Radiative Transfer

Why radiative transfer matters

Almost everything we know about the universe beyond the Solar System arrives as light, and light never reaches us untouched. Between the source and the detector it is absorbed, re-emitted, and scattered by gas, dust, and plasma. Radiative transfer is the theory that decodes this — the bridge between what a star or nebula is and what its spectrum looks like. Without it, a spectrum is just a wiggly line; with it, that line yields temperature, pressure, density, chemical abundance, magnetic field, velocity, and even the age of a stellar population.

• Stellar atmospheres. Effective temperature, surface gravity, and elemental abundances are all extracted by matching model spectra to observed line strengths — pure radiative transfer.
• Photospheres and limb darkening. The τ ≈ 2/3 rule defines the visible surface of every star and explains why the solar disk fades toward its edge.
• Line formation. The 100,000+ absorption lines in the solar spectrum are each a resolved radiative-transfer problem in a specific atomic transition.
• Interstellar medium & nebulae. Emission and absorption along a sightline reveal column densities of H I, CO, and dust.
• Exoplanet atmospheres. Transmission spectroscopy is transfer through a thin annulus of atmosphere backlit by the host star.
• The cosmic microwave background. The CMB is the radiation field that decoupled when the universe's optical depth to Thomson scattering fell below one at recombination.

How it works, step by step

Follow a pencil-thin beam of radiation of frequency ν as it crosses a slab of gas of thickness ds along its direction of travel.

1. Define the specific intensity. I_ν is the energy carried per unit time, area, solid angle, and frequency (units W m⁻² Hz⁻¹ sr⁻¹). Unlike flux, intensity is conserved along a ray in empty space — it does not fall off with distance — which is exactly why it is the natural variable for transfer.
2. Account for losses. Over ds the beam loses dI_ν = −α_ν I_ν ds, where α_ν is the absorption (extinction) coefficient in m⁻¹, combining true absorption and scattering out of the beam.
3. Account for gains. The same slab emits, adding dI_ν = +j_ν ds, where j_ν is the emission coefficient (W m⁻³ Hz⁻¹ sr⁻¹). Combine both: dI_ν/ds = −α_ν I_ν + j_ν.
4. Switch to optical depth. Define dτ_ν = −α_ν ds, measured inward from the observer. Dividing through by −α_ν gives the clean form dI_ν/dτ_ν = I_ν − S_ν, where S_ν ≡ j_ν/α_ν is the source function. Optical depth, not physical distance, is the natural ruler: it counts photon mean free paths.
5. Read off the two limits. If I_ν > S_ν the beam is brighter than the local emission, so it loses net energy; if I_ν < S_ν it is dimmer and gains. Radiation always relaxes toward the source function.
6. Write the formal solution. Integrating along the ray gives the emergent intensity I_ν(0) = ∫₀^∞ S_ν(t) e^(−t) dt. The exponential e^(−t) is the fog of optical depth: deeper layers contribute, but exponentially attenuated. You see an emission-weighted average of S_ν, dominated by the region near τ ≈ 1.
7. Locate the photosphere. Expand S_ν linearly in τ, S_ν(τ) ≈ a + bτ, and the integral collapses to the Eddington–Barbier result I_ν(0) ≈ a + b = S_ν(τ = 1) for a normal ray, and S_ν(τ ≈ 2/3) for the disk-integrated flux. That depth is the layer we actually observe.
8. Form the lines. At a frequency where a bound-bound transition boosts α_ν, the τ = 2/3 surface floats higher into cooler gas with a smaller B_ν(T). The core of the line is therefore darker than the continuum — a Fraunhofer absorption line.

The equation of transfer and its symbols

The compact equation of transfer, valid for a static, plane-parallel or arbitrary geometry once τ is measured along the ray, is:

dI_ν / dτ_ν = I_ν − S_ν

with the source function defined as S_ν = j_ν / α_ν and optical depth by dτ_ν = −α_ν ds. Its along-the-ray formal solution between optical depths τ₁ (source) and τ₂ (observer) is:

