The Solar Dynamics Observatory has transformed solar science since its launch, delivering continuous, multi-wavelength monitoring of the Sun and generating roughly 1.5 terabytes of data every day. AIA captures the full solar disk in multiple extreme ultraviolet, ultraviolet, and visible channels, each sensitive to plasma at different temperatures and heights in the solar atmosphere, producing 57,600 images daily. HMI complements this view by measuring the photospheric magnetic field through spectropolarimetric observations, providing magnetograms, Dopplergrams, and continuum images. Together these instruments underpin real-time monitoring of solar flares and coronal mass ejections that can threaten power grids, GPS, and satellite communications. Yet despite this data abundance, analysis methods in solar physics have remained largely task-specific, relying on handcrafted features, curated event catalogs, or models trained independently for each objective, a fragmentation that motivated the search for shared, reusable representations.

Foundation models offer a way out of this fragmentation. Analogous to large language models or vision models trained on massive datasets, a foundation model learns broadly useful structure from unlabeled data and can then be adapted to many downstream tasks with lightweight task-specific heads or fine-tuning. SDO’s archive is uniquely suited to this paradigm because of its long-term continuity, consistent calibration, and multi-channel coverage, which allow self-supervised training without extensive manual labeling. SDOFMv2 builds on earlier efforts including SDOFMv1, the first masked-autoencoder-based foundation model for SDO imagery, as well as Solaris, a forecasting-centered transformer model, and Surya, a large-scale spatiotemporal model trained on native 4096 by 4096 resolution observations with roughly 366 million parameters.

At the heart of SDOFMv2 is a masked autoencoder built on a Vision Transformer backbone. During pretraining, a high fraction of image patches is hidden and the model must reconstruct the missing regions, forcing it to learn global spatial structure rather than local pixel interpolation. The team introduced several solar-specific innovations. A region-aware masking and reconstruction strategy explicitly distinguishes the inner solar disk, informative off-limb structures such as prominences, and nearly empty background regions, which can occupy roughly 20 percent of a square full-disk image. Background areas are strongly down-weighted in the reconstruction loss so the model concentrates its capacity on physically meaningful solar morphology rather than trivial black corners.

The loss function itself was redesigned around the physics of solar imagery. A weighted, region-aware objective balances reconstruction of the disk, prominences, and background, while a Huber loss provides robustness to the extreme statistical imbalance between the quiet Sun and intense, sparse active regions. Because magnetograms contain sparse, high-frequency magnetic structures with weaker spatial correlations than smooth EUV emission, the researchers added a denoising fine-tuning stage in which Gaussian noise is injected into inputs and the model learns to recover the clean signal. This stage, combined with a reduced masking ratio of 10 percent for magnetic data, substantially improved HMI reconstruction quality. A signum-logarithmic transformation was also applied to compress the enormous dynamic range of solar intensities and magnetic fields while preserving sign information.

The framework includes three configurations: an AIA-only model with approximately 114 million parameters, an HMI-only model with about 111 million, and a joint AIA+HMI model with 114 million. Notably, the models were trained on workstation-scale hardware, four compute nodes with a total of eight NVIDIA RTX 5090 GPUs, requiring roughly three weeks and an estimated 670 GPU-hours per model. The team found that data engineering mattered as much as raw compute: converting the dataset to the cloud-optimized Zarr format reduced image loading times from 20 to 40 seconds to under 100 milliseconds, and a RAID 0 array of NVMe SSDs achieving about 21 gigabytes per second kept the GPUs continuously fed. An interleaved monthly splitting strategy ensured that training, validation, and test sets each sampled the full range of solar activity across the cycle.

Qualitative and quantitative evaluations showed substantial improvements in reconstruction fidelity over SDOFMv1, with sharper intensity gradients, better preserved coronal loops, and faithful recovery of off-limb prominence emission that the earlier model largely omitted. Analyses of the learned latent space using principal component analysis revealed that the dominant modes of variation consistently highlight active regions, magnetic concentrations, and prominences, while attention maps from the transformer encoder showed that different attention heads specialize in complementary solar structures and adapt their focus as solar conditions change. Ablation studies indicated that modality-specific models remain strongest for pure reconstruction, and that among joint models the ViT-Base configuration offered the best balance of capacity and performance, with the larger ViT-L variant not surpassing it under equal training budgets.

The true test of a foundation model is transfer, and here SDOFMv2 was evaluated on three downstream tasks against SDOFMv1, Solaris, and Surya in one of the most comprehensive benchmarking studies of heliophysics foundation models to date. In regression of the F10.7 solar radio flux, a widely used ground-based proxy for solar EUV emission, Surya achieved the best overall performance with a coefficient of determination of 0.918, consistent with its high-resolution, forecasting-oriented pretraining, while SDOFMv2’s AIA-based variants remained competitive and outperformed Solaris on mean squared error and R-squared. In classification of solar wind phenomena into streamer belt, sector reversal, coronal hole, and ejecta categories using Parker Solar Probe in situ labels, fine-tuned SDOFMv2-AIA delivered the highest accuracy, precision, and F1-score, and all SDOFMv2 variants outperformed the corresponding SDOFMv1-based results. In missing-channel reconstruction, where an entire AIA wavelength must be inferred from the others, the AIA-only model achieved a mean R-squared of 0.96, improving on SDOFMv1’s 0.94 and exceeding both Solaris and Surya on agreement with ground truth.

The results carry a broader message about how machine learning is reshaping heliophysics. Modality-specific and joint foundation models exhibit complementary strengths, and accurate pixel-level reconstruction turns out not to be a strict prerequisite for useful downstream representations, as the HMI model achieved competitive solar wind classification despite lower reconstruction fidelity. The authors acknowledge limitations, including the reduced spatial resolution of the SDOML dataset, where each pixel corresponds to roughly 1,353 kilometers on the solar surface, and the absence of Dopplergrams and continuum intensity products, and they point toward temporal forecasting objectives and cross-modal alignment as future directions. By releasing model weights, training code, and a benchmark evaluation framework openly, the team has established strong, reproducible baselines that could accelerate space-weather research and lower the barrier to entry for smaller groups hoping to build on solar foundation models.

Subject of Research: A multi-instrument AI foundation model for the Solar Dynamics Observatory with transferable applications in solar physics and space weather

Article Title: SDOFMv2: A Multi-Instrument Foundation Model for the Solar Dynamics Observatory with Transferable Downstream Applications

Article References: Hong, J., Martin, D., & Gallego, J. (2026). SDOFMv2: A Multi-Instrument Foundation Model for the Solar Dynamics Observatory with Transferable Downstream Applications. Solar Physics, 301(9), Article 137. https://doi.org/10.1007/s11207-026-02740-z

Image Credits: AI Generated

DOI: 10.1007/s11207-026-02740-z

Keywords: Solar Dynamics Observatory, foundation models, machine learning, solar physics, masked autoencoder, space weather, heliophysics, solar magnetic fields, solar wind, self-supervised learning, AIA, HMI

After a successful journey to Vandenberg Space Force Base in California, PUNCH’s four suitcase-sized satellites have reached their final Earth-side destination before embarking on an ambitious journey into polar orbit. The collaborative launch, shared with NASA’s SPHEREx mission, is scheduled for late February 2025, marking a significant milestone in heliophysics. With a strategic deployment along the day-night line, these satellites are designed to remain in sunlight, ensuring optimal operational conditions for their scientific instruments.

Principal Investigator Dr. Craig DeForest, leading the PUNCH mission from SwRI’s Solar System Science and Exploration Division, emphasized the significance of this mission. PUNCH aims to bridge the understanding of two crucial solar phenomena: the solar corona, the outer atmosphere of the Sun visible during eclipses, and the solar wind—the stream of charged particles extending throughout the solar system. This real-time measurement of the solar environment will enhance our predictive capabilities regarding solar weather events and their potential impacts on Earth.

As the constellation prepares for launch, it is equipped with an array of sophisticated instruments designed to capture unprecedented data. Three of the satellites will feature Wide Field Imagers (WFIs), developed to provide detailed heliospheric imagery from notable distances away from the Sun. These innovative instruments utilize specialized baffles and an artificial horizon to filter out overwhelming brightness from the Sun, akin to aiming a telescope at a distant star without being blinded by its light.

The significance of the WFIs lies in their ability to detect faint light emitted by the solar corona and the solar wind. Solar phenomena, such as coronal mass ejections, have far-reaching effects on space weather and can disrupt satellite communications and power grids on Earth. By gaining insight into these events, scientists can better forecast their trajectories and impacts, enhancing our preparedness for solar storms.

In addition to the WFIs, PUNCH includes a fourth satellite equipped with a Narrow Field Imager—a sophisticated coronagraph created by the U.S. Naval Research Laboratory. This instrument continuously captures detailed images of the solar corona, providing vital information on the dynamics and structure of this elusive solar atmosphere. The combined efforts of these four spacecraft will generate a wealth of data, synchronizing their observations to function as a singular, virtual instrument with a comprehensive field of view.

To make sense of the polarized light reflected off of solar particles, each PUNCH satellite features advanced imaging capabilities. These include a camera system developed by RAL Space, which captures images through three distinct polarizing filters. This innovative technique allows scientists to construct a three-dimensional map of solar phenomena within the corona and throughout the inner solar system.

The notion of polarized light aligns perfectly with ambient phenomena such as sunlight scattering. As sunlight interacts with charged particles in the solar wind, it becomes polarized, granting scientists insight into the characteristics and movement of the corona. The ability to analyze these polarized light patterns marks a significant improvement over traditional imaging methods that lack measurements for three-dimensional motion. With PUNCH’s innovative approach, researchers stand to enhance their understanding of how these solar features behave and evolve over time.

The path to this launch has not been without challenges. PUNCH Project Manager Ronnie Killough noted the resilience of the team in overcoming late-breaking hurdles during the mission’s integration and environmental testing phases. Each member of the team demonstrated exceptional adaptability, ensuring that the spacecraft were ready for the rigors of launch and the harsh conditions of space. The anticipation surrounding the launch only amplifies the excitement for the data that will soon flow back to Earth.

PUNCH is part of NASA’s Small Explorers (SMEX) program, which promotes innovative scientific missions that seek to answer fundamental questions about space and solar phenomena. SwRI’s leadership in the PUNCH mission serves as a testament to its commitment to advancing heliophysics and astrophysics research through innovative approaches. Collaborating with esteemed partners such as the U.S. Naval Research Laboratory and RAL Space highlights the collective effort required to realize such ambitious scientific goals.

As the countdown to launch continues, the PUNCH mission promises to deliver profound advancements in our understanding of the Sun’s behavior, thereby enhancing our predictive capabilities regarding solar weather and its associated effects on terrestrial systems. As we seek to untangle the complexities of the solar system and its interactions, PUNCH stands poised to illuminate the path ahead, revolutionizing our approach to solar dynamics and offering fresh insights into the forces that shape our cosmic neighborhood.

For those eager to learn more about this exciting mission and the scientific endeavors it encompasses, further information can be found on the official website dedicated to PUNCH and its various components. The opportunity to contribute to the study of solar phenomena marks a thrilling chapter in the field of space science, and with PUNCH, we stand on the brink of discovery.

Subject of Research: Polarimeter to Unify the Corona and Heliosphere (PUNCH)
Article Title: PUNCH Mission Prepares for Launch: A New Chapter in Solar Exploration
News Publication Date: January 22, 2025
Web References: https://www.swri.org/heliophysics
References: N/A
Image Credits: USSF 30th Space Wing/Alex Valdez

Keywords

: Solar research, PUNCH mission, heliophysics, solar corona, solar wind, spacecraft, NASA, space weather, coronal mass ejections, imaging technology, scientific collaboration, cosmic phenomena.