Specific Intensity and Radiative Flux

Every measurement of light in astrophysics is, at bottom, a bookkeeping question: how much energy crosses a small surface, travelling in a particular direction, within a narrow band of frequencies, during a short interval of time? Package that answer into a single quantity and you have the specific intensity $I_\nu(\mathbf{r},\hat{\mathbf{n}},t)$ — the most fundamental variable of radiative transfer. It is what the equation of transfer is written in, and every other radiative quantity an astronomer quotes — mean intensity, flux, energy density, radiation pressure — is one of its angular moments, an integral $\int I_\nu\cos^n\theta\,\mathrm{d}\Omega$ with a different power of the projection cosine. Two facts make the intensity the right variable and separate it from the flux a telescope actually collects. First, it is directionally resolved: unlike the flux, a net quantity summed over all incoming directions, $I_\nu$ gives the brightness along one line of sight. Second, and more surprising, it is distance-invariant — in vacuum, along a ray with no emission or absorption, $I_\nu$ does not change. As a beam travels outward the source subtends a smaller solid angle in exactly the proportion needed to keep the intensity fixed. This is why the surface brightness of a galaxy, a nebula, or the solar disc is the same near or far, even though the total flux you receive falls as $1/d^2$ . That single distinction — intensity constant, flux diluting — sits under a huge amount of observational astrophysics, from galaxy photometry to the Tolman dimming test. And the moments of $I_\nu$ carry the story inward: deep in a star they close into the photon-gas law $P=aT^4/3$ , while at the surface they deliver the emergent flux $\sigma T_{\rm eff}^4$ that fixes a star's luminosity and colour.

Key takeaways

The specific intensity $I_\nu$ is the fundamental quantity of radiative transfer: energy per unit area, per unit solid angle, per unit frequency, per unit time, crossing a surface in a given direction.
In free space, along a ray with no emission or absorption, $I_\nu$ is conserved — this is why surface brightness does not change with distance, while flux falls as $1/d^2$ .
Every radiative observable is an angular moment of $I_\nu$ : mean intensity $J_\nu$ (no cosine), flux $F_\nu$ (one $\cos\theta$ ), radiation pressure $P_\nu$ (two), forming the Eddington hierarchy $J_\nu,H_\nu,K_\nu$ .
A uniformly bright surface radiates $F_\nu=\pi I_\nu$ into the outward hemisphere — the $\cos\theta$ weighting turns $2\pi$ steradians into a factor $\pi$ , and this is the $\pi$ in $F=\sigma T_{\rm eff}^4$ .
For an isotropic field $F_\nu=0$ , $u_\nu=(4\pi/c)J_\nu$ , and $P_\nu=u_\nu/3$ — the one-third closure that governs photon gases, stellar interiors, and the radiation-dominated early universe.
The Sun makes it concrete: surface brightness $I=2.0\times10^{10}\ \mathrm{erg\,s^{-1}cm^{-2}sr^{-1}}$ is identical at the photosphere and at Earth; only the flux dilutes, from $6.3\times10^{10}$ to the solar constant $1.36\times10^{6}\ \mathrm{erg\,s^{-1}cm^{-2}}$ .
Key idea: $I_\nu$ is the directionally-resolved, distance-invariant carrier of radiant energy; every observable radiative quantity is an angular moment $\int I_\nu\cos^n\theta\,\mathrm{d}\Omega$ of it.

The core insight

$I_\nu$ is the directionally-resolved, distance-invariant carrier of radiant energy; every observable radiative quantity — mean intensity, flux, energy density, radiation pressure — is an angular moment $\int I_\nu\cos^n\theta\,\mathrm{d}\Omega$ of it, distinguished only by the power of the projection cosine.

Where this lives today

The intensity–flux distinction is the daily grammar of observational astrophysics. Galaxy surveys quote surface brightness in magnitudes per square arcsecond precisely because it is distance-independent — letting astronomers compare a nearby spiral's disc with one at high redshift on equal footing, once the $(1+z)^{-4}$ Tolman dimming is folded in, a live test of the expanding universe. Low-surface-brightness galaxies and the diffuse intracluster light, both fainter than the night sky, are detectable only because their intensity does not fade with distance even as their flux sinks below the background. In stellar physics the angular dependence of the emergent intensity $I_\nu(0,\mu)$ is measured directly as limb darkening — the Sun's edge is dimmer because a slanted line of sight reaches only cooler upper layers — and the same signal in transiting-exoplanet and eclipsing-binary light curves reads the vertical temperature structure of stars we cannot resolve. Radiation pressure through the $P=u/3$ closure sets the Eddington luminosity that caps accreting neutron stars and quasars and regulates the most massive stars. And modern radiative-transfer codes — stellar-atmosphere synthesis, cosmological reionisation simulations, the general-relativistic ray-tracing behind the Event Horizon Telescope's black-hole images — all integrate the specific intensity along rays, because it remains, seventy years after Chandrasekhar, the one quantity that propagates cleanly.

