betaFeedback

WorksPhysics and Astronomy

Model Choice or Model Bias? How stellar spectral libraries and activity modelling choices influence Ariel spectra- I. Forward model

Abstract

The Ariel mission will conduct a population-level survey of approximately 1000 exoplanet atmospheres across its nominal lifetime, requiring unprecedented consistency in the treatment of stellar contamination of transmission spectra. We present a systematic comparison of stellar spectral libraries and modelling approaches for computing facular and spot contamination spectra across the full Ariel wavelength range (0.5-7.8 microns) for six stellar spectral types spanning F3 to M2. Using a surface resolved forward model with approximately 200 randomly generated active region configurations per spectral type, we compute contamination spectra at Ariel Tier 3 resolution and compare three 1D spectral libraries against 3D MHD stellar atmosphere spectra as a physical ground truth. We find that library-dependent differences in the facular treatment introduce systematic biases exceeding the Ariel Tier 3 sensitivity threshold of approximately 20 ppm across most of the wavelength range, for most spectral types and activity levels considered. These biases align with the absorption bands of water, carbon dioxide, carbon monoxide, and methane, rendering them indistinguishable from genuine atmospheric signals at Ariel's precision. Our results demonstrate hat the choice of stellar spectral library, in particular the faculae treatment is a critical and currently underappreciated source of systematic error for Ariel atmospheric characterisation. The impact on atmospheric retrievals is quantified in a companion paper.

Introduction

Since the discovery of the first exoplanet around a Sun-like star in the mid-1990s, the field has advanced from detection to detailed atmospheric characterization. Out of over 6,000 known exoplanets, the James Webb Space Telescope (JWST) has already probed the atmospheres of roughly 100 targets using transmission spectroscopy [36, 8, 42, 46, 35]. As a planet transits its host star, wavelength-dependent variations in atmospheric opacity alter the observed transit depth, encoding information about atmospheric composition. However, disentangling planetary signatures from stellar spectra remains challenging due to stellar surface heterogeneities, brought about by stellar magnetism. Occulted features crossed by the planet leave distinct imprints in transit light curves (e.g., [1, 12, 32] (Petit dit de la Roche et al. 2024)), whereas unocculted active regions drive the Transit Light Source Effect (TLSE; [30]). Standard corrections rely on disc-averaged representations of active regions ([30] (Rackham et al. 2019)) or joint atmospheric retrievals ([28, 18, 39]). Additional stellar biases, such as fixing limb-darkening coefficients or adopting inadequate limb-darkening laws, can introduce errors of ∼3%\sim3\% in inferred planetary radii [10] (Espinoza & Jordán 2016) and wavelength-dependent biases exceeding 100 ppm [19] (Columbre et al. 2024). The Ariel mission [43, 41], launching in the early 2030s, will conduct a population-level survey of ∼1000\sim1000 transiting exoplanets. Ariel’s tiered observing strategy involves repeated visits to high-priority targets, making epoch-to-epoch consistency essential. Drawing lessons from JWST is vital to fully leverage these multi-visit observations. Recently, Thompson et al. [40] demonstrated that uncorrected inter-visit stellar variability introduces significant biases in retrieved atmospheric parameters. To balance computational efficiency, scalability, and interpretability, they adopt a simplified 1D stellar activity model that is widely used in the retrieval context. Spots and faculae are treated as separate, distinct components, with each active region represented by a spectral energy distribution (SED) from the PHOENIX BT-Settl library [4] characterised by its temperature contrast. However, such models neglect center-to-limb variations (CLV) and cannot reproduce the more complex temperature and spectral structure of faculae. As shown by Nèmec et al. [26] in the solar context, disc-averaged models fail to reproduce true contamination signals, underscoring the need for physically motivated facular models incorporating CLV. Recent 3D radiative magnetohydrodynamic (MHD) simulations combining MURaM [44, 33] and MPS-ATLAS (Witzke et al. [45]) offer first-principles models of stellar atmospheres that account for interactions among convection, magnetic fields, and radiation (Kostogryz et al. [21], hereafter K26). Applying magnetic fields above 100 G (associated with faculae) reveals that faculae brighten hot F-type stars but weaken toward cooler stars, even appearing dark in the coolest models (K26). Furthermore, standard 1D and 3D CLV coefficients in forward models can produce transit depth discrepancies up to 3% in common photometric bands (Nèmec et al., submitted). In this work, we investigate how different treatments of faculae and choices of stellar atmosphere models propagate into transmission spectra. We compare simple TLSE prescriptions against surface-resolved active region modeling using StarSim (Herrero et al. [16]; Rosich et al. [34]) across publicly available spectral libraries, focusing on Ariel’s full spectral range (0.5–7.8 μ\mum) at Tier 3 resolution. Section 2 describes the spectral libraries, Section 3 details the spatial modeling of active regions, and Section 4 presents the model comparison.

