The Transit Light Source Effect: How Starspots Fake Exoplanet Atmosphere Signals

What the transit light source effect is

Transmission spectroscopy measures how deep a planet's transit is at each wavelength. The transit depth is D = (R_p/R_star)², so if the planet's atmosphere is more opaque at some wavelength (say a water band), the planet blocks slightly more light there and the transit looks deeper. Astronomers turn a spectrum of transit depths into a spectrum of the atmosphere.

The hidden assumption is that the light shining through the planet's atmosphere is spectrally identical to the star's disk-integrated light, which serves as the reference. The transit light source effect is the failure of that assumption. A real stellar surface is heterogeneous: dark, cool magnetic starspots and bright, hot faculae speckle the disk. The planet only crosses one narrow chord, so its true light source is a local mixture of photosphere and spots that generally differs from the whole-disk average.

• Cool spots are redder, so unocculted spots make the star look bluer along the chord.
• Hot faculae push the opposite way.
• Because spot/facula contrasts vary with wavelength, the error is chromatic — exactly the signature we hunt for in atmospheres.

The mechanism and the governing equation

Rackham, Apai & Giampapa (2018) formalized the effect with a wavelength-dependent contamination factor ε(λ) that multiplies the true transit depth:

D_obs(λ) = ε(λ) · D_true(λ)

For a star whose spots along the transit chord differ from the disk-averaged spot coverage, the factor is

ε(λ) = [ (1 − f_spot)·S_phot(λ) + f_spot·S_spot(λ) ] / [ (1 − F_spot)·S_phot(λ) + F_spot·S_spot(λ) ]

Here f_spot is the spot covering fraction in the transit chord, F_spot is the disk-integrated spot fraction, and S_phot, S_spot are the photosphere and spot spectra. If the chord and disk have the same spot mix, ε = 1 and there is no distortion. When unocculted spots dominate the disk (F_spot > f_spot), the reference is artificially dimmed at short wavelengths, so ε rises toward the blue and spots inflate transit depths toward shorter wavelengths. Faculae reverse the slope. The key insight: because most of a rocky planet's atmospheric signal is only ~10–100 parts per million, even a percent-level spot fraction can dominate.

Characteristic numbers and a worked example

Spot temperature contrasts scale roughly with the photosphere. Rackham et al. adopt a relation where an M-dwarf spot sits near T_spot ≈ 0.86 · T_phot — so for a ~2560 K late-M dwarf, spots are only ~250–360 K cooler, while for the Sun (5800 K) spots run ~1500 K cooler.

• TRAPPIST-1 (M8V, T ≈ 2560 K): the modeled spot covering fraction is f_spot ≈ 8% (+18/−7) and facula fraction f_fac ≈ 54% (+16/−46). These heterogeneities alter transit depths by roughly 1–15× the strength of the planetary features across 0.3–5.5 μm.
• Density bias: because contamination changes the apparent radius, TRAPPIST-1 planet densities are underestimated by about 3% (−3, +3/−8%), nudging inferred bulk compositions toward too much water/volatiles.

A concrete worst case: if a 1.4 μm water feature in a rocky planet is ~50 ppm, an unocculted spot field on an active M dwarf can add a spurious slope-plus-bump of several hundred ppm — more than 10× the real signal — so a spotless planet can masquerade as a wet one.

How it shows up in real observations

The TLS effect appears as a smooth, sloped, or molecularly structured contribution to the transmission spectrum that cannot be removed by simply monitoring the star's brightness variability. Rackham et al. showed that variability amplitude only tracks the asymmetry of spot coverage, not the total immaculate (fully spotted) baseline — so corrections tied to the light curve's amplitude systematically underestimate contamination.

• Spot-crossing events: if the planet transits directly over a spot, a bump appears mid-transit — a distinct, occulted-spot signature separate from the unocculted-spot TLS slope.
• Molecular fingerprints: cool spot spectra carry water and TiO/VO bands, so contamination can mimic atmospheric H₂O near 0.9 and 1.4 μm — the very bands JWST NIRISS/NIRSpec target.
• Multi-epoch drift: because spot coverage evolves over a rotation (days) and activity cycle (years), the same planet yields different "atmospheres" on different nights — a red flag for contamination.

HST WFC3 and now JWST spectra of M-dwarf worlds routinely fit a stellar-contamination component alongside any planetary model.

How it differs from related effects

Several activity systematics are easy to confuse:

• TLS (unocculted spots/faculae): features outside the transit chord bias the disk-integrated reference. Chromatic, present even in a perfectly clean chord, and the focus of Rackham et al.
• Spot-crossing (occulted spots): the planet crosses a feature inside the chord, producing an in-transit brightening bump; it distorts the light-curve shape and radius, but is time-localized and often modelable.
• Stellar flares: transient brightening, not a persistent spectral slope.
• Instrumental slopes / clouds & hazes: aerosols also mute or slope a spectrum, but they are grey-to-blue in a way that can, with care, be distinguished from the redder spot signatures.

The regime also matters: inactive FGK dwarfs produce little detectable TLS contamination, whereas active M and K dwarfs — precisely the small, bright-contrast stars best for finding habitable-zone rocky planets — are the worst offenders. This is the cruel irony of TLS: the easiest planets to detect orbit the hardest stars to characterize.

Significance, famous cases, and open questions

The transit light source effect reframed a decade of exoplanet atmosphere claims around cool stars. It explained why some reported "features" drifted between epochs and warned that JWST's exquisite precision would be limited not by photon noise but by how well we know the host star's surface.

• TRAPPIST-1 is the canonical case: multiple JWST programs must jointly fit stellar heterogeneity to interpret whether planets b–h have atmospheres at all.
• GJ 1214 b and sub-Neptunes around M dwarfs face the same ambiguity between real hazes and spot contamination.

Open questions include: What is the true immaculate spot filling factor of quiet M dwarfs? Photometry can't measure it directly. Are spot and facula spectra well modeled? Real active regions aren't single-temperature blackbodies. And can multi-wavelength, multi-epoch data plus stellar models break the degeneracy? Tools like spectral-decontamination pipelines and high-resolution spot-mapping (e.g., AU Mic studies) are pushing toward robust corrections, but the consensus is that stellar contamination, not the telescope, now sets the floor for small-planet atmospheric characterization.

Stellar contamination by host-star type: how strongly the transit light source effect distorts transmission spectra.

Host star (type, Teff)	Spot temperature contrast	Nominal spot covering fraction	TLS contamination vs. planet signal
Sun-like G2V (~5800 K)	Spots ~1500 K cooler	Small, ~0.03–1% (quiet Sun)	Minor for inactive stars; up to ~few× if active
K dwarf (~4500 K)	Spots ~800–1200 K cooler	~1–3% typical, up to ~10s%	Can exceed planet signal in active cases
Early M dwarf (~3500 K)	Spots ~200–400 K cooler	Few % nominal, up to ~10–40%	Often 2–10× the atmospheric feature
Late M / TRAPPIST-1 (M8V, ~2560 K)	T_spot ≈ 0.86 × T_phot	f_spot ≈ 8%, f_fac ≈ 54%	~1–15× planetary signal; density biased ~3%