The physical picture

Imagine standing in a slow drift of luminous sparks. The flux is the net rain of sparks through a hoop you hold up: tilt the hoop and the count changes, walk away and it falls off. The specific intensity is more robust — it is the brightness you perceive looking in one fixed direction, and it does not care how far away you stand, only what lies along that line of sight. The operational definition follows a single ray onto a detector: place a patch of area $\mathrm{d}A$ with normal $\hat{\mathbf{n}}$ , accept radiation arriving within solid angle $\mathrm{d}\Omega$ about a direction at angle $\theta$ to the normal, and the energy collected in time $\mathrm{d}t$ and band $\mathrm{d}\nu$ defines $I_\nu$ . The $\cos\theta$ is pure geometry: a ray striking obliquely spreads its energy over a larger footprint, so the true area is foreshortened to $\mathrm{d}A\cos\theta$ as seen along the beam.

Figure

A pencil of radiation strikes the surface patch dA at angle θ to the normal n̂, filling solid angle dΩ. The energy crossing in time dt and band dν is dE_ν = I_ν cosθ dA dΩ dν dt, where the cosθ is the foreshortening of dA seen along the ray

Once you have $I_\nu$ , the whole zoo of radiative quantities is generated by integrating over direction with different weights. Count photons regardless of heading and you get energy density and mean intensity — no cosine. Ask for the net energy crossing a surface and each ray is foreshortened by $\cos\theta$ — that is the flux, and for an isotropic field the inward and outward contributions cancel it to zero. Ask for the momentum flux and you get two powers of cosine — one from projecting momentum flux onto the surface, one from the direction of the momentum itself — which is why an isotropic photon gas obeys $P=u/3$ , the average of $\cos^2\theta$ over a sphere being exactly $1/3$ . The second figure turns the abstract invariance of $I_\nu$ into hard numbers with the nearest star: the Sun's surface brightness is the same at its photosphere and here at Earth, while the flux we receive has diluted by more than four orders of magnitude.

Figure

Along the ray from the Sun to Earth the specific intensity is invariant: the frequency-integrated surface brightness I = 2.0×10¹⁰ erg s⁻¹ cm⁻² sr⁻¹ is the same at the photosphere and at Earth. The bolometric flux dilutes as 1/d², from F = πI = 6.3×10¹⁰ at the surface to the solar constant 1.36×10⁶ erg s⁻¹ cm⁻², because the Sun subtends a solid angle Ω = 6.8×10⁻⁵ sr that shrinks as 1/d²

The core equations

The starting point is operational — a definition, not a derivation. Everything downstream is an angular integral of it. The energy $\mathrm{d}E_\nu$ registered by a detector fixes $I_\nu$ through the projected-area, solid-angle, bandwidth, and time factors.

\mathrm{d}E_\nu = I_\nu\,\cos\theta\,\mathrm{d}A\,\mathrm{d}\Omega\,\mathrm{d}\nu\,\mathrm{d}t \qquad (1)

where: $I_\nu$ — specific (monochromatic) intensity, units $\mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}\,sr^{-1}}$ ; $\mathrm{d}E_\nu$ — energy crossing the surface in the given band and time; $\theta$ — angle between ray direction and surface normal $\hat{\mathbf{n}}$ ; $\mathrm{d}A$ — element of surface area; $\mathrm{d}\Omega$ — element of solid angle about the ray; $\mathrm{d}\nu,\ \mathrm{d}t$ — frequency bandwidth and time interval. Reading it: $I_\nu$ is energy per unit everything, and the $\cos\theta$ says a beam hitting edge-on delivers less across the surface than one hitting face-on. The zeroth moment averages $I_\nu$ over all $4\pi$ steradians with no directional weighting. It is the natural driver of atomic rates, since an atom does not care which way a photon came from.

J_\nu = \frac{1}{4\pi}\oint I_\nu\,\mathrm{d}\Omega \qquad (2)

where: $J_\nu$ — mean intensity, the angular average of $I_\nu$ over the full sphere, same units as $I_\nu$ . Reading it: for an isotropic field $J_\nu = I_\nu$ exactly; in general $J_\nu$ is what appears in the emission–absorption balance of the source function. The first moment inserts one power of $\cos\theta$ to pick out the net transport across the surface. This is the flux — the quantity a telescope of known area actually collects — and its Eddington-normalised cousin $H_\nu$ .