I_ν(τ₂) = I_ν(τ₁) e^{−(τ₁−τ₂)} + ∫ S_ν(t) e^{−(t−τ₂)} dt

Symbol	Meaning	Units
Iν	Specific (monochromatic) intensity	W m⁻² Hz⁻¹ sr⁻¹
ν	Frequency of the radiation	Hz
αν	Absorption / extinction coefficient (opacity per length)	m⁻¹
jν	Emission coefficient	W m⁻³ Hz⁻¹ sr⁻¹
τν	Optical depth (mean free paths along the ray)	dimensionless
Sν = jν/αν	Source function	W m⁻² Hz⁻¹ sr⁻¹
Bν(T)	Planck blackbody function (Sν in LTE)	W m⁻² Hz⁻¹ sr⁻¹
Jν	Mean intensity (Sν for pure scattering)	W m⁻² Hz⁻¹ sr⁻¹
μ = cos θ	Cosine of angle between ray and outward normal	dimensionless

Kirchhoff's law of thermal radiation (Gustav Kirchhoff, 1859) fixes the ratio in thermal equilibrium: j_ν/α_ν = B_ν(T). A good absorber at a frequency is an equally good emitter there — which is why the same atomic transition that carves an absorption line can produce a bright emission line when the geometry is reversed.

Optical depth: the numbers that matter

Optical depth is the single most important number in the whole subject because the observable behaviour flips completely across τ = 1. The transmitted fraction of a background beam is e^−τ:

Optical depth τ	Transmitted fraction e−τ	Regime	What you observe
0.01	99.0%	Very optically thin	See straight through; faint self-emission
0.1	90.5%	Optically thin	Thin nebula; intensity ∝ column density
1	36.8%	Unit optical depth	≈ 63% of a background beam absorbed
2/3 ≈ 0.67	51.3%	Photosphere	Layer whose Sν sets the emergent flux
5	0.67%	Optically thick	Only the outer skin visible
> 10⁶	≈ 0	Extremely thick (stellar interior)	Radiation diffuses; near-perfect blackbody

Inside the Sun the optical depth from centre to surface is of order 10¹¹, so a photon random-walks for tens of thousands of years to escape, even though a straight-line crossing at c would take about 2.3 seconds. In the deep, thick interior the equation of transfer reduces to the diffusion approximation, in which the radiative flux F_ν = −(4π/3α_ν) dB_ν/dz flows down the temperature gradient — this is what carries most of the Sun's luminosity through its radiative zone.

Worked example: the solar photosphere

Consider a ray leaving the centre of the solar disk (μ = 1). By Eddington–Barbier, the emergent continuum intensity equals B_ν evaluated where τ_ν = 1 along that ray; the disk-integrated flux corresponds to τ ≈ 2/3. The dominant continuum opacity in the visible is the negative hydrogen ion, H⁻, formed when a neutral H atom loosely binds an extra electron — a source Rupert Wildt identified in 1939. At τ_500nm = 2/3 the local temperature is ≈ 5772 K (the Sun's effective temperature, set by L_⊙ = 4πR_⊙²σT_eff⁴). Move to the core of a strong line — say the Ca II K line at 393.4 nm — and the bound-bound opacity is orders of magnitude higher, so τ = 2/3 is reached far higher, in gas near 4400 K. The intensity ratio between line core and continuum is roughly B_ν(4400 K)/B_ν(5772 K), a deep, dark line. This is the quantitative origin of the Fraunhofer spectrum (Joseph von Fraunhofer catalogued 574 solar lines in 1814–1815).

Limb darkening falls out of the same relation. Looking toward the edge of the disk you view along a slanted ray (small μ), which reaches τ = 2/3 higher up in cooler gas, so the limb looks dimmer and redder than disk centre. The measured limb-darkening curve of the Sun is a direct probe of the temperature-versus-optical-depth structure of its atmosphere.

Common misconceptions

• Optical depth is a distance. No — it is dimensionless, counting mean free paths. One centimetre of dense fog and a light-year of thin interstellar gas can share the same τ.
• We see the "surface" of a star. Stars have no solid surface; the photosphere is simply the fuzzy layer where τ falls to ≈ 2/3, a few hundred kilometres thick out of a 700,000 km radius.
• The source function is a property of the light. It is a property of the matter — the ratio j_ν/α_ν set by the gas's temperature, density, and atomic state.
• Absorption lines mean the element is missing. The opposite — a strong line means that species is abundant enough to raise τ and lift the visible layer into cooler gas.
• Emission and absorption lines require different physics. Same transition, same Kirchhoff ratio; only the temperature gradient along the sightline decides whether you see a dip or a spike.
• LTE always holds. It fails in thin, low-density, or strongly-irradiated regions where radiative rates outrun collisions, forcing a full non-LTE treatment.