The effect at the heart of the study is called limb darkening, and it is exactly what the name suggests: a star appears fainter at its edge than at its center. This happens because stars are not uniformly bright disks. When we look toward the center of a stellar disk, our line of sight penetrates deeper into the star’s atmosphere, reaching hotter and brighter layers. Near the edge, the light we receive originates from shallower, cooler layers, and it must pass through more stellar material on its way out. The result is a gradual fade from a brilliant center to a dimmer rim, a signature that encodes the temperature and density structure of the star’s outer atmosphere.

Lead author Narsireddy Anugu, a staff scientist at Georgia State University’s CHARA Array, emphasized that the survey goes beyond simply sizing up stars. “We are not just measuring how large these stars are,” Anugu said. “We are measuring how their light is distributed across the stellar disk, which directly tests stellar-atmosphere models.” That distinction is crucial. Stellar radii can be measured in many ways, but the distribution of brightness across a stellar surface is a far more demanding test, one that probes the physics of how energy flows through a star’s outermost layers.

The CHARA Array itself is a marvel of engineering that makes such measurements possible. Located on Mount Wilson, the facility combines light from six telescopes positioned at different sites across the observatory grounds. By interfering the light together, CHARA achieves the resolving power of a much larger telescope, allowing astronomers to resolve details on the surfaces of stars that would otherwise remain forever blurred into points of light. Rather than capturing direct images, CHARA measures how the contrast of the interference pattern changes with the spacing between telescopes. These measurements reveal both the size of a star and how brightness changes across its surface, a technique that has made CHARA one of the most productive stellar-imaging facilities in the world.

Gail Schaefer, director of the CHARA Array, highlighted the facility’s unique capabilities in enabling the survey. “This study demonstrates the powerful capabilities of our facility,” Schaefer said. “By combining light from telescopes across the mountaintop, we can image stars with enough detail to see what their surfaces actually look like.” The observations were conducted by observing the same stars simultaneously through two different near-infrared filters, allowing the team to measure how the center-to-edge fading changes with wavelength. This dual-wavelength strategy is what transformed the survey from a collection of stellar portraits into a quantitative test of atmospheric physics.

The sample consisted of 31 bright stars in late stages of their lives, stars whose outer layers have expanded outward far beyond the dimensions they had during the prime of their existence. These evolved subgiant, giant, and supergiant stars provide especially valuable tests of stellar-atmosphere models because the large convective motions churning through their extended atmospheres can produce complex brightness profiles. Unlike the Sun, whose surface is a relatively thin layer governed by well-understood radiative processes, these puffed-up stars have atmospheres where convection dominates, making them natural laboratories for testing whether theoretical models can capture the messy reality of stellar surfaces.

The quantitative result is striking. Across the sample, the limb-darkening strength decreased by about 38 percent going from near-infrared wavelengths at 1.6 microns to 2.2 microns. The atmosphere models tested in the study predicted the same overall trend, weaker limb darkening at longer infrared wavelengths, but with a smaller decrease of only about 17 to 22 percent. In other words, the models capture the broad behavior correctly, but they do not fully reproduce how strongly the wavelength dependence plays out in real stars. The discrepancy means that current models may underestimate how stellar brightness changes across a star’s surface, a systematic error that could propagate into downstream measurements that astronomers rely on every day.

Why does this matter so much? Limb darkening directly affects measurements of stellar diameters, since the apparent edge of a star depends on how its brightness fades. More dramatically, it impacts how astronomers characterize exoplanets that transit across the surface of their parent star. When a planet crosses in front of a star, the amount of light blocked during the transit depends on the brightness across the stellar disk. A planet passing in front of the bright center blocks more light than one skimming the dimmer edge. If models misrepresent that brightness distribution, the inferred sizes and other properties of transiting exoplanets inherit the error. With thousands of confirmed exoplanets and missions like TESS and the upcoming PLATO survey depending on precise transit modeling, an empirical correction to limb darkening is a gift to the exoplanet community.

The survey also delivered a reassuring null result: the team found no evidence for surface features such as large starspots or hidden companion stars anywhere in the sample. That cleanliness matters, because spots and companions can mimic or mask limb-darkening signatures, and their absence confirms that the measured wavelength dependence is a genuine atmospheric effect rather than contamination from surface asymmetries or stellar multiplicity. The result strengthens the case that the discrepancy with models is real and systematic, not an artifact of a few unusual stars.

Looking ahead, the team plans to extend the comparison to broader wavelength coverage, stretching from visible light into the near-infrared, which will further constrain how well models reproduce the full spectral behavior of stellar atmospheres. The researchers also intend to image smaller main-sequence stars, which are far more difficult to resolve but are especially important for understanding transiting exoplanets around Sun-like stars, the primary targets in the search for habitable worlds. For now, the CHARA Array’s survey of 31 fading disks stands as a reminder that even the most familiar objects in the sky still hold surprises, and that pushing observational precision to its limits remains one of the most reliable ways to find where our theories fall short.

Subject of Research: Near-infrared limb darkening measurements of evolved stars using optical interferometry

Article Title: Georgia State Telescope array reveals how nearby stars fade from center to edge

Article References: Georgia State Telescope array reveals how nearby stars fade from center to edge. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: CHARA Array, limb darkening, stellar atmospheres, optical interferometry, evolved stars, near-infrared astronomy, exoplanet transits, stellar diameters, Mount Wilson, supergiant stars, stellar convection, The Astronomical Journal

Radiation transfer is, in essence, the calculation of how emitted light travels through a medium before reaching an observer. In a simplified one-dimensional model, an atmosphere can be represented as a stack of horizontal layers whose properties vary smoothly with depth. Real atmospheres are less orderly. Stars contain rising hot granules and sinking cooler material, magnetic structures, winds, and large-scale convective cells. Exoplanets have day-night temperature contrasts, circulation, clouds, and externally supplied stellar radiation. Kilonovae consist of rapidly expanding, chemically complex ejecta whose geometry and physical state evolve. A three-dimensional model retains variations in all spatial directions, while a time-dependent treatment can follow changes in the gas rather than assuming a permanent steady state.

The NLTE part addresses a second limitation of standard modeling. Local thermodynamic equilibrium assumes that collisions dominate the internal energy distribution of atoms and molecules, allowing their populations to be estimated with Saha-Boltzmann statistics from the local temperature and density. That assumption often fails near an open surface, where photons escape and the radiation field can control excitation and ionization. In NLTE calculations, the population of each energy level is determined by balancing radiative and collisional transitions. The radiation field affects those populations, while the populations in turn alter the opacity and emissivity that shape the radiation field. Solving the problem therefore requires repeated, coupled calculations rather than a single local evaluation.

At the center of the calculation is the radiative-transfer equation for the specific intensity, the amount of radiation traveling in a particular direction at a particular frequency. In a common time-independent form, the change in intensity along a ray depends on the difference between the intensity and the source function, which is the ratio of emissivity to extinction. The calculation integrates this relationship over optical depth, a measure of how opaque the material is. In three dimensions, the code must trace many rays through a spatial grid, interpolate temperature, density, velocity, opacity, and source-function values onto each photon path, and then combine the directional intensities to obtain the mean radiation field. That mean field enters the rate equations governing atomic and molecular populations.

The review emphasizes that numerical choices can influence the answer as much as the underlying physics. Long-characteristics methods follow rays through an entire model and preserve sharp spectral structures well, but they can be expensive. Short-characteristics methods connect neighboring layers and are easier to parallelize, although interpolation can diffuse intense beams. Higher-order interpolation can improve accuracy but may create artificial overshoots, including unphysical negative opacities or intensities. Monotonic schemes suppress those artifacts but may sacrifice accuracy or complicate convergence. The angle quadrature, which selects and weights the rays used to integrate the radiation field, also matters. Relatively few directions may be sufficient for calculating mean intensities in statistical-equilibrium equations, whereas emergent line profiles and centre-to-limb variations generally require more.

One practical compromise is the so-called 1.5D approach. Each vertical column in a three-dimensional atmospheric simulation is treated as an independent one-dimensional atmosphere, preserving the local temperature, density, and velocity structure but ignoring horizontal radiation exchange between columns. The resulting fluxes can then be averaged across the model. This approach can greatly reduce computational demands and has produced useful results for several stellar problems, especially when the photon mean free path is short. However, it is not universally reliable. In extremely metal-poor stars, for example, the review describes cases in which 1.5D and full 3D NLTE calculations produced abundance differences as large as 0.22 dex for iron lines. The approximation must therefore be tested against the specific diagnostic and physical regime.

The scientific payoff is clearest in stellar spectroscopy. Convection gives spectral lines distinctive asymmetric shapes and Doppler shifts because rising and sinking gas contribute different amounts of light at different velocities. Three-dimensional models reproduce observed line bisectors more successfully than traditional hydrostatic one-dimensional models in several comparisons, while predicted convective shifts can reach hundreds of metres per second. Such effects matter for radial-velocity measurements, chemical-abundance studies, and efforts to separate stellar surface variability from planetary signals. Centre-to-limb observations provide another stringent test. For the solar oxygen line near 7772 angstroms, the review reports that 1D LTE models can overestimate the inferred abundance by 0.6 dex when the limb is analyzed, whereas 3D NLTE calculations offer a way to account for the changing geometry and radiation field.

These corrections extend directly to exoplanet research. During a transit, a planet blocks different portions of its host star, each with its own brightness, velocity, magnetic activity, and spectral-line shape. The resulting Rossiter-McLaughlin signal can reveal the projected alignment between stellar rotation and the planetary orbit, but it can also be distorted by inaccurate models of the stellar surface. Studies summarized in the review find that 3D NLTE treatment improves some diagnostics, including those based on sodium and potassium lines, although other features remain to be explored. Three-dimensional and NLTE methods are also being adapted to irradiated planetary atmospheres, where the host star supplies an external radiation field and can drive photoionization, photodissociation, heating, and atmospheric escape. Clouds and global circulation add further spatial complexity.

Kilonovae present a different but equally demanding challenge. Their spectra arise from rapidly expanding material produced in compact-object mergers, with radioactive decays supplying energy and heavy elements providing dense forests of spectral transitions. Expanding shells are often modeled with spherical symmetry and escape-probability approximations, but their composition, velocity structure, and ionization state can evolve rapidly. The review places such systems within the broader push toward time-dependent, multidimensional NLTE calculations, while noting that direct spatially resolved tests are not currently available for exoplanet or kilonova photospheres. Progress will require reliable atomic and molecular data, including transition probabilities, photoionization cross-sections, collision rates, and information for complex ions and molecules. It will also require algorithms that balance physical fidelity with the immense cost of solving millions of coupled radiation and population equations. The central message is not that every observation needs the most elaborate possible model, but that astronomers must understand when common simplifications introduce systematic errors. As high-resolution spectrographs, transit surveys, and time-domain observatories deliver increasingly precise data, 3D NLTE radiation transfer provides a framework for turning subtle spectral details into more dependable knowledge of some of the universe’s most dynamic environments.

A useful distinction in these calculations is between the radiation field inside a model and the observables ultimately compared with data. A simulation can predict an angle- and frequency-dependent specific intensity at the surface, as well as a flux integrated over directions. The intensity contains information about viewing angle and spatial structure, whereas the integrated flux provides a spectral energy distribution for the object as a whole. This difference is important for phenomena such as stellar surface inhomogeneities, where two observers may receive different line profiles from the same model depending on which regions are visible.

The transfer calculation is also only one part of a larger physical modeling chain. A model must first specify or compute the state of the gas, including quantities such as density, temperature, velocity, and composition. Radiation transfer then uses detailed opacities and emissivities on a finer frequency grid to produce a more realistic spectrum than the coarse radiative description used in many underlying fluid or atmosphere calculations. In NLTE work, the sequence is not strictly one-way: the radiation field changes the energy states of atoms and molecules, and those changed populations modify the opacity and emissivity. Iteration is therefore needed until the matter and radiation descriptions become mutually consistent.

The relevant microscopic data can be a major source of uncertainty. The review stresses the need for atomic and molecular information alongside fluid dynamics, statistical mechanics, and energy transport. Rates for radiative and collisional processes determine how strongly particles respond to the radiation field, while the available transitions establish which frequencies can absorb or emit. These inputs become especially consequential when spectra are used to infer detailed chemical abundances. A numerical solution may be internally converged yet still inherit systematic limitations from incomplete or inaccurate physical data.

Geometry is not an optional refinement in every regime. Multidimensional treatment becomes necessary when the structure encountered by a photon varies substantially across space or changes non-monotonically along its path. This criterion can apply to convective stellar surfaces, strongly irradiated atmospheres, or expanding merger ejecta. The review consequently treats 3D NLTE as a family of coupled problems rather than a single universal algorithm. Different systems demand different compromises among spatial resolution, frequency coverage, angular sampling, time dependence, and the representation of matter-radiation coupling.

These methodological issues connect radiation-transfer modeling to several broader observational programs. In stellar studies, synthetic spectra help constrain chemical evolution, convection, magnetism, and mass loss. In exoplanet work, they support interpretation of transit and atmosphere measurements. For kilonovae, evolving spectra can provide clues to the composition and physical state of rapidly changing ejecta. The review also places related applications in a wider computational landscape that includes interstellar-medium diagnostics, circumstellar polarization, dusty galaxy discs, active-galaxy accretion discs, and supernova modeling. Across these settings, the value of greater realism is measured by whether it changes an inferred physical parameter or resolves a discrepancy with observations, not simply by the number of dimensions in the calculation.

Subject of Research: Three-dimensional non-local thermodynamic equilibrium radiation transfer in astrophysical atmospheres

Article Title: 3D NLTE radiation transfer: theory and applications to stars, exoplanets, and kilonovae

Article References: Bergemann, M., & Hoppe, R. (2026). 3D NLTE radiation transfer: theory and applications to stars, exoplanets, and kilonovae. Living Reviews in Computational Astrophysics, 12(1), Article 7. https://doi.org/10.1007/s41115-026-00029-3

Image Credits: AI Generated

DOI: 10.1007/s41115-026-00029-3

Keywords: radiative transfer, 3D modeling, NLTE, stellar atmospheres, exoplanets, kilonovae, spectroscopy, astrophysical simulations, radiation, transfer, theory, applications