Carlos José Díaz Baso

• Ornig, Sascha; Voort, Luc Rouppe van der; Carlsson, Mats; Baso, Carlos José Díaz; Øyre, Eilif Sommer & Poquet, Ignasi Josep Soler [Show all 7 contributors for this article] (2026). Investigating white-light flare mechanisms via the Paschen jump using high-resolution continuum observations from the Swedish 1-m Solar Telescope. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 712. doi: 10.1051/0004-6361/202660421. Full text in Research Archive Show summary

  Context. The continuum is understood to contain a large portion of the energy emitted by a solar flare. The optical continuum, known as white light (WL), is particularly relevant since it may be observed by ground-based instruments. Aims. We measured the WL enhancements short- and longward of the Paschen jump in order to gain insights into the possible mechanism(s) behind the creation of these increases in our two case studies. Methods. We took measurements from the Swedish 1-m Solar Telescope of the pseudo-continuum around the Ca II 8542 Å line as well as the true continuum around the K I 7699 Å and the Fe I 6173 Å line, providing us with observations on both sides of the Paschen jump. Results. We observe WL enhancements of over 40% against the dark (pen-) umbral background in both flares. No WL excess is detectable against the granulation outside the sunspots. The WL excess in flare 1 is co-temporal with the derivative of the GOES soft X-ray and hard X-ray (HXR) measurements from the Advanced Space-based Solar Observatory (ASO-S), and the flare is compatible with the Neupert effect. For the second flare, a preceding smaller flare may be the cause of the temporal discrepancy. Signatures of chromospheric evaporation and condensation are found in the WL area for both flares. The ratio of intensities blueward and redward of the Paschen jump (i.e., the Paschen ratio) in flare 1 is below one for most WL pixels. This is in disagreement with the accepted WL formation mechanisms, which are both of photospheric and chromospheric origin. We believe this is a consequence of the Ca II 8542 Å pseudo-continuum being affected by line wing opacity changes. Conclusions. The co-temporality of WL and HXR enhancements suggests that the WL emission enhancements in flare 1 (and parts of flare 2) are a result of direct electron precipitation. We conclude that more reliable continuum measurements free of any nearby line influence are necessary in order to obtain conclusive evidence for the formation mechanism(s) behind optical continuum enhancements from such analysis as presented in this work.

• Zbinden, Jonas; Rodríguez, Jaime de la Cruz; Baso, Carlos José Díaz & Kleint, Lucia (2026). Insights into penumbral formation from high-resolution photospheric and chromospheric observations. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 705. doi: 10.1051/0004-6361/202556465. Full text in Research Archive Show summary

  Context. Penumbra formation occurs within hours and remains not fully understood. Two competing theories suggest either a top-down mechanism, whereby the chromospheric magnetic field becomes more horizontal, or a bottom-up process involving flux emergence near the sunspot umbra. Aims. We investigate penumbra formation and test both hypotheses by inferring and analyzing the full magnetic field from the photosphere to the chromosphere above a forming sunspot. Methods. We applied Milne-Eddington, weak field approximation, and the multiatom non-local thermodynamic equilibrium inversion code STiC to obtain the model atmosphere that reproduces the Stokes profiles of Fe I 6302 Å and Ca II 8542 Å observed at the Swedish Solar Telescope, complementing the analysis with images taken in Hβ. Results. A penumbra formed only in the southern sector of a sunspot in AR 13010 on May 15, 2022. We find that a preexisting, strong (> 500 G) and nearly horizontal magnetic field in the chromosphere is a key prerequisite. Penumbral filaments grew southward where localized flux emergence was observed, accompanied by strong redshifts (2–3 km s−1) throughout the atmospheric layers. These redshifts likely represent field-aligned horizontal flows, possibly combining the onset of the Evershed flow and a siphon-like flow driven by field-aligned pressure gradients. Conclusions. Our findings indicate that penumbra formation results from both top-down and bottom-up processes. While strong horizontal chromospheric magnetic fields are a necessary boundary condition, flux emergence initiates filament growth. The fallen flux tube model is ruled out, based on clear bottom-up signatures, though it remains unclear why the chromospheric and photospheric fields become more horizontal concurrently. Extended spectropolarimetric data are needed to get a deeper insight.

• Ramos, A. Asensio; Löfdahl, M. G.; Baso, C. Díaz; Kuckein, C. & Pozuelo, S. Esteban (2025). torchmfbd: A flexible multi-object, multi-frame blind deconvolution code. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 703. doi: 10.1051/0004-6361/202555530. Full text in Research Archive Show summary