F_\nu = \oint I_\nu\,\cos\theta\,\mathrm{d}\Omega,\qquad H_\nu = \frac{F_\nu}{4\pi} = \frac{1}{4\pi}\oint I_\nu\,\cos\theta\,\mathrm{d}\Omega \qquad (3)

where: $F_\nu$ — net monochromatic flux, $\mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}}$ ; $H_\nu$ — first Eddington moment, the flux normalised like $J_\nu$ and $K_\nu$ . Reading it: the cosine can be positive or negative, so opposing beams subtract; an isotropic field gives $F_\nu=0$ . Counting the energy resident in a unit volume — regardless of direction — gives the radiation energy density, tied directly to the mean intensity by the speed of light.

u_\nu = \frac{1}{c}\oint I_\nu\,\mathrm{d}\Omega = \frac{4\pi}{c}\,J_\nu \qquad (4)

where: $u_\nu$ — monochromatic energy density, $\mathrm{erg\,cm^{-3}\,Hz^{-1}}$ ; $c$ — speed of light, converting an energy flow into an energy content. Reading it: dividing a flow by $c$ turns "energy crossing per second" into "energy present per cm³". The second moment carries two powers of cosine and measures the flux of momentum — the radiation pressure — with $K_\nu$ its Eddington form.

P_\nu = \frac{1}{c}\oint I_\nu\,\cos^2\theta\,\mathrm{d}\Omega = \frac{4\pi}{c}\,K_\nu,\qquad K_\nu = \frac{1}{4\pi}\oint I_\nu\,\cos^2\theta\,\mathrm{d}\Omega \qquad (5)

where: $P_\nu$ — monochromatic radiation pressure (momentum flux), $\mathrm{dyne\,cm^{-2}\,Hz^{-1}}$ ; $K_\nu$ — second Eddington moment. Reading it: photons carry momentum $E/c$ , so the pressure is an energy-density-like quantity; the two cosines are the reason the isotropic result is $u/3$ , not $u$ . The deepest structural fact is that $I_\nu$ is not merely conserved along a ray in vacuum — the combination $I_\nu/\nu^3$ is a relativistic invariant, tracking the photon phase-space density that Liouville's theorem holds constant.

\frac{I_\nu}{\nu^3} = \text{invariant along a ray} \qquad (6)

where: $\nu^3$ — the phase-space Jacobian relating $I_\nu$ to the photon occupation number. Reading it: in static vacuum $\nu$ is fixed so $I_\nu$ itself is constant (surface brightness independent of distance); under a redshift $1+z$ both $I_\nu$ and $\nu$ transform, and surface brightness dims as $(1+z)^{-4}$ . Finally, at a stellar surface the frequency-integrated first moment delivers the emergent flux, defining the effective temperature; deep inside, the isotropic closure delivers the photon-gas pressure. These two lines connect the microphysics of $I_\nu$ to the macroscopic Stefan–Boltzmann law.

F = \pi\int_0^\infty I_\nu\,\mathrm{d}\nu = \sigma T_{\rm eff}^4,\qquad P_{\rm rad}=\frac{1}{3}u = \frac{1}{3}aT^4,\quad a=\frac{4\sigma}{c} \qquad (7)

where: $\sigma = 5.670\times10^{-5}\ \mathrm{erg\,cm^{-2}\,s^{-1}\,K^{-4}}$ — Stefan–Boltzmann constant; $a = 7.566\times10^{-15}\ \mathrm{erg\,cm^{-3}\,K^{-4}}$ — radiation constant; $T_{\rm eff}$ — the temperature of a blackbody with the same total emergent flux. Reading it: the factor $\pi$ is the hemispheric $\cos\theta$ integral of a uniform surface, derived below.

Derivation AThe angular moments and the one-third closure

The goal is to show, from the moment definitions alone, that an isotropic radiation field obeys the closure relations $J_\nu = I_\nu$ , $F_\nu = 0$ , $u_\nu = (4\pi/c)I_\nu$ , and above all $P_\nu = u_\nu/3$ — the single most important closure in radiative astrophysics, the one that makes a photon gas behave like a fluid with pressure one-third of its energy density and that underlies both stellar structure and the thermodynamics of the early universe. We take each moment in turn, exploit isotropy to pull the constant $I_\nu$ outside, and evaluate the elementary angular integrals so the factors of $4\pi$ and $1/3$ appear explicitly.