Spectral libraries

We note that the choice of libraries was based on public availability and therefore we by no means claim the following list to be complete.

BT-Settl

The BT-Settl models (Allard [4]) are based on the PHOENIX code (Hauschildt et al. [14]). Spectra are calculated in 1D spherical geometry and are given only as a disc averaged flux. Since the models were developed also for ultra-cool dwarfs, the models include molecular lines (H2O, NH3, CH4, TiO, VO, CO, CO2, CN, FeH, CrH, MgH, CaH) and atomic linelists from combined databases (VALD, Kurucz Atomic Line Lists, NIST Atomic Spectra Database). Additionally, the model allows for 1D hydrodynamic cloud formation and settling which is required to accurately model ultra-cool dwarfs, although we note that the spectral types we consider in this work are too hot for clouds to form.

PHOENIX NewEra

The Hauschildt et al. [13] NewEra model grid follows the setup of the commonly used Husser et al. [17] 1D LTE stellar atmospheres library. The atmospheres are calculated with the PHOENIX code V20 in spherical geometry. The equation of state is described and solved through the Astrophysical Chemical Equilibrium Solver (ACES; Barman et al. [5]), treating 839 different species (84 elements, 289 ions, 249 molecules, and 217 condensates). Atomic line opacities are taken from the Kurucz database Kurucz (2005a,b). Molecular line opacities are primarily drawn from the ExoMol database Tennyson et al. [38], cut to ∼20.5 billion lines for the temperature range of the grid. Since the release of the Husser models over a decade ago, the molecular line lists have been substantially revised, in some scenarios introducing noticeable spectral differences, especially at optical wavelengths and for cool stars. These spectral differences arise despite minimal changes to the underlying atmospheric temperature–pressure structures. For the complete list of molecular lines, we refer to Hauschildt et al. [13] Table 1 and tbl:2. In the present work we utilised both the PHOENIX-NewEraV2-LowRes-SPECTRA which reach up to 1.6 microns, sufficient to cover the Ariel wavelength regime we utilise the PHOENIX-NewEraV2-JWST-SPECTRA.

Spectral typePhotospheric Temperature [K]Spot Temperature [K]Faculae Temperature [K]log g
F36800426068804.0
G25780395058204.4
K05250376052804.4
K44550350045704.6
M03750312037504.8
M23560303035205.0

Table 1. Spectral types and temperatures for the photosphere, faculae and spot component and log g considered in this work. Log g, photospheric and faculae temperatures were taken from K26 and correspond to their ’SSD’ and ’200G’ case. The spot temperatures were calculated using Eq. (3).

MPS-ATLAS

We utilise two different Merged Parallelised Simplified (MPS)-ATLAS Witzke et al. [45] spectral libraries in the context of the present work that use the same base radiative transfer (RT) code but rely on different constructions of the underlying atmosphere. The MPS-ATLAS code is based on the ATLAS Kurucz [22] and DFSYN-THE Kurucz (2005b); Castelli [6] codes, including an extension in its ability to treat optimum opacity distribution functions (Cernetic et al. [7]).

