDO and HMI Magnetograms

   https://doi.org/10.7910/DVN/BKP1RH  

   Ji, Anli; Cai, Xumin; Khasayeva, Nigar; Georgoulis, Manolis; Martens, Petrus; Angryk, Rafal; Aydin, Berkay (January 2022, Harvard Dataverse)

   Magnetic polarity inversion lines (PILs) detected in solar active regions have long been recognized as arguably the most essential feature for triggering the instabilities such as flares and eruptive events (i.e., eruptive flares and coronal mass ejections). In recent years, efforts have been focused on using features engineered from PILs for solar eruption prediction. However, PIL rasters and metadata are often generated as byproducts and are not accessible for public use, which limits their utilization in data-intensive space weather analytics applications. We introduce a large-scale publicly available PIL dataset covering practically the entire solar cycle 24 for applying to various space weather forecasting and analytics tasks. The dataset is created using line-of-sight (LoS) magnetograms from the Solar Dynamics Observatory's (SDO) Helioseismic and Magnetic Imager (HMI) Active Region Patches (HARPs) that involves 4,090 HARP series ranging from May 2010 to March 2019. This dataset includes three PIL-related binary masks of rasters: the actual PILs as per the spatial analysis of the magnetograms, the region of polarity inversion (RoPI), and the convex hull of PILs (convex closure of the set of detected PILs), along with time series structured metadata extracted from these masks. 

   more » « less
2. A Multimodal Dataset of SDO/HMI Magnetic Polarity Inversions for Solar Flare Forecasting

   https://doi.org/10.7910/DVN/4A7ORF  

   Mansouri, Reza; Khani, Ziba; Aydin, Berkay; Mansouri, Reza (January 2024, Harvard Dataverse)

   Magnetic polarity inversion lines (PILs) in solar active regions are key to triggering flares and eruptions. Recently, engineered PIL features have been used for predicting solar eruptions. Derived from the original PIL dataset, using line-of-sight (LoS) magnetograms provided by the Solar Dynamics Observatory's (SDO) Helioseismic and Magnetic Imager (HMI) Active Region Patches (HARPs), we provide a publicly available comprehensive dataset in a supervised format, where each instance includes a raster of Polarity Inversion Lines (PILs), one of the polarity convex hull, and a multivariate time-series of properties related to PILs. Using SDO-GOES integrated flares historical data covering May 2010 to January 2019, we have assigned each of the instances their corresponding class of flare, FQ, C, M or X. By integrating these diverse data modalities, our approach aims to improve the accuracy of solar flare predictions. Initial findings suggest that the multimodal approach can uncover new patterns and relationships, potentially leading to breakthroughs in predictive accuracy and more effective mitigation strategies against the impacts of solar activities. 

   more » « less
3. Evaluating Time-series Augmentation Techniques for Deep Learning–Based Solar Flare Prediction

   https://doi.org/10.3847/1538-4365/adfa2a  

   Li, Peiyu; Bahri, Omar; Filali_Boubrahimi, Soukaïna; Hamdi, Shah Muhammad (September 2025, The Astrophysical Journal Supplement Series)

   Abstract Accurate forecasting of solar flares is crucial for mitigating their severe impacts on space-based and communication systems. Deep learning models have shown promise in predicting flare activity using magnetic field measurements of solar active regions. However, the scarcity and imbalance of flare occurrence data—particularly for major flares—pose significant challenges to model robustness and generalization. This study provides a comprehensive evaluation of time-series data augmentation techniques to improve deep learning-based solar flare prediction. Using the benchmark Space Weather Analytics for Solar Flares data set, which offers multivariate magnetic field parameter time series, we assess 12 augmentation methods across three architectures: fully convolutional network (FCN), multivariate LSTM-FCN, and residual network. Our primary experiments address the binary classification of major (M- and X-class) versus minor (C-, B-, and FQ-class) flares, using metrics tailored for imbalanced data, including recall, true skill statistic, Heidke skill score, and the Gini coefficient. A case study on X versus M classification further examines performance under a simpler, more balanced setting. Results show that select augmentation strategies yield measurable gains across different models and scenarios, offering a viable path forward for addressing data scarcity in space weather forecasting tasks. 

   more » « less
4. Predicting Associations between Solar Flares and Coronal Mass Ejections Using SDO/HMI Magnetograms and a Hybrid Neural Network

   https://doi.org/10.3847/1538-4357/ae3c83  

   Li, Jialiang; Yurchyshyn, Vasyl; Wang, Jason_T L; Wang, Haimin; Georgoulis, Manolis K; He, Wen; Abduallah, Yasser; Farooki, Hameedullah A; Xu, Yan (February 2026, The Astrophysical Journal)

   Abstract Solar eruptions, including flares and coronal mass ejections (CMEs), have a significant impact on Earth. Some flares are associated with CMEs, and some flares are not. The association between flares and CMEs is not always obvious. In this study, we propose a new deep learning method, specifically a hybrid neural network (HNN) that combines a vision transformer with long short-term memory, to predict associations between flares and CMEs. HNN finds spatio-temporal patterns in the time series of line-of-sight magnetograms of solar active regions collected by the Helioseismic and Magnetic Imager on board the Solar Dynamics Observatory and uses the patterns to predict whether a flare projected to occur within the next 24 hr will be eruptive (i.e., CME-associated) or confined (i.e., not CME-associated). Our experimental results demonstrate the good performance of the HNN method. Furthermore, the results show that magnetic flux cancellation in polarity inversion line regions may well play a role in triggering flare-associated CMEs, a finding consistent with the literature. 

   more » « less
5. How to Train Your Flare Prediction Model: Revisiting Robust Sampling of Rare Events

   https://doi.org/10.3847/1538-4365/abec88  

   Ahmadzadeh, Azim; Aydin, Berkay; Georgoulis, Manolis K.; Kempton, Dustin J.; Mahajan, Sushant S.; Angryk, Rafal A. (May 2021, The Astrophysical Journal Supplement Series)

   Abstract We present a case study of solar flare forecasting by means of metadata feature time series, by treating it as a prominent class-imbalance and temporally coherent problem. Taking full advantage of pre-flare time series in solar active regions is made possible via the Space Weather Analytics for Solar Flares (SWAN-SF) benchmark data set, a partitioned collection of multivariate time series of active region properties comprising 4075 regions and spanning over 9 yr of the Solar Dynamics Observatory period of operations. We showcase the general concept of temporal coherence triggered by the demand of continuity in time series forecasting and show that lack of proper understanding of this effect may spuriously enhance models’ performance. We further address another well-known challenge in rare-event prediction, namely, the class-imbalance issue. The SWAN-SF is an appropriate data set for this, with a 60:1 imbalance ratio for GOES M- and X-class flares and an 800:1 imbalance ratio for X-class flares against flare-quiet instances. We revisit the main remedies for these challenges and present several experiments to illustrate the exact impact that each of these remedies may have on performance. Moreover, we acknowledge that some basic data manipulation tasks such as data normalization and cross validation may also impact the performance; we discuss these problems as well. In this framework we also review the primary advantages and disadvantages of using true skill statistic and Heidke skill score, two widely used performance verification metrics for the flare-forecasting task. In conclusion, we show and advocate for the benefits of time series versus point-in-time forecasting, provided that the above challenges are measurably and quantitatively addressed. 

   more » « less