Step 1. For an isotropic field $I_\nu$ is the same in every direction, so it comes outside all angular integrals; the zeroth moment then gives $J_\nu = (1/4\pi)I_\nu\oint\mathrm{d}\Omega = (1/4\pi)I_\nu(4\pi) = I_\nu$ . Why: an unweighted average of a constant returns the constant, so the mean intensity simply equals the intensity — the baseline against which anisotropy is measured.

Step 2. The first moment carries $\cos\theta$ : $F_\nu = I_\nu\int_0^{2\pi}\!\!\int_0^\pi\cos\theta\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\phi$ , and the $\theta$ integral $\int_0^\pi\cos\theta\sin\theta\,\mathrm{d}\theta = [\tfrac12\sin^2\theta]_0^\pi = 0$ kills it. Why: over the full sphere every outgoing ray is matched by an incoming one of equal strength, so an isotropic field transports no net energy — there is brightness everywhere but no flux.

Step 3. The energy density needs no cosine: $u_\nu = (1/c)I_\nu\oint\mathrm{d}\Omega = (4\pi/c)I_\nu$ , which is also $(4\pi/c)J_\nu$ by Step 1. Why: energy density counts all photons present irrespective of heading, so the full solid angle $4\pi$ enters and the speed of light converts the flow into a stored density.

Step 4. The pressure carries $\cos^2\theta$ : $P_\nu = (I_\nu/c)\int_0^{2\pi}\!\!\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\phi$ , and the $\theta$ integral is $[-\tfrac13\cos^3\theta]_0^\pi = 2/3$ , so $P_\nu = (I_\nu/c)(2\pi)(2/3) = (4\pi/3)(I_\nu/c) = u_\nu/3$ . Why: the angular average of $\cos^2\theta$ over a sphere is exactly $1/3$ , and this single number — not any deep dynamics — is the origin of $P=u/3$ for every isotropic photon gas. Integrating over frequency for a blackbody, $u = aT^4$ gives $P_{\rm rad} = aT^4/3$ . The three key integrals are: $\oint\mathrm{d}\Omega = 4\pi$ ; $\int_0^\pi\cos\theta\sin\theta\,\mathrm{d}\theta = 0$ ; $\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta = 2/3$ ; together they give $P_\nu = u_\nu/3$ .

Go deeper · Derivation B · 1 minFlux from a uniformly bright surface and the factor pi

Derivation BFlux from a uniformly bright surface and the factor pi

The goal is to establish that a surface radiating uniform intensity $I_\nu$ outward into a hemisphere emits a flux $F_\nu = \pi I_\nu$ — not $2\pi I_\nu$ , as one might guess from the $2\pi$ steradians of a hemisphere — because the $\cos\theta$ projection weights face-on rays more than grazing ones. This is exactly the factor $\pi$ that sits in $F=\sigma T_{\rm eff}^4$ , so pinning it down connects the abstract moment $F_\nu$ to the Stefan–Boltzmann law that fixes a star's luminosity. We restrict the flux integral to outgoing directions, factor out the intensity, and evaluate the hemispheric integral.

Step 1. At a photosphere radiation escapes only outward, so the flux integral runs over the upper hemisphere: $F_\nu = \int_0^{2\pi}\!\!\int_0^{\pi/2} I_\nu\cos\theta\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\phi$ . Why: there is (to good approximation) no radiation coming back down onto the surface from outside, so only $0\le\theta\le\pi/2$ contributes and the flux does not cancel as it did in the isotropic full-sphere case.

Step 2. With $I_\nu$ uniform across the outward directions it leaves the integral, giving $F_\nu = 2\pi I_\nu\int_0^{\pi/2}\cos\theta\sin\theta\,\mathrm{d}\theta$ . Why: treating the emergent intensity as direction-independent is the defining idealisation of a uniformly bright ("Lambertian") surface, and it isolates the pure geometry in the remaining integral.

Step 3. Evaluate: $\int_0^{\pi/2}\cos\theta\sin\theta\,\mathrm{d}\theta = [\tfrac12\sin^2\theta]_0^{\pi/2} = \tfrac12$ , so $F_\nu = 2\pi I_\nu\cdot\tfrac12 = \pi I_\nu$ . Why: the cosine down-weights the grazing rays that make up most of the solid angle, halving the naive $2\pi$ to $\pi$ — the geometric heart of the result.

Step 4. Integrate over frequency with $I_\nu = B_\nu(T)$ : $F = \pi\int_0^\infty B_\nu(T)\,\mathrm{d}\nu = \pi B(T) = \sigma T^4$ , since $\int_0^\infty B_\nu\,\mathrm{d}\nu = (\sigma/\pi)T^4$ . Why: the blackbody surface radiates the Planck intensity, and the hemispheric $\pi$ converts intensity into the emergent flux we call $\sigma T_{\rm eff}^4$ . *The result rests on $\int_0^{\pi/2}\cos\theta\sin\theta\,\mathrm{d}\theta = 1/2$ , which gives $F_\nu = \pi I_\nu$ and links directly to $\sigma T^4$ .*

Go deeper · Derivation C · 2 minWhy intensity is invariant along a ray

Derivation CWhy intensity is invariant along a ray

The goal is to prove the claim that makes surface brightness distance-independent: in vacuum, with no emission or absorption, the specific intensity is the same at both ends of a ray, $I_{\nu,1} = I_{\nu,2}$ . The argument is not thermodynamic but purely geometric — it rests on the conservation of étendue, the product of area and solid angle, which is the phase-space volume of a ray bundle. It explains at a stroke why a nebula is no dimmer per unit solid angle when farther away, and why the intensity, not the flux, survives propagation unchanged.

Step 1. Consider two surface elements $\mathrm{d}A_1$ and $\mathrm{d}A_2$ facing each other across a distance $r$ along a common ray bundle, and follow the energy $\mathrm{d}E_\nu$ that leaves the first and arrives at the second with no losses in between. Why: energy conservation in vacuum equates what departs area 1 into the cone toward area 2 with what lands on area 2 from the cone back toward area 1 — nothing is added or removed along the way.

Step 2. The solid angle subtended by $\mathrm{d}A_2$ at $\mathrm{d}A_1$ is $\mathrm{d}\Omega_1 = \mathrm{d}A_2/r^2$ , and vice versa, so the throughput obeys $\mathrm{d}A_1\,\mathrm{d}\Omega_1 = \mathrm{d}A_1\,\mathrm{d}A_2/r^2 = \mathrm{d}A_2\,\mathrm{d}\Omega_2$ . Why: this symmetric product $\mathrm{d}A\,\mathrm{d}\Omega$ is the étendue, and its equality at the two ends is a geometric identity — the beam's phase-space volume is conserved as it propagates.

Step 3. Writing the transported energy at each end with the definition (1), $\mathrm{d}E_\nu = I_{\nu,1}\,\mathrm{d}A_1\,\mathrm{d}\Omega_1\,\mathrm{d}\nu\,\mathrm{d}t = I_{\nu,2}\,\mathrm{d}A_2\,\mathrm{d}\Omega_2\,\mathrm{d}\nu\,\mathrm{d}t$ ; since the étendue factors are equal, $I_{\nu,1} = I_{\nu,2}$ . Why: with the geometric factors cancelling and the energy fixed, the only way to balance the books is for the intensities themselves to be equal — surface brightness is conserved along the ray.

Step 4. Restore the two effects excluded above: emission and absorption add the source and sink terms of the transfer equation ( $\mathrm{d}I_\nu/\mathrm{d}s = -\alpha_\nu I_\nu + j_\nu$ ), while a cosmological redshift multiplies both $I_\nu$ and $\nu^3$ by $(1+z)^{-3}$ , leaving $I_\nu/\nu^3$ invariant. Why: the bare geometric result is the vacuum, static limit; the full transfer equation and the $(1+z)^{-4}$ surface-brightness dimming are its two important generalisations. *In summary: conservation of the étendue $\mathrm{d}A\,\mathrm{d}\Omega$ gives $I_{\nu,1}=I_{\nu,2}$ in vacuum, and the broader invariant $I_\nu/\nu^3$ follows from photon phase-space conservation.*

Limiting cases

The moments of $I_\nu$ collapse to memorable relations in the idealised radiation fields that recur throughout astrophysics.

Case	Result	Meaning
Isotropic field	$J_\nu = I_\nu,\ \ F_\nu = 0,\ \ u_\nu = \dfrac{4\pi}{c}I_\nu,\ \ P_\nu = \dfrac{u_\nu}{3}$	No net transport; pressure is one-third the energy density. Holds deep inside a star and in any photon gas.
Uniform outward hemisphere	$F_\nu = \pi I_\nu\ \ (\text{outward only})$	A Lambertian surface radiates $\pi I_\nu$ , not $2\pi I_\nu$ : the $\cos\theta$ weighting halves the naive hemisphere. This is the $\pi$ in $\sigma T_{\rm eff}^4$ .
Distant point source	$F_\nu = \dfrac{L_\nu}{4\pi d^2},\ \ I_\nu\ \text{const along ray}$	Flux dilutes as $1/d^2$ while intensity is unchanged; the source merely subtends a smaller solid angle.
Thermal equilibrium	$I_\nu = B_\nu(T) = \dfrac{2h\nu^3}{c^2}\dfrac{1}{e^{h\nu/k_BT}-1}$	Intensity is isotropic and equals the Planck function everywhere, independent of position or direction.
Cosmological ray (redshift $z$ )	$\dfrac{I_\nu}{\nu^3}\ \text{invariant}\ \Rightarrow\ \Sigma\propto(1+z)^{-4}$	Bolometric surface brightness $\Sigma$ dims as $(1+z)^{-4}$ — the Tolman test of the expanding universe.