Kostogryz et al. [20] have employed the MPS-ATLAS code to compute an extensive stellar atmosphere grid in plane-parallel geometry, and provided both specific intensity Iλ(μ)I_\lambda(\mu) and disc-integrated values. Note that the disc-integrated values were calculated by disc-integrating the specific intensities. The atomic and molecular line lists were taken from Kurucz (2005b).

More recently, Kostogryz2026 have presented a small grid of magnetized stellar atmospheres, where the atmospheric structures used in the RT calculations are computed with MURaM, which solves the conservative form of the MHD equations for a compressible, partially ionized plasma to simulate stellar atmospheres. In particular, it models convection self-consistently and is able to take into account initial magnetic field seeds which are important to simulate faculae.

Forward model

For the forward model we use two different approaches. One is the classical TLSE effect, neglecting the spatial distribution of magnetic features and working with different temperatures for photosphere, faculae and spots. The spectrum I(λ)I(\lambda) can be written as

I(λ)=I(λ)phot⋅αphot+I(λ)fac⋅αfac+I(λ)spot⋅αspot,(1)I(\lambda) = I(\lambda)_{\mathrm{phot}} \cdot\alpha_{\mathrm{phot}} + I(\lambda)_{\mathrm{fac}} \cdot\alpha_{\mathrm{fac}} + I(\lambda)_{\mathrm{spot}} \cdot\alpha_{\mathrm{spot}}, \tag*{(1)}

where α\alpha and I(λ)I(\lambda) are the disc filling factors and corresponding temperatures of the 3 distinct features, respectively. Another, though more complex approach is to model for a spotted (with spots and faculae) rotating photosphere built from the integration of the spectral contribution of a fine grid of surface elements. Such an approach is taken by the StarSim (Herrero et al. [16]; Rosich et al. [34]) framework. While the current StarSim version allows for combined modeling of photometric and spectral variability, we incorporate the extensions of the model by Porqueras-León et al. [29] and Stucki in prep. which allow the calculation of the transmission spectrum and usage of the 3D models. A new version of the code, including those extensions will be released soon (Gomes et al., in prep.).

Similarly to Eq. (1), one could write the resulting spectrum as

Ijk(λ)=Ijk(λ)phot⋅αphotjk+Ijk(λ)fac⋅αfacjk+Ijk(λ)spot⋅αspotjk,(2)I^{jk}(\lambda) = I^{jk}(\lambda)_{\mathrm{phot}} \cdot\alpha^{jk}_{\mathrm{phot}} + I^{jk}(\lambda)_{\mathrm{fac}} \cdot\alpha^{jk}_{\mathrm{fac}} + I^{jk}(\lambda)_{\mathrm{spot}} \cdot\alpha^{jk}_{\mathrm{spot}}, \tag*{(2)}

with Ijk(λ)XI^{jk}(\lambda)_{\mathrm{X}} being the spectra of feature X either from the 3D library directly or the 1D MPS-ATLAS intensities with the corresponding temperatures from Table 1 and αXjk\alpha^{jk}_{\mathrm{X}} being the disc coverage of feature X at a surface element with coordinates (j,k)(j,k).

We allow for 1 to 10 magnetic features (1 features consists of spot and faculae pair) and for each number of features, we randomly construct 20 different surfaces, each with randomly drawn size distributions, longitudes, and latitudes. Spot sizes are drawn randomly, and a corresponding facula patch is assigned to each spot with an area equal to a randomly drawn integer multiplier between 1 and 3 times the spot area. A multiplier of 1 therefore corresponds to equal spot and facula coverage, while a multiplier of 3 means the facular area is 3 times larger than the associated spot. Since a multiplier of 1 results in a facular area equal to the spot area, the StarSim implementation currently defaults to discarding the faculae area. In that case, a new facular region is drawn independently with a size in the same range as the spot, ensuring that both components are always represented in the surface configuration. This approach is used to calculate the StarSim-3D (meaning the combination of StarSim and the 3D-MHD spectra) and StarSim-1D (meaning StarSim together with the 1D-MPS-ATLAS-CLV) spectra. All calculations are based on a stellar inclination of 90∘90^\circ. The projected disc filling factors of spots αsp\alpha_{\mathrm{sp}} and faculae αfc\alpha_{\mathrm{fc}} obtained from each StarSim realisation are then inserted into Eq. (1) together with the spectra from the BT-Settl, MPS-ATLAS, and NewEra libraries at the corresponding component temperatures, yielding a disc-average contamination spectrum for each library and each surface configuration.

For the spectral types considered, we focus on the 3D MHD MPS-ATLAS grid as our ‘ground-truth’. Consequently, we choose stellar atmospheres with similar effective temperatures, metallicities, and log gg values for the other spectral libraries, assuming solar metallicity. For the faculae component, we use the spectra computed for a surface magnetic field strength of 200G in Kostogryz2026, which is representative of the average field strength of solar faculae. The temperatures associated with these magnetized atmosphere calculations are listed in Table 1. The spot temperatures are calculated with respect to the photospheric temperature following Herbst et al. [15]:

Tspot=−3.58⋅10−5Tstar2+1.0188⋅Tstar+239.3(3)T_{\mathrm{spot}} = -3.58 \cdot10^{-5}T_{\mathrm{star}}^2 + 1.0188 \cdot T_{\mathrm{star}} + 239.3 \tag*{(3)}

where TstarT_{\mathrm{star}} is taken to be the photospheric temperature. We note, that since there are no 3D MHD spot spectra available, we revert to using the 1D spectra in that case, noting that those spectra have known problems reproducing the contrast of the 3D spectra. There are no MPS-ATLAS library spectra below Teff=3500T_{\mathrm{eff}} = 3500K. For spot temperatures below this limit (M0 and M2 in Table 1), we rescale the 3500K spectrum by the ratio of the Planck function at the target temperature to that at 3500K, computed at each wavelength. While this preserves the overall spectral shape, it does not capture the changes in molecular band strengths at lower temperatures and should be regarded as an approximation. We emphasize that since Eq. (3) is based on observations, the temperatures are a mix of the umbral and penumbral component that make up spots and we therefore do not distinguish between those two components in the present work.

Results of the forward model exercise

We generated roughly 200 spectra using the StarSim code for each spectral type considered. The heterogeneous stellar spectra are subsequently binned to the Ariel Tier 3 spectral resolution using the ArielRad radiometric simulator (Mugnai et al. [25]) and TauREx3 ([3, 2]).

The set of spectra is shown in Figure 1. We normalized each spectrum to their respective photospheric spectrum. Intensities over 1 therefore correspond to an excess brightening of the star caused by faculae, dimming is observed when the spot spectral contribution exceeds that of the faculae. Visually, some differences are clearly noticeable: for the G2 star, there is a steeper slope for higher intensities for all models compared to the 3D-MHD ones, while the calculations for the K-dwarfs appear relatively consistent. Once the cooler M-dwarfs are considered, differences become more pronounced. We quantify the differences later in this section.

Forward modelled stellar spectra

Figure 1. Forward modelled stellar spectra as observed at Ariel Tier 3 spectral resolution with the different approaches. For more information on the different approaches representing each column we refer to the text.

In Figure 2 and Figure 3 we colourcode the disk coverage of spots and faculae respectively. The spot filling factors reach up to 40%, and even up to 90% for the faculae (mostly in cases where the spot filling factors are small). For all spectra, it is evident that with increasing spot coverage, the normalized intensity becomes smaller, meaning that there is net-darkening compared to the photosphere. However, if the faculae area coverages are considered, the trends are less clear. Generally, high faculae filling factors can counteract intermediate values of spot area coverages, and both a combination of low spot and low to intermediate faculae areas can lead to brightening with respect to the photosphere, specifically at wavelengths below 1 micron. In the case of the M0 and M2 calculations, high faculae filling factors also lead to net-darkening, which is counterintuitive, since ‘faculae’ are usually associated with brightening. However, as discussed in K26 the effect of adding magnetic fields (which in this case are associated with faculae) becomes less pronounced for stars cooler than spectral classes of M0. While K26 presented this result in the form of CLV in dependence of μ\mu for selected wavelengths, we calculate the contrast across the full Ariel wavelength regime. The contrast is defined as the difference between the photosphere and the considered magnetic region (faculae or spots), normalized to the photosphere. In this nomenclature, contrasts above 1 mean net-brightening with respect to the photosphere and below 1 darkening.

Same spectra as shown in Fig. 1 with the projected spot area filling factors colorcoded.

Figure 2. Same spectra as shown in Figure 1 with the projected spot area filling factors colorcoded.

Same as Fig. 2 but for the projected faculae area filling factors.

Figure 3. Same as Figure 2 but for the projected faculae area filling factors.

The contrasts are shown in Figure 4 for the spots and Figure 5 for the faculae. We note, that we only chose the MPS-ATLAS flux spectra in this comparison, and that the conclusions drawn from this experiment are the same for the BT-Settl and the NewEra spectra.

Contrasts of the spot spectra to the photospheric spectra for 3 selected spectral libraries

Figure 4. Contrasts of the spot spectra to the photospheric spectra for 3 selected spectral libraries. Colorcoded is the μ\mu-angle for the StarSim 3D MPS-ATLAS spectra and the StarSim 1D MPS-ATLAS CLV spectra.

Figure 5. Same as Figure 4 but for the faculae contrasts.

We note, that the MPS-ATLAS-1D CLV spectra have a coarser μ\mu-grid compared to the MPS-ATLAS-3D spectra, particularly towards the limb, as the 3D spectra cut-off at μ=0.1\mu= 0.1. The spot contrast is very similar, independent of whether a 3D or 1D quiet spectrum was used, and the MPS-ATLAS-flux spectra are most similar to the CLV spectra at μ=0.5\mu= 0.5. This is consistent with the fact that the MPS-ATLAS-flux spectra were calculate from the MPS-ATLAS-1D CLV spectra. It is in the faculae component where the differences are striking. Again, the MPS-ATLAS-1D CLV flux spectra are similar to the μ=0.5\mu= 0.5 values. Across the spectral types considered the MPS-ATLAS-3D faculae show a clear, non-linear trend with increasing contrast towards the limb, in particular with a strong slope shortward of 3 microns. As discussed in K26 the impact of 200G magnetic fields in the 3D calculations is decreasing with decreasing effective temperatures, with the smallest contrast still associated with brightening for the M0 calculations. Even though the resulting photospheric and facular temperatures are the same according to Table 1 shows that for M0, there is still a difference in the contrast for the 3D calculations, which is because of the highly complex interplay of the magnetic fields in those simulations with both the granulation and the radiation field, which cannot be taken into account in 1D approaches.

The brightening towards the limb is explained by the ‘hot wall’ model (Spruit [37]): magnetic flux concentrations cause a local depression of the (τ=1\tau= 1) surface, exposing hot lateral granulation walls. At disc centre the line-of-sight of the observer passes vertically through the depression, where the magnetic flux evacuates the flux tube (analogous to the process occurring within star spots), revealing cooler atmospheric layers. Towards the limb, the steeper viewing angle makes the hot walls increasingly visible, producing the characteristic brightening. Crucially, the brightening is a result of magnetic fields and geometry, both effects that cannot be captured by 1D models. As discussed in K26 the magnetic suppression of convection dominates over the effect of hot walls and faculae appear locally dark for stars below M0 spectral classes. The comparison presented in Figure 5 highlights the need of including proper, physically motivated faculae brightening (or in specific instances, dimming) in forward models of stellar contamination.

Lastly, we emphasize the differences between the models by treating the MPS-ATLAS-3D spectra as the ‘ground-truth’ and calculate the differences between this and the other libraries. This is shown in Figure 6, where we additionally added colored vertical lines for several molecular features (blue: H2_2O, red: CO2_2, orange: CO, and green: CH4_4) that are predominately found in the atmospheres of both (sub) Neptunes and hot Jupiters, populations that represent the predominant targets of the Ariel mission.

Differences between the other libraries and the 3D-MHD spectra

Figure 6. Similar as Figure 1 but this time we calculated the difference between the other libraries and the 3D-MHD spectra. The blue shaded vertical lines indicate H2_2O, red shaded CO2_2, orange shaded CO, and green shaded CH4_4 features.

Conclusions and future work

We have shown that across a range of stellar spectral types, from F3 to M2, it is insufficient to describe faculae as a spectrally hotter version of the photosphere. The facular contrast (defined as the ratio of the facular to photospheric spectrum) is not only highly wavelength dependent, but also shows a strong dependence on the μ\mu-angle, i.e. the position of the facula on the stellar disc, with both dependencies being functions of the stellar effective temperature.

Treating the 3D MHD MPS-ATLAS spectra as the physical ground truth, the differences between those forward models spectra and those predicted by the 1D libraries (BT-Settl, MPS-ATLAS-1D, and NewEra) exceed the Ariel Tier 3 sensitivity threshold of ∼200\sim200 ppm across most of the wavelength range considered (0.5–7.8 μ\mum), for most spectral types, and most activity levels considered. The differences are particularly pronounced for hotter stars and decrease towards cooler spectral types, with the M2 calculations showing the counter-intuitive result that faculae cause a net dimming rather than brightening. The differences between spectral libraries are large enough to mimic planetary atmospheric signals in several relevant

bands (we considered H2_2O, CO2_2, CO, and CH4_4). As noted, fixed limb-darkening coefficients introduce errors in the inferred planetary radius at the level of ∼3%\sim3\%, and an inadequate limb-darkening law introduces wavelength-dependent biases exceeding 100 ppm ([19] (Coloumb et al. 2024)). The library-dependent facular contrasts we find are of comparable magnitude, and in several cases substantially larger, yet they arise from a modelling choice, namely the representation of faculae as a hotter photosphere, that is still routinely made without question. It is therefore crucial that both forward modelling exercises and machine learning approaches that rely on stellar models (e.g. [29]) to incorporate physically motivated facular CLV. The immediate next step, which forms the subject of Paper II of this series, is to quantify the impact of model and spectral library mismatches on atmospheric retrievals, following a similar approach to [40], and to assess the extent of any biases in retrieved atmospheric compositions, specifically those arising from the simplified treatment of faculae.

Acknowledgements

This publication has been made possible by the spanish grants PID2021-125627OB-C31 funded by MCIU/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”, PID2020-120375GB-I00 funded by MCIU/AEI, by the programme Unidad de Excelencia María de Maeztu CEX2020-001058-M, by the Generalitat de Catalunya/CERCA programme, by the SGR 01526/2021, the European Research Council (ERC) under the European Union’s Horizon Europe programme (ERC Advanced Grant SPOTLESS; no. 101140786, ERC Synergy Grant REVEaL grant No. 101118581) and the Marie Skłodowska-Curie Actions grant agreement No 101149286 (INCITE, and the ASI-INAF agreement 2021-5-HH.2-2024.) A.T. acknowledges support from the ASI-INAF agreement 2021-5-HH.2-2024.

CONFLICT OF INTEREST

Authors declare no conflict of interest.

DATA AVAILABILITY

This work has utilized publicly available open source software such as the StarSim (https://github.com/dbarochlopez/starsim/tree/main) and Taurex3 (https://github.com/ucl-exoplanets/taurex3). The PHOENIX NewEra spectra are hosted by https://www.fdr.uni-hamburg.de/record/17612, the MPS-ATLAS 1D spectra by https://edmond.mpg.de/dataset.xhtml?persistentId=doi:10.17617/3.NJ56TR and the MPS-ATLAS 3D by https://edmond.mpg.de/dataset.xhtml?persistentId=doi:10.17617/3.FBTIYY.

References

  1. [1]Ahrer, E.-M., Radica, M., Piaulét-Goraveb, C., et al. 2025, ApJ, 985, L10, doi: 10.3847/2041-8213/add010DOI
  2. [2]Al-Refaie, A. F., Changeat, Q., Venot, O., Waldmann, I. P., & Tinetti, G. 2022, ApJ, 932, 123, doi: 10.3847/1538-4357/ac6dcdDOI
  3. [3]Al-Refaie, A. F., Changeat, Q., Waldmann, I. P., & Tinetti, G. 2021, ApJ, 917, 37, doi: 10.3847/1538-4357/ac0252DOI
  4. [4]Allard, F. 2014, in IAU Symposium, Vol. 299, Exploring the Formation and Evolution of Planetary Systems, ed. M. Booth, B. C. Matthews, & J. R. Graham, 271–272, doi: 10.1017/S1743921313008545
  5. [5]Barman, T. S., Madhusudhan, N., Konopacky, Q. M., & Marois, C. 2011, ApJ, 733, 65, doi: 10.1088/0004-637X/733/1/65DOI
  6. [6]Castelli, F. 2005, Memorie della Societa Astronomica Italiana Supplementi, 8, 34
  7. [7]Cernetic, M., Shapiro, A. I., Witzke, V., et al. 2019, A&A, 627, A157, doi: 10.1051/0004-6361/201935723DOI
  8. [8]Charbonneau, D., Brown, T. M., Noyes, R. W., & Gilliland, R. L. 2002, ApJ, 568, 377, doi: 10.1086/338770DOI
  9. [10]Espinoza, N., & Jordán, A. 2015, MNRAS, 450, 1879, doi: 10.1093/mnras/stv744DOI
  10. [12]Fournier-Tondreau, M., Pan, Y., Morel, K., et al. 2025, MNRAS, 539, 422, doi: 10.1093/mnras/staf489DOI
  11. [13]Hauschildt, P. H., Barman, T., Baron, E., Aufdenberg, J. P., & Schweitzer, A. 2025, A&A, 698, A47, doi: 10.1051/0004-6361/202554171DOI
  12. [14]Hauschildt, P. H., Baron, E., & Starrfield, S., & Allard, F. 1996, ApJ, 462, 386, doi: 10.1086/177160DOI
  13. [15]Herbst, K., Papaioannou, A., Arapietian, V. S., & Atri, D. 2021, ApJ, 907, 89, doi: 10.3847/1538-4357/abcc04DOI
  14. [16]Herrero, E., Ribas, I., Jordi, C., et al. 2016, A&A, 586, A131, doi: 10.1051/0004-6361/201425369DOI
  15. [17]Husser, T.-O., Wende-von Berg, S., Dreizler, S., et al. 2013, A&A, 553, A6, doi: 10.1051/0004-6361/201219058DOI
  16. [18]Iyer, A. R., & Line, M. R. 2020, ApJ, 889, 78, doi: 10.3847/1538-4357/ab612eDOI
  17. [19]Keers, R. E., Shapiro, A. I., Kostogryz, N. M., et al. 2024, ApJ, 977, L7, doi: 10.3847/2041-8213/ad8b51DOI
  18. [20]Kostogryz, N., Shapiro, A., Witzke, V., et al. 2023, Research Notes of the AAS, 7, 39
  19. [21]Kostogryz, N., Shapiro, A. I., Witzke, V., et al. 2026, arXiv e-prints, arXiv:2606.21912, doi: 10.48550/arXiv.2606.21912DOI
  20. [22]Kurucz, R. L. 1970, SAO Special Report, 309
  21. [25]Mugnai, L. V., Pascale, E., Edwards, B., Papageorgiou, A., & Sarkar, S. 2020, Experimental Astronomy, 50, 303, doi: 10.1007/s10686-020-09676-7
  22. [26]Němec, N.-E., Porqueras-León, Ó., Ribas, I., & Shapiro, A. I. 2026, A&A, 705, A111, doi: 10.1051/0004-6361/202557325DOI
  23. [28]Pinhas, A., Rackham, B. V., Madhusudhan, N., & Apai, D. 2018, MNRAS, 480, 5314, doi: 10.1093/mnras/sty2209DOI
  24. [29]Porqueras-León, Ó., Perger, M., Morales, J. C., et al. 2026, arXiv e-prints, arXiv:2609.35721, doi: 10.48550/arXiv.2609.35721
  25. [30]Rackham, B. V., Apai, D., & Giampapa, M. S. 2018, ApJ, 853, 122, doi: 10.3847/1538-4357/aaa08cDOI
  26. [32]Radica, M., Piauleth-Goraveb, C., Taylor, J., et al. 2025, ApJ, 979, L5, doi: 10.3847/2041-8213/ada381DOI
  27. [33]Rempel, M. 2014, ApJ, 789, 132, doi: 10.1088/0004-637X/789/2/132DOI
  28. [34]Rosich, A., Herrero, E., Mallonn, M., et al. 2020, A&A, 641, A82, doi: 10.1051/0004-6361/202037586DOI
  29. [35]Rustamkulov, Z., Sing, D. K., Mukherjee, S., et al. 2023, Nature, 614, 659, doi: 10.1038/s41586-022-05677-yDOI
  30. [36]Seager, S., & Sasselov, D. D. 2000, The Astrophysical Journal, 537, 916, doi: 10.1086/309088
  31. [37]Spruit, H. C. 1976, Sol. Phys., 50, 269, doi: 10.1007/BF00155292DOI
  32. [38]Tennyson, J., Yurchenko, S. N., Al-Refaie, A. F., et al. 2016, Journal of Molecular Spectroscopy, 327, 73, doi: 10.1016/j.jms.2016.05.002
  33. [39]Thompson, A., Biagini, A., Cracchiolo, G., et al. 2024, ApJ, 960, 107, doi: 10.3847/1538-4357/ad0369DOI
  34. [40]Thompson, A. H., Changeat, Q., Luftinger, T., & Tinetti, G. 2026, arXiv e-prints, arXiv:2610.08128 https://arxiv.org/abs/2610.08128
  35. [41]Tinetti, G., Eccleston, P., Leuftinger, T., et al. 2022, in European Planetary Science Congress, EPSC2022–1114, doi: 10.5194/epsc2022-1114
  36. [42]Tinetti, G., Vidal-Madjar, A., Liang, M.-C., et al. 2007, Nature, 448, 169, doi: 10.1038/nature06002DOI
  37. [43]Tinetti, G., Drossart, P., Eccleston, P., et al. 2018, Experimental Astronomy, 46, 135, doi: 10.1007/s10686-018-9598-x
  38. [44]Vögler, A., Shelyag, S., Schüssler, M., et al. 2005, A&A, 429, 335, doi: 10.1051/0004-6361:20041507DOI
  39. [45]Witzke, V., Shapiro, A. I., Cernetic, M., et al. 2021, A&A, 653, A65, doi: 10.1051/0004-6361/202140275DOI
  40. [46]Yan, F., Casasayas-Barris, N., Molaverdikhani, K., et al. 2019, A&A, 632, A69, doi: 10.1051/0004-6361/201936396DOI

Paper details

vlnc:W_2yrfzdnn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents