M dwarf convection can fake exoplanet atmospheric signals, simulations show

A research team led by Andrea Chiavassa used 3D radiation-hydrodynamics simulations to show that turbulent convection on M dwarf stars can imprint false or distorted atmospheric signals onto high-resolution spectroscopy of small, temperate exoplanets, according to a paper posted to arXiv on 3 Nov 2026 and accepted for publication in Astronomy and Astrophysics.

M dwarfs are small, cool stars with masses from 0.08 to 0.65 solar masses, radii from 0.10 to 0.65 solar radii, bolometric luminosities roughly 3 times 10 to the minus 4 to 8 times 10 to the minus 2 solar luminosities, effective temperatures about 2700 to 4100 Kelvin, and surface gravities log g around 4.6 to 5.1.

Those properties make M dwarfs the most prolific hosts for terrestrial exoplanets, and many of them harbor rocky planets; the TRAPPIST-1 system, for example, contains seven rocky planets. GJ 3470 is a bright early M1.5 dwarf with a Ks magnitude of 7.99, a slow rotation of 1.1 plus or minus 0.1 kilometers per second, and it hosts a Neptune-size companion.

GJ 3470 b is a moderately eccentric warm Neptune on a polar orbit with an escaping atmosphere; its atmosphere was detected from space by Hubble and JWST, but no ground-based detection exists yet. The benchmark observations used in the study were taken on January 27, 2020, with the GIANO-B spectrograph on the 3.6-meter Telescopio Nazionale Galileo as part of the GAPS program, spanning orbital phases from minus 0.017 to 0.027 around mid-transit.

GIANO-B covers 0.95 to 2.45 micrometers at a resolving power of 50,000, a range matching the expected ANDES instrument on the Extremely Large Telescope, and the team extracted the spectra with the GOFIO v3.0 pipeline. The team ran the Stagger code, a compressible radiation-(magneto)hydrodynamic code, to calculate the M dwarf convection simulation, using the opacity binning method with 12 opacity bins and assuming a solar chemical composition.

The total simulated time was about 2 hours to capture several convective turnovers, and radiative transfer was solved along nine rays totaling eighteen directions. The team computed synthetic spectra with the Optim3D code at a constant resolving power of 200,000 from 300 to 2500 nanometers, covering visible and infrared wavelengths, producing 120 synthetic spectra at 60-second steps over the two-hour simulation.

Convection on M dwarf surfaces appears as stochastic granulation patterns of downflowing cooler plasma and bright rising hot plasma, causing non-stationary flux that shifts or distorts spectral lines during transits. During a modeled transit of GJ 3470 b, the combined amplitude of centre-to-limb variations and the Rossiter-McLaughlin effect is comparable to the exoplanet’s own atmospheric signal. The researchers processed a high-resolution transit sequence using principal component analysis, a statistical method that separates overlapping signals, to remove stellar trends.

Principal component analysis can in principle remove stellar convection effects from transit data, but its performance depends heavily on whether telluric and planetary signals are present together and on the observational noise budget. The team tested PCA on their modeled GJ 3470 b transit sequence and found it could recover the planet without clear bias from stellar leftovers, but only within a narrow wavelength range and one specific orbital setup, limiting how broadly researchers can apply the method.

Current water-vapor line lists remain inaccurate, showing only modest agreement in line strength and position within the chosen K-band interval (a near-infrared wavelength band) when researchers compare synthetic spectra with actual observations. The researchers used GIANO-B observations of GJ 3470 as a benchmark to test molecular line data, and inaccurate H2O line lists mean atmospheric retrievals relying on these lists could misinterpret the strength or location of water features in M dwarf spectra.

The study concludes that 3D radiation-hydrodynamics simulations are essential for accurately modeling convection-driven spectral variability in M dwarfs and for obtaining unbiased measurements from future Extremely Large Telescope observations. Researchers consider high-resolution spectroscopy a key ingredient for future exoplanet observations with the Extremely Large Telescope, enabling planet studies in reflected light and potential biomarker detection in nearby rocky exoplanet atmospheres.

The ELT will use instruments METIS, HARMONI, ANDES, and PCS in that order to study planets around M dwarfs, and high-resolution spectroscopy enables unambiguous molecular detection through line-by-line matching and measures Doppler shifts from planetary motion and winds. An independent 2024 grid of non-magnetic 3D hydrodynamical CO5BOLD model atmospheres found that convection has a limited impact on M-dwarf atmospheric structures and observable properties at solar metallicity, though effects are more pronounced at [M/H] = minus 2.

In the CO5BOLD grid, the optically thick layers of 1D hydrostatic models failed to replicate the 3D thermodynamic structures regardless of convection parametrization, and 3D-to-1D abundance corrections for analyzed spectral lines were below 0.01 dex (one decimal order of magnitude, or a factor of ten), indicating negligible convective influence on line formation at the studied grid points.

The newer work models a specific Neptune transit with 3D radiation-hydrodynamics and principal component analysis, whereas the earlier grid compared 3D and 1D hydrostatic atmospheres across M0 to M5 parameters at solar and sub-solar metallicities, finding convection’s imprint small except at [M/H] = minus 2 where outer layers cool by several hundred kelvin.

TOI-4616 b is an Earth-sized planet transiting a nearby mid-M4 dwarf and serves as a benchmark for studying terrestrial planets around M dwarfs, reported by a team led by doctoral researcher Francis Zong Lang at the University of Bern’s Center for Space and Habitability. TOI-4616 is about 91 light-years away, with a stellar radius of about 0.1889 solar radii, a mass of about 0.1881 solar masses, and a temperature of about 3150 Kelvin.

TOI-4616 b has a radius of 1.22 Earth radii, an orbital period of 1.55 days, and an equilibrium temperature of around 525 Kelvin, placing it in an extreme irradiation regime for an Earth-sized planet around a mid-M dwarf. “TOI-4616 b constitutes a valuable benchmark system for comparative studies of terrestrial planets around mid-M dwarfs”

Rocky planets around M dwarfs typically orbit close to their stars, exposing them to powerful radiation and flaring that can strip atmospheres, and M dwarfs take up to one or two billion years to reach the main sequence with higher early luminosity. Thicker carbon-dioxide-dominated atmospheres might resist dissipation, secondary atmospheres could form through volcanism and outgassing after the main sequence, and strong magnetic fields may help planets retain atmospheres, though no clear conclusions exist yet, and researchers say more study is needed.

Many M-dwarf Earth-like planets are poor JWST targets due to imprecise stellar measurements or few transit observations. Stellar convection on M dwarfs can produce spectral distortions comparable in strength to exoplanet atmospheric signals, which threatens unbiased habitability assessments for small temperate worlds targeted by JWST and future ELT instruments.