Common trapConfusing intensity with flux — the single most common error

Intensity and flux are not the same thing scaled differently; they respond to distance in opposite ways. The specific intensity (surface brightness) is conserved along a ray in vacuum — a galaxy's central brightness in $\mathrm{erg\,s^{-1}cm^{-2}Hz^{-1}sr^{-1}}$ is the same near or far, which is why you cannot make a resolved nebula look brighter per unit area by moving closer, only bigger. The flux you collect falls as $1/d^2$ , because the source subtends a shrinking solid angle. The two are linked by $F_\nu = \int I_\nu\cos\theta\,\mathrm{d}\Omega$ , and the trap has a twin: the flux integral requires the $\cos\theta$ projection, while the energy density and mean intensity do not. Dropping the cosine from the flux, or sneaking it into the energy density, is the classic way to get the wrong power of $\pi$ or the wrong factor of $3$ . Remember the rule of moments: no cosine counts photons in a volume, one cosine counts energy across a surface, two cosines count momentum across a surface.

Worked examples

1 · The Sun proves that intensity does not fade (fully worked)

The Sun has effective temperature $T_{\rm eff}=5772\ \mathrm{K}$ , radius $R_\odot=6.96\times10^{10}\ \mathrm{cm}$ , and lies at $d=1.496\times10^{13}\ \mathrm{cm}$ . Show that its surface brightness (frequency-integrated intensity) is identical at the photosphere and at Earth, and that the flux we receive is the solar constant.

Step 1. The emergent flux at the photosphere is Stefan–Boltzmann: $F_{\rm surf} = \sigma T_{\rm eff}^4 = (5.670\times10^{-5})(5772)^4 = \mathbf{6.29\times10^{10}\ erg\,s^{-1}cm^{-2}}$ . Since a uniformly bright surface obeys $F=\pi I$ , the surface brightness is $I = F_{\rm surf}/\pi = \mathbf{2.00\times10^{10}\ erg\,s^{-1}cm^{-2}sr^{-1}}$ .

Step 2. The luminosity is the flux times the surface area: $L = 4\pi R_\odot^2 F_{\rm surf} = 4\pi(6.96\times10^{10})^2(6.29\times10^{10}) = \mathbf{3.83\times10^{33}\ erg\,s^{-1}}$ — the standard solar luminosity $L_\odot$ .

Step 3. The flux arriving at Earth is that luminosity spread over a sphere of radius $d$ : $F_\oplus = L/(4\pi d^2) = (3.83\times10^{33})/[4\pi(1.496\times10^{13})^2] = \mathbf{1.36\times10^{6}\ erg\,s^{-1}cm^{-2}} = 1361\ \mathrm{W\,m^{-2}}$ — the measured solar constant.

Step 4. Now recover the intensity from Earth. The Sun subtends solid angle $\Omega_\odot = \pi(R_\odot/d)^2 = \pi(4.65\times10^{-3})^2 = \mathbf{6.80\times10^{-5}\ sr}$ (an angular diameter of $0.53^\circ$ ). For a resolved disc, $I = F_\oplus/\Omega_\odot = (1.36\times10^{6})/(6.80\times10^{-5}) = \mathbf{2.00\times10^{10}\ erg\,s^{-1}cm^{-2}sr^{-1}}$ — identical to the photospheric value of Step 1. The flux fell by a factor $(R_\odot/d)^2 = 2.2\times10^{-5}$ , but the intensity did not move: every bit of the dimming lives in the shrinking solid angle, exactly as the invariance of $I_\nu$ demands.

2 · Radiation pressure versus gas pressure in the solar core

At the centre of the Sun $T\approx1.5\times10^{7}\ \mathrm{K}$ and the gas pressure is roughly $P_{\rm gas}\approx2.4\times10^{17}\ \mathrm{dyne\,cm^{-2}}$ . How much of the support against gravity comes from radiation?

Step 1. Deep in the star the field is very nearly isotropic, so the one-third closure applies: $P_{\rm rad} = \tfrac13 aT^4$ with $a = 7.566\times10^{-15}\ \mathrm{erg\,cm^{-3}K^{-4}}$ .

Step 2. Evaluate: $aT^4 = (7.566\times10^{-15})(1.5\times10^7)^4 = (7.566\times10^{-15})(5.06\times10^{28}) = 3.83\times10^{14}\ \mathrm{erg\,cm^{-3}}$ , so $P_{\rm rad} = \tfrac13(3.83\times10^{14}) = \mathbf{1.28\times10^{14}\ dyne\,cm^{-2}}$ .

Step 3. The ratio is $P_{\rm rad}/P_{\rm gas} = (1.28\times10^{14})/(2.4\times10^{17}) \approx \mathbf{5\times10^{-4}}$ — radiation contributes only about half a part in a thousand of the Sun's central support. In the Sun gas pressure dominates, but $P_{\rm rad}\propto T^4$ while $P_{\rm gas}\propto \rho T$ , so in a star ten times more massive the radiation term climbs to rival gas pressure and sets the Eddington limit $L_{\rm Edd}=4\pi GMc/\kappa\approx1.3\times10^{38}\ \mathrm{erg\,s^{-1}}$ for a solar mass — the luminosity at which radiation would blow the star apart.

3 · Energy density of the cosmic microwave background

The CMB is an isotropic blackbody field at $T=2.725\ \mathrm{K}$ . Find its energy density and pressure.

Step 1. Isotropy plus thermal equilibrium means $I_\nu = B_\nu(T)$ in every direction, so the energy density is simply $u = aT^4$ .

Step 2. Evaluate: $u = (7.566\times10^{-15})(2.725)^4 = (7.566\times10^{-15})(55.1) = \mathbf{4.2\times10^{-13}\ erg\,cm^{-3}}$ .

Step 3. Because the field is isotropic, the net flux vanishes, $F=0$ , in the comoving frame, and the pressure follows the closure $P = u/3 = \mathbf{1.4\times10^{-13}\ dyne\,cm^{-2}}$ . This same $P=u/3$ , with $u\propto T^4\propto(1+z)^4$ , is what made radiation dominate the dynamics of the universe before $z\approx3400$ .

Physics of the portrait

The specific intensity is one of those quantities everyone uses and few can name a discoverer for — it grew up over ninety years, from Kirchhoff's first clear statement of a radiation field, through Schwarzschild's and Eddington's stellar atmospheres, to the man who turned the scattered results into a single mathematical discipline.

Origins · 1 minSubrahmanyan Chandrasekhar and how this was discovered

Subrahmanyan Chandrasekhar

1910–1995 · radiative transfer and the theory of stellar atmospheres

The formalism of specific intensity and its moments was assembled over three generations, and Subrahmanyan Chandrasekhar is the figure who welded it into a coherent theory. Its root is Gustav Kirchhoff, who in 1859 stated the law relating a body's emission to its absorption and so isolated the idea of a temperature-controlled radiation field. Karl Schwarzschild, in 1906, wrote the equation of radiative equilibrium for a stellar atmosphere and introduced following intensity along a ray. Arthur Eddington, building the first real theory of stellar interiors through the 1910s and 1920s, defined the moment hierarchy $J,H,K$ and the closure $K=J/3$ that bears his name, and recognised radiation pressure as a pillar of stellar support. It was Chandrasekhar, in his 1950 monograph Radiative Transfer, who systematised the whole subject: he solved the transfer equation exactly for scattering atmospheres, derived the Sun's limb-darkening law, introduced the discrete-ordinate method still used in modern codes, and treated scattered-light polarisation with unprecedented rigour. His broader career — the Chandrasekhar mass limit for white dwarfs at age nineteen, then stellar dynamics, hydrodynamic stability, and the mathematical theory of black holes — won him the 1983 Nobel Prize. But it is Radiative Transfer that made the specific intensity the working language of stellar atmospheres, and every moment relation here traces to the tradition he completed.

History

1859

— Kirchhoff states his law relating emission to absorption, isolating the thermal radiation field

1879

— Stefan measures the fourth-power radiation law empirically

1884

— Boltzmann derives F = σT⁴ thermodynamically from Maxwell's radiation pressure

1906

— Schwarzschild writes the equation of radiative equilibrium for a stellar atmosphere

1926

— Eddington formalises the moment equations and the K = J/3 closure in stellar interiors

1950

— Chandrasekhar's Radiative Transfer systematises the intensity formalism and solves limb darkening

Going further

Texts. Rybicki and Lightman, Radiative Processes in Astrophysics, Ch. 1 (definitions and moments) · Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 3 and 9 · Mihalas, Stellar Atmospheres, 2nd ed., Ch. 1 · Chandrasekhar, Radiative Transfer, Ch. 1 · Ryden, Introduction to Cosmology, 2nd ed., Ch. 2 (surface brightness and Tolman dimming). The key numerical checks underlying this article are: (1) $\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta = 2/3$ , giving $P=u/3$ ; (2) $\int_0^{\pi/2}\cos\theta\sin\theta\,\mathrm{d}\theta = 1/2$ , giving $F=\pi I$ ; (3) constants $\sigma = 5.670\times10^{-5}\ \mathrm{erg\,cm^{-2}s^{-1}K^{-4}}$ and $a=4\sigma/c = 7.566\times10^{-15}\ \mathrm{erg\,cm^{-3}K^{-4}}$ ; (4) the solar chain $F_{\rm surf}=\sigma T_{\rm eff}^4 = 6.29\times10^{10}$ , $I=F/\pi = 2.00\times10^{10}$ , $L=3.83\times10^{33}$ , $F_\oplus=1.36\times10^{6}$ , $\Omega_\odot=6.80\times10^{-5}$ , and $I=F_\oplus/\Omega_\odot$ recovering the same $2.00\times10^{10}$ ; (5) CMB $u=aT^4=4.2\times10^{-13}\ \mathrm{erg\,cm^{-3}}$ at $T=2.725\ \mathrm{K}$ ; (6) étendue $\mathrm{d}A\,\mathrm{d}\Omega$ conserved. All constants are standard CGS values.

Prerequisites. Solid angle and spherical geometry · Electromagnetic energy transport. Leads to. The Equation of Radiative Transfer · Blackbody Radiation and the Planck Law · Emission and the Source Function · Stefan–Boltzmann and Wien Displacement Laws.

Ask the tutor: Want to see how these same moments $J_\nu, H_\nu, K_\nu$ turn the ray-by-ray transfer equation into the moment equations that build a whole stellar atmosphere?

Is the specific intensity the same thing as brightness?

Essentially, yes — "surface brightness" is the everyday name for the specific intensity of a resolved source. For an extended object like the Moon, a nebula, or a galaxy, the intensity is the energy per second reaching you from each patch of the source, per unit area of your pupil, per unit solid angle it covers, per unit frequency. Its defining feature is distance-independence in vacuum: the Moon's disc is exactly as bright per square arcsecond to the naked eye or through a telescope — the telescope makes it bigger, gathering more total flux, but no brighter per unit area. This is why a telescope cannot make an extended nebula outshine the sky background; it can only resolve more of it.

Why does the flux fall as 1/d² but the intensity does not?

Because the two quantities measure different things. Flux is the total energy per unit area per second you collect from the whole source, and as you retreat the source subtends a smaller solid angle — its angular size shrinks as $1/d$ , its solid angle as $1/d^2$ — so the collected energy drops as $1/d^2$ . Intensity is energy per unit solid angle of the source, so the shrinking angular size is divided back out: each individual patch of the source sends you the same brightness no matter how far you are. Formally, $F_\nu = \int I_\nu\cos\theta\,\mathrm{d}\Omega$ , and for a resolved source of solid angle $\Omega$ the flux is $I_\nu\Omega$ ; since $I_\nu$ is fixed and $\Omega\propto1/d^2$ , the flux carries all the distance dependence.

Where does the factor of 1/3 in radiation pressure come from?

It is a purely geometric average, not a dynamical effect. Radiation pressure is the second moment of the intensity, carrying two powers of $\cos\theta$ : one because only the component of a photon's momentum along the surface normal pushes on it, and one because the flux of that momentum across the surface is itself projected by $\cos\theta$ . For an isotropic field the intensity factors out and you are left with the average of $\cos^2\theta$ over the sphere, exactly $1/3$ (from $\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta = 2/3$ over $\int_0^\pi\sin\theta\,\mathrm{d}\theta = 2$ ). Hence $P = u/3$ for any isotropic photon gas — the CMB, a stellar interior, or blackbody radiation in a box.

What is the difference between the flux F and the Eddington flux H?

They are the same quantity in different normalisations. The astrophysical flux $F_\nu = \oint I_\nu\cos\theta\,\mathrm{d}\Omega$ is the net energy crossing unit area per second, the thing a detector measures. The Eddington flux $H_\nu = F_\nu/4\pi$ is that same flux divided by $4\pi$ so it sits on the same footing as the other Eddington moments $J_\nu$ and $K_\nu$ , all of which are $(1/4\pi)\oint I_\nu\cos^n\theta\,\mathrm{d}\Omega$ with $n=0,1,2$ . The Eddington normalisation is convenient because the transfer equation and closures like $K=J/3$ look cleanest when all three moments are averages over $4\pi$ . Which convention a book uses is a matter of taste; the physics is identical.

Why is I_ν/ν³ the invariant rather than I_ν itself?

Because