  Context . Post-facto image restoration techniques are essential for improving the quality of ground-based astronomical observations, which are affected by atmospheric turbulence. Multi-object, multi-frame blind deconvolution (MOMFBD) methods are widely used in solar physics to achieve diffraction-limited imaging. Aims . We present torchmfbd , a new open-source code for MOMFBD that leverages the PyTorch library to provide a flexible, GPU-accelerated framework for image restoration. The code is designed to handle spatially variant point spread functions (PSFs) and includes advanced regularization techniques. Methods . The code implements the MOMFBD method using a maximum a posteriori estimation framework. It supports both wavefront-based and data-driven PSF parameterizations, including a novel experimental approach using nonnegative matrix factorization. Regularization techniques, such as smoothness and sparsity constraints, can be incorporated to stabilize the solution. The code also supports dividing large fields of view into patches and includes tools for apodization and destretching. The code architecture is designed to become a flexible platform over which new reconstruction and regularization methods can also be straightforwardly implemented. Results . We demonstrate the capabilities of torchmfbd on real solar observations, showing its ability to produce high-quality reconstructions efficiently. The GPU acceleration significantly reduces computation time, making the code suitable for large datasets.

• Kuckein, C.; Collados, M.; Ramos, A. Asensio; Baso, C. J. Díaz; Felipe, T. & Noda, C. Quintero [Show all 9 contributors for this article] (2025). Inferring chromospheric velocities in an M3.2 flare using He I 1083.0 nm and Ca II 854.2 nm. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 699. doi: 10.1051/0004-6361/202554798. Full text in Research Archive Show summary

  Aims. Our aim was to study the chromospheric line-of-sight (LOS) velocities during the GOES M3.2 flare (SOL2013-05-17T08:43) using simultaneous high-resolution ground-based spectroscopic data of the He  I 10830 Å triplet and Ca  II 8542 Å line. A filament was present in the flaring area. Methods. The observational data were acquired with the Vacuum Tower Telescope (VTT, Tenerife, Spain) and covered the pre-flare, flare, and post-flare phases. Spectroscopic inversion techniques (HAZEL and STiC) were applied individually to He  I and Ca  II lines to recover the atmospheric parameters of the emitting plasma. Different inversion configurations were tested for Ca  II , and two families of solutions were found to explain the red-asymmetry of the profiles: a redshifted emission feature or a blueshifted absorption feature. These solutions could explain two different flare scenarios (condensation vs. evaporation). The ambiguity was solved by comparing these results to the He  I inferred velocities. Results. At the front of the flare ribbon, we observed a thin short-lived blueshifted layer. This is seen in both spectral regions, but is much more pronounced in He  I , with velocities of up to −10 km s −1 . In addition, at the front we found the coexistence of multiple He  I profiles within one pixel. The central part of the ribbon is dominated by He  I and Ca  II redshifted emission profiles. A flare-loop system, visible only in He  I absorption and not in Ca  II , becomes visible in the post-flare phase and shows strong downflows at the footpoints of up to 39 km s −1 . In the flare the Ca  II line represents lower heights compared to the quiet Sun, with peak sensitivity shifting from log τ ≃−5.2 to log τ ≃−3.5. The loop system's downflows persist for over an hour in the post-flare phase. Conclusions. The inferred LOS velocities support a cool-upflow scenario at the leading edge of the flare, with rapid transition from blueshifts to redshifts likely to occur within seconds to tens of seconds. Although the flare had a significant impact on the surrounding atmosphere, the solar filament in the region remained stable throughout all flare phases. The inclusion of the He  I triplet in the analysis helped resolve the ambiguity between two possible solutions for the plasma velocities detected in the Ca  II line. This highlights the importance of combining multiple chromospheric spectral lines to achieve a more comprehensive understanding of flare dynamics.

• Leenaarts, J.; Noort, M. van; Rodríguez, J. de la Cruz; Danilovic, S.; Baso, C. J. Díaz & Hillberg, T. [Show all 10 contributors for this article] (2025). High flow speeds and transition-region-like temperatures in the solar chromosphere during flux emergence: Evidence from imaging spectropolarimetry in He i 1083 nm and numerical simulations. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 696. doi: 10.1051/0004-6361/202453355. Full text in Research Archive Show summary

  Context. Flux emergence in the solar atmosphere is a complex process that causes a release of magnetic energy as heat and acceleration of solar plasma on a variety of spatial scales. Aims. We aim to investigate temperatures and velocities in small-scale reconnection episodes during flux emergence. Methods. We analyzed imaging spectropolarimetric data taken in the He  I 1083 nm line with a spatial resolution of 0.26, a time cadence of 2.8 s, and a spectral range corresponding to ±220 km s −1 around the line. This line is sensitive to temperatures higher than 15 kK, unlike diagnostics such as Mg  II h&k, Ca  II H&K, and H α , which lose sensitivity already at 15 kK. The He  I data is complemented by imaging spectropolarimetry in the Fe  I 617.3 nm and Ca  II 854.2 nm lines and imaging spectroscopy in Ca  II K and H α at a cadence between 12 s and 36 s. We employed inversions to determine the magnetic field and vertical velocity in the solar atmosphere. We computed He  I 1083 nm profiles from a radiation-magneto-hydrodynamics simulation of the solar atmosphere to help in the interpretation of the observations. Results. We find fast-evolving blob-like emission features in the He  I 1083 nm triplet at locations where the magnetic field is rapidly changing direction, and these are likely sites of magnetic reconnection. We fit the line with a model consisting of an emitting layer located below a cold layer representing the fibril canopy. The modeling provides evidence that this model, while simple, catches the essential characteristics of the line formation. The morphology of the emission in the He  I 1083 nm is localized and blob-like, unlike the emission in the Ca  II K line, which is more filamentary. Conclusions. The modeling shows that the He  I 1083 nm emission features and their Doppler shifts can be caused by opposite-polarity reconnection and/or horizontal current sheets below the canopy layer in the chromosphere. Based on the high observed Doppler width and the blob-like appearance of the emission features, we conjecture that at least a fraction of them are produced by plasmoids. We conclude that transition-region-like temperatures in the deeper layers of the active region chromosphere are more common than previously thought.

• Poquet, I. J. Soler; Baso, C. J. Díaz; Voort, Luc Rouppe van der & Vissers, G. J. M. (2025). Automatic detection of Ellerman bombs using deep learning. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 699. doi: 10.1051/0004-6361/202453524. Full text in Research Archive Show summary

  Context. Ellerman bombs (EBs) are observable signatures of photospheric small-scale magnetic reconnection events. They can be seen as intensity enhancements of the H α 6563 Å line wings and as brightenings in the SDO/AIA 1600 Å and 1700 Å passbands. Reliable automatic and systematic detection of EBs would enable the study of the impact of magnetic reconnection on the Sun’s dynamics. Aims. We aim to develop a method to automatically detect EBs in H α observations from the Swedish 1-m Solar Telescope (SST) and in SDO/AIA observations using the 1600 Å, 1700 Å, 171 Å, and 304 Å passbands. Methods. We trained models based on neural networks (NNs) to perform automatic detection of EBs. Additionally, we used different types of NNs to study how different properties – such as local spatial information, the spectral shape of each pixel, or the center-to-limb variation – contribute to the detection of EBs. Results. We find that for SST observations, the NN-based models are proficient at detecting EBs. With sufficiently high spectral resolution, the spatial context is not required to detect EBs. However, as we degrade the spectral and spatial resolution, the spatial information becomes more important. Models that include both dimensions perform best. For SDO/AIA, the models struggle to reliably distinguish between EBs and bright patches of different origin. Permutation feature importance revealed that the H α line wings (around ± 1 Å from line center) are the most informative features for EB detection. For the SDO/AIA case, the 1600 Å channel is the most relevant one when used in combination with 171 Å and 304 Å. Conclusions. The combination of the four different SDO/AIA passbands is not informative enough to accurately classify EBs. From our analysis of a few sample SDO/AIA 1600 Å and 1700 Å light curves, we conclude that inclusion of the temporal variation may be a significant step towards establishing an effective EB detection method that can be applied to the extensive SDO/AIA database of observations.

• Baso, Carlos José Díaz; Miliać, I.; Voort, Luc Rouppe van der & Schlichenmaier, R. (2025). Spectral resolution effects on the information content in solar spectra. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 693. doi: 10.1051/0004-6361/202452331. Full text in Research Archive Show summary

  Context. When interpreting spectropolarimetric observations of the solar atmosphere, wavelength variations in the emergent intensity and polarization translate into information on the depth stratification of physical parameters such as the temperature, velocity, and magnetic field. Resolving the fine details in the shapes of the spectral lines and their polarization gives us the capability to resolve small-scale depth variations in these physical parameters. With the advent of large-aperture solar telescopes and the development of state-of-the-art instrumentation, the requirements on spectral resolution have become a prominent question. Aims. We aim to quantify how the information content contained in a representative set of polarized spectra of photospheric spectral lines depends on the spectral resolution and spectral sampling of that spectrum. Methods. We used a state-of-the-art numerical simulation of a sunspot and the neighboring quiet Sun photosphere to synthesize polarized spectra of magnetically sensitive neutral iron lines. We then applied various degrees of spectral degradation to the synthetic spectra and analyzed the impact on its dimensionality using principal component analysis and the wavelength power spectrum using wavelet decomposition. Finally, we applied the Stokes Inversion based on Response functions (SIR) code to the degraded synthetic data to assess the effect of spectral resolution on the inferred parameters. Results. We find that the dimensionality of the Stokes spectra and the power contained in the small spectral scales significantly change with the spectral resolution. We find that regions with strong magnetic fields where convection is suppressed have more homogeneous atmospheres and produce less complex Stokes profiles. On the other hand, regions with strong gradients in the physical quantities give rise to more complex Stokes profiles that are more affected by spectral degradation. The degradation also makes the inversion problem more ill-defined, so inversion models with a larger number of free parameters overfit and give wrong estimates. Conclusions. The impact of spectral degradation in the interpretation of solar spectropolarimetric observations depends on multiple factors, including the spectral resolution, noise level, line spread function (LSF) shape, complexity of the solar atmosphere, and degrees of freedom in our inversion methods. To mitigate this impact, incorporating a good estimation of the LSF into the inversion process is recommended. Having a finely sampled spectrum may be more beneficial than achieving a higher signal-to-noise ratio per wavelength bin. Considering the inclusion of different spectral lines that can counter these effects, and calibrating the effective degrees of freedom in modeling strategies, are also important considerations. These strategies are crucial for the accurate interpretation of such observations and have the potential to offer more cost-effective solutions.

• Baso, Carlos José Díaz; Ramos, A. Asensio; Rodríguez, J. de la Cruz; Santos, J. M. da Silva & Voort, Luc Rouppe van der (2025). Exploring spectropolarimetric inversions using neural fields: Solar chromospheric magnetic field under the weak-field approximation. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 693. doi: 10.1051/0004-6361/202452172. Full text in Research Archive Show summary

  Context. Full-Stokes polarimetric datasets, originating from slit-spectrograph or narrow-band filtergrams, are routinely acquired nowadays. The data rate is increasing with the advent of bi-dimensional spectropolarimeters and observing techniques that allow long-time sequences of high-quality observations. There is a clear need to go beyond the traditional pixel-by-pixel strategy in spectropolarimetric inversions by exploiting the spatiotemporal coherence of the inferred physical quantities that contain valuable information about the conditions of the solar atmosphere. Aims. We explore the potential of neural networks as a continuous representation of the physical quantities over time and space (also known as neural fields), for spectropolarimetric inversions. Methods. We have implemented and tested a neural field to perform one of the simplest forms of spectropolarimetric inversions, the inference of the magnetic field vector under the weak-field approximation (WFA). By using a neural field to describe the magnetic field vector, we regularized the solution in the spatial and temporal domain by assuming that the physical quantities are continuous functions of the coordinates. This technique can be trivially generalized to account for more complex inversion methods. Results. We have tested the performance of the neural field to describe the magnetic field of a realistic 3D magnetohydrodynamic (MHD) simulation. We have also tested the neural field as a magnetic field inference tool (approach also known as physics-informed neural networks) using the WFA as our radiative transfer model. We investigated the results in synthetic and real observations of the Ca II 8542 Å line. We also explored the impact of other explicit regularizations, such as using the information of an extrapolated magnetic field, or the orientation of the chromospheric fibrils. Conclusions. Compared to traditional pixel-by-pixel inversions, the neural field approach improves the fidelity of the reconstruction of the magnetic field vector, especially the transverse component. This implicit regularization is a way of increasing the effective signal to noise of the observations. Although it is slower than the pixel-wise WFA estimation, this approach shows a promising potential for depth-stratified inversions, by reducing the number of free parameters and inducing spatiotemporal constraints in the solution.

• Poirier, Nicolas Aurelien Corentin; Danilovic, S.; Kohutova, Petra; Baso, Carlos José Díaz; Voort, Luc Rouppe van der & Calchetti, D. [Show all 7 contributors for this article] (2025). Coronal kink oscillations and photospheric driving: Combining SolO/EUI and SST/CRISP high-resolution observations. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 696. doi: 10.1051/0004-6361/202453025. Full text in Research Archive Show summary

  Context. The driving and excitation mechanisms of decay-less kink oscillations in coronal loops remain under active debate. The photospheric dynamics may provide the continuous energy supply required to sustain these oscillations. Aims. We aim to quantify and provide simple observational constraints on the photospheric driving of coronal loops in a few typical active region configurations: sunspot, plage, pores, and enhanced-network regions. We then aim to investigate the possible interplay between the photospheric driving and the properties of kink oscillations in the connected coronal loops. Methods. We analysed two unique datasets of the corona and photosphere taken at a high spatial and temporal resolution during the first coordinated observation campaign between Solar Orbiter and the Swedish 1-m Solar Telescope (SST). We applied a local correlation tracking method on the SST/CRISP data to quantify the photospheric motions at the base of coronal loops. The same loops were then analysed in the corona by exploiting data from the Extreme Ultraviolet Imager (EUI) on Solar Orbiter and by using a wavelet analysis to characterise the detected kink oscillations. Results. Each type of photospheric region shows varying dynamics but with an overall increase in strength going from pore to plage to enhanced-network to sunspot regions. Differences can also be seen in the amplitudes of the fundamental kink mode measured in the corresponding coronal loops. This suggests the photosphere is involved in the driving of coronal kink oscillations. However, the few samples available do not allow the excitation mechanism to be further established yet. Conclusions. Despite oscillating coronal loops being anchored in seemingly “static” strong magnetic field regions, as seen from coronal EUV observations, photospheric observations provide evidence for a continuous and significant driving at their base. The precise connection between photospheric driving and coronal kink oscillations remains to be further investigated. Upcoming coordinated observations between Solar Orbiter and ground-based telescopes will provide crucial additional observational constraints, with this pilot study serving as a baseline for future works. Finally, this study provides critical constraints on both the quasi-steady and broadband photospheric driving that can be tested in existing numerical models of coronal loops.

• Pozuelo, S. Esteban; Ramos, A. Asensio; Baso, Carlos José Díaz & Cobo, B. Ruiz (2024). Properties of sunspot light bridges on a geometric height scale. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 689. doi: 10.1051/0004-6361/202348181. Full text in Research Archive Show summary

  Context. Investigating light bridges (LBs) helps us comprehend key aspects of sunspots. However, few studies have analyzed the properties of LBs in terms of the geometric height, which is a more realistic perspective given the corrugation of the solar atmosphere. Aims. We aim to shed light on LBs by studying the variation in their physical properties with geometric height. Methods. We used the SICON code to infer the physical quantities in terms of the optical depth and the Wilson depression values of three LBs hosted by a sunspot observed with Hinode/SP in the Fe I 630 nm pair lines. We also used SIR inversions to cross-check the height variation of the field inclination in the LBs. In both output sets, we performed linear interpolation to convert the physical parameters from optical depth into a geometric height scale in each pixel. Results. Depending on their general appearance, we classified each LB as filamentary, grainy, or umbral. They appear as ridges that reach different maximum heights, with the umbral LB being the deepest. While the filamentary LB hosts a plasma inflow from the penumbra, the results for the grainy LB are compatible with an injection of hot plasma through convective cells of reduced field strength. Only a few positions reveal hints suggesting a cusp-like magnetic canopy. Moreover, strong gradients in the magnetic field strength and inclination usually exhibit enhanced electric currents, with the filamentary LB having remarkably strong currents that appear to be related to chromospheric events. Conclusions. The height stratification in filamentary and grainy LBs differ, indicating diverse mechanisms at work. Our results are in general incompatible with a magnetic canopy scenario, and further analysis is needed to confirm whether it exists along the entire LB or only at specific locations. Furthermore, this work assesses the usefulness of the SICON code when determining the height stratification of solar structures.

• Pietrow, A.G.M.; Kiselman, D.; Andriienko, O.; Roche, D. J. M. Petit dit de la; Baso, Carlos José Díaz & Calvo, F. (2023). Center-to-limb variation of spectral lines and continua observed with SST/CRISP and SST/CHROMIS. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 671. doi: 10.1051/0004-6361/202244811. Full text in Research Archive Show summary

  Context. Observations of center-to-limb variations (CLVs) of spectral lines and continua provide a good test for the accuracy of models with a solar and stellar atmospheric structure and spectral line formation. They are also widely used to constrain elemental abundances, and are becoming increasingly more important in atmospheric studies of exoplanets. However, only a few such data sets exist for chromospheric lines. Aims. We aim to create a set of standard profiles by means of mosaics made with the CRISP and CHROMIS instruments of the Swedish 1-m Solar Telescope (SST), as well as to explore the robustness of said profiles obtained using this method. Methods. For each spectral line, we used a mosaic that ranges from the center to the limb. Each of these mosaics were averaged down to 50 individual spectral profiles and spaced by 0.02 in the μ scale. These profiles were corrected for p-mode oscillations, and their line parameters (equivalent width, line shift, full-width at half-maximum, and line depth) were then compared against literature values whenever possible. Results. We present a set of 50 average profiles that are spaced equidistantly along the cosine of the heliocentric angle ( μ ) by steps of 0.02 for five continuum points between 4001 and 7772 Å, as well as ten of the most commonly observed spectral lines at the SST (Ca II H & K, Η β , Mg I 5173 Å, C I 5380 Å, Fe I 6173 Å, Fe I 6301 Å, Ha, O I 7772 Å, and Ca II 8542 Å). Conclusions. The CLV of line profiles and continua are shared in the CDS as machine readable tables, providing a quantitative constraint on theoretical models that aim to model stellar atmospheres.

• Baso, Carlos José Díaz; Voort, Luc Rouppe van der; Rodriguez, Jaime de la Cruz & Leenaarts, Jorrit (2023). Designing wavelength sampling for Fabry–Pérot observations: Information-based spectral sampling. Astronomy and Astrophysics (A & A). ISSN 0004-6361. 673. doi: 10.1051/0004-6361/202346230. Full text in Research Archive Show summary

  Context. Fabry–Pérot interferometers (FPIs) have become very popular in solar observations because they offer a balance between cadence, spatial resolution, and spectral resolution through a careful design of the spectral sampling scheme according to the observational requirements of a given target. However, an efficient balance requires knowledge of the expected target conditions, the properties of the chosen spectral line, and the instrumental characteristics. Aims. Our aim is to find a method that allows the optimal spectral sampling of FPI observations in a given spectral region to be found. The selected line positions must maximize the information content in the observation with a minimal number of points. Methods. In this study, we propose a technique based on a sequential selection approach in which a neural network is used to predict the spectrum (or physical quantities, if the model is known) from the information at a few points. Only those points that contain relevant information and improve the model prediction are included in the sampling scheme. Results. We have quantified the performance of the new sampling schemes by showing the lower errors in the model parameter reconstructions. The method adapts the separation of the points according to the spectral resolution of the instrument, the typical broadening of the spectral shape, and the typical Doppler velocities. The experiments that use the Ca  II 8542 Å line show that the resulting wavelength scheme naturally places more points in the core than in the wings (by almost a factor of 4), consistent with the sensitivity of the spectral line at each wavelength interval. As a result, observations focused on magnetic field analysis should prioritize a denser grid near the core, while those focused on thermodynamic properties would benefit from a larger coverage. The method can also be used as an accurate interpolator to improve the inference of the magnetic field when using the weak-field approximation. Conclusions. Overall, this method offers an objective approach for designing new instrumentation or observing proposals with customized configurations for specific targets. This is particularly relevant when studying highly dynamic events in the solar atmosphere with a cadence that preserves spectral coherence without sacrificing much information.

View all works in NVA

• Baso, Carlos José Díaz (2026). Information content of solar spectra. Full text in Research Archive Show summary

  High-resolution observations and advanced data analysis techniques are essential for uncovering fine-scale physical processes in the solar atmosphere. In this presentation, we explore three recent methodological and observational lessons learned from high-cadence solar telescope data. First, we examine the evolution of an eruptive flare observed with the TRIPPEL instrument at the Swedish 1-m Solar Telescope, demonstrating how combining magnetic extrapolations with non-local thermodynamic equilibrium inversions bridges observational and theoretical domains to quantify released magnetic energy and atmospheric heating. Second, we address the challenge of disentangling complex, multi-component spectral line profiles during mini-filament eruptions, showing how linear optimization techniques efficiently resolve overlapping atmospheric components. Finally, we introduce the application of normalizing flows as powerful density estimators to model multi-dimensional spectral probability distributions, enabling the automated detection of statistically rare and unusual phenomena in quiet Sun time-series datasets. Together, these approaches highlight how advanced modeling and machine learning techniques can optimize our analysis of massive observational data rates and reveal hidden solar dynamics.

• Baso, Carlos José Díaz (2026). AI in Solar Physics. Full text in Research Archive Show summary

  Solar physics is entering an unprecedented data-rich era driven by advanced solar observatories generating massive volumes of high-resolution spatiotemporal and spectropolarimetric observations. Analyzing these extensive datasets presents severe computational bottlenecks, as traditional numerical modeling and radiative transfer inversions are extremely time-consuming. In this work, we explore how artificial intelligence and modern machine learning frameworks can revolutionize solar data analysis across three critical domains. First, we implement probabilistic generative models, including diffusion models, to perform efficient unsupervised anomaly detection, allowing researchers to automatically isolate rare and physically interesting phenomena within terabytes of observations by evaluating data probability distributions. Second, we integrate large language model agents as automated optimizers to accelerate complex mathematical inversions, solving systems with hundreds of thousands of unknowns orders of magnitude faster. Finally, we examine the application of large-scale foundation models in heliophysics, such as spatiotemporal transformers trained on extensive Solar Dynamics Observatory archives, demonstrating their potential for zero-shot forecasting of solar conditions and the prediction of explosive events. Together, these approaches demonstrate how advanced AI methodologies help overcome the challenges of big data, opening new avenues for automated discovery and physical understanding in solar physics.

• Baso, Carlos José Díaz (2026). Inspectorch: Efficient rare event exploration. Full text in Research Archive Show summary

  Next-generation facilities like the European Solar Telescope (EST) will observe the Sun with unprecedented spatial and temporal resolution, providing simultaneous multi-wavelength and multi-height diagnostics. However, the sheer volume of high-resolution spectropolarimetric data will make it impossible to fully analyze using conventional methods. While standard machine learning identifies general trends, it routinely overlooks the highly localized, transient, or weak signals crucial for understanding chromospheric heating and flux emergence. Our aim is to introduce an unsupervised probabilistic method to efficiently identify these rare, extreme events in multidimensional solar observations, optimizing our computational resources for the phenomena EST is designed to uncover. To achieve this, we introduce Inspectorch, an open-source framework utilizing flow-based models. These flexible density estimators learn the multidimensional distribution of spectropolarimetric data and assign a probability to each sample, immediately isolating unusual events. We validate this approach using datasets from the Hinode Spectro-Polarimeter, the Interface Region Imaging Spectrograph (IRIS), the Microlensed Hyperspectral Imager (MiHI) at the Swedish 1-m Solar Telescope (SST), and space-based imagers such as the Atmospheric Imaging Assembly (AIA) on board the Solar Dynamics Observatory and the Extreme Ultraviolet Imager (EUI) on board Solar Orbiter. Across these datasets, the algorithm successfully isolates profiles with uncommon broadening, strong Doppler shifts associated with small-scale reconnection, and complex profiles exhibiting height-dependent magnetic polarity reversals. Its probabilistic nature makes it highly effective at identifying those cases that occupy a tiny fraction of the total number of samples and are overlooked by traditional methods. Inspectorch demonstrates that flow-based density estimation is robust, requires minimal parameter tuning, and can be applied to large datasets with minimal effort. By providing probabilistic anomaly scores, this publicly available framework prepares the solar physics community to rapidly identify and analyze the most physically relevant and elusive phenomena in the massive datasets soon to be produced by EST.

• Baso, Carlos José Díaz (2026). Thermodynamic and magnetic evolution of an eruptive C-class solar flare: Linking small-scale triggers to eruptive potential. Full text in Research Archive Show summary

  Solar flares and filament eruptions are complex phenomena driven by the release of magnetic energy, and they act as primary drivers of space weather. However, a large energy reservoir alone is insufficient to determine their eruptive potential; the localized magnetic topology and small-scale plasma dynamics play a critical role. We investigated the thermodynamic and magnetic properties of the solar atmosphere during the rise, peak, and decay phases of an eruptive C5.1-class flare in active region NOAA 12561 to understand how fine-scale atmospheric responses dictate broader eruptive behavior. High spatial and spectral resolution spectropolarimetric observations of the chromospheric Ca II 8542 Å line and nearby photospheric lines were obtained with the TRIPPEL-SP spectropolarimeter at the Swedish 1-m Solar Telescope. By combining nonlocal thermodynamic equilibrium (NLTE) inversions with non-force-free field (NFFF) magnetic extrapolations, we bridged the observational and theoretical domains to follow the event's evolution. Before the flare, our analysis revealed a complex, sheared magnetic topology with a high free energy content (~2 x 10^30 erg). In this precursor phase, we detected persistent localized heating (temperature increase of ~2000 K) with strong downflows (10-20 km/s) deep in the atmosphere. This heating was co-spatial with a bald-patch region, demonstrating how low-altitude magnetic reconnection can destabilize a larger filament structure. The flare's rise phase was marked by the filament's eruption with speeds exceeding 70 km/s. Following the eruption, free energy decreased by ~30% as post-flare loops channeled the released energy into the lower atmosphere, driving ribbon heating up to 8500 K and significant downflows. These results highlight how combining high-resolution ground-based spectropolarimetry with advanced modeling is essential for understanding how small-scale localized magnetic phenomena scale up to trigger eruptive events with space weather implications.

• Baso, Carlos José Díaz (2026). Inspectorch: a probabilistic framework for efficient rare event exploration in the EST era. Full text in Research Archive Show summary

  Next-generation facilities like the European Solar Telescope (EST) will observe the Sun with unprecedented spatial and temporal resolution, providing simultaneous multi-wavelength and multi-height diagnostics. However, the sheer volume of high-resolution spectropolarimetric data will make it impossible to fully analyze using conventional methods. While standard machine learning identifies general trends, it routinely overlooks the highly localized, transient, or weak signals crucial for understanding chromospheric heating and flux emergence. Our aim is to introduce an unsupervised probabilistic method to efficiently identify these rare, extreme events in multidimensional solar observations, optimizing our computational resources for the phenomena EST is designed to uncover. To achieve this, we introduce Inspectorch, an open-source framework utilizing flow-based models. These flexible density estimators learn the multidimensional distribution of spectropolarimetric data and assign a probability to each sample, immediately isolating unusual events. We validate this approach using datasets from the Hinode Spectro-Polarimeter, the Interface Region Imaging Spectrograph (IRIS), the Microlensed Hyperspectral Imager (MiHI) at the Swedish 1-m Solar Telescope (SST), and space-based imagers such as the Atmospheric Imaging Assembly (AIA) on board the Solar Dynamics Observatory and the Extreme Ultraviolet Imager (EUI) on board Solar Orbiter. Across these datasets, the algorithm successfully isolates profiles with uncommon broadening, strong Doppler shifts associated with small-scale reconnection, and complex profiles exhibiting height-dependent magnetic polarity reversals. Its probabilistic nature makes it highly effective at identifying those cases that occupy a tiny fraction of the total number of samples and are overlooked by traditional methods. Inspectorch demonstrates that flow-based density estimation is robust, requires minimal parameter tuning, and can be applied to large datasets with minimal effort. By providing probabilistic anomaly scores, this publicly available framework prepares the solar physics community to rapidly identify and analyze the most physically relevant and elusive phenomena in the massive datasets soon to be produced by EST.

• Baso, Carlos José Díaz (2026). Physics-informed machine learning and probabilistic methods for spectropolarimetry. Full text in Research Archive Show summary

  In the last decade, machine learning and neural networks have emerged as powerful tools for analyzing huge collections of solar data, demonstrating versatility in tasks ranging from image reconstruction to the acceleration of spectropolarimetric inversions. However, while these algorithms offer unprecedented speed, they often lack the precision required for detailed physical interpretation and fail to exploit the inherent spatial coherence of the solar atmosphere to increase fidelity. Moreover, the straightforward application of these tools risks merely shifting the bottleneck—from managing huge raw datasets to sifting through massive volumes of inversion results—without effectively isolating the poorly understood phenomena that drive scientific discovery. In this contribution, we address these limitations by reviewing novel applications that combine physical constraints with advanced statistical methods. We specifically highlight Neural Fields, which leverage continuous parameterization to impose spatio-temporal constraints and improve magnetic field reconstruction, and Normalizing Flows, a probabilistic technique optimized for efficiently detecting rare events and extreme spectra that traditional methods overlook. We demonstrate the potential of these approaches using observations from telescopes such as the Swedish 1-m Solar Telescope (SST) and the Interface Region Imaging Spectrograph (IRIS). Finally, we discuss how integrating these tools is essential to maximize the scientific return from current and future missions.

• Poquet, Ignasi Josep Soler; Baso, Carlos José Díaz & Voort, Luc Rouppe van der (2026). Multi-wavelength observations of Ellerman bombs. Full text in Research Archive Show summary

  Context. Ellerman bombs (EBs) are observational signatures of small-scale magnetic reconnection, key to understanding the lower solar atmosphere. While their role in active regions has been widely studied using the Hα line, near-ultraviolet (NUV) spectra routinely observed by the Interface Region Imaging Spectrograph (IRIS) offer a promising alternative for EB identification, enabling large-scale studies. Aims. We aim to identify the most important spectral signatures of EBs in the IRIS NUV spectra. With this, we seek to develop a robust criterion for their detection solely using the IRIS NUV spectra. In parallel, we determine the typical atmospheric stratification associated with EBs. Methods. We used four coordinated observations between the Swedish 1-m Solar Telescope (SST) and IRIS. Using the Hα line as a reference, we detected 18 different EBs and studied their associated IRIS NUV spectra. In addition, we used the IRIS2+ inversion tool to infer the temperature, line-of-sight velocity, and non-thermal broadening from the EB spectra. Results. The defining feature of EBs in the IRIS NUV is the enhancement of the wings of the subordinated Mg ii triplet between the Mg ii h&k lines. Inversions reveal that these signatures are produced by a localized temperature increase of ∆T ∼ 1670 K around log(τ)= −3.8. Using only the Mg ii triplet signatures, we found a detection criterion that successfully recovered 14 of 18 Hα-detected EBs. In addition, the shape of the Mg ii h&k lines in relation to the Mg ii triplet can serve as a proxy for the EB formation height. Conclusions. The NUV spectrum observed by IRIS is a good candidate for detecting EBs, opening the doors to large-scale studies across the extensive IRIS database, removing the dependence on Hα observations. This capability bridges the gap between small- and large-scale event studies, which was not possible before due to the lack of coordinated observations.

• Poquet, Ignasi Josep Soler; Baso, Carlos José Díaz & Voort, Luc Rouppe van der (2025). Ellerman bombs detection in SST and SDO using Deep Learning. Full text in Research Archive
• Baso, Carlos José Díaz (2024). Advances in machine learning applications on photospheric and chromospheric studies. Full text in Research Archive
• Poquet, Ignasi Josep Soler; Baso, Carlos José Díaz & Voort, Luc Rouppe van der (2024). Detecting Ellerman bombs with Deep Learning. Full text in Research Archive
• Baso, Carlos José Díaz (2024). Chromospheric Magnetic Field Reconstruction through Neural Field Assisted Spectropolarimetric Inversions. Full text in Research Archive
• Baso, Carlos José Díaz (2024). Information content of solar spectra: Insights from numerical simulations and neural field inversion techniques. Full text in Research Archive
• Baso, Carlos José Díaz (2024). The Force Awakens: Unleashing Deep Learning in Solar Physics. Full text in Research Archive
• Baso, Carlos José Díaz (2024). Deep Learning 2.0: Combining physics-based models with data-driven solutions. Full text in Research Archive
• Baso, Carlos José Díaz (2023). Differentiable programming for spectra modeling and inference. Full text in Research Archive
• Poquet, Ignasi Josep Soler; Voort, Luc H.M. Rouppe van der & Baso, Carlos José Díaz (2023). Interpretación de observaciones solares usando redes neuronales. Full text in Research Archive
• Poquet, Ignasi Josep Soler; Voort, Luc H.M. Rouppe van der & Baso, Carlos José Díaz (2023). Interpretation of Solar Observations with Deep Learning. Full text in Research Archive
• Baso, Carlos José Díaz (2023). Deep Learning 2.0: Combining physics-based models with data-driven solutions. Full text in Research Archive
• Baso, Carlos José Díaz (2023). Deep Learning 2.0: Combining physics-based models with data-driven solutions. Full text in Research Archive
• Baso, Carlos José Díaz (2023). Unleashing the Potential of Machine Learning for Efficient Analysis of Solar Observations. Full text in Research Archive
• Baso, Carlos José Díaz (2022). Unlocking the potential of deep learning for the analysis of spectropolarimetric observations. Full text in Research Archive
• Baso, Carlos José Díaz (2022). Webinar Lead Talk - Inteligencia artificial al servicio de la física solar. Full text in Research Archive

View all works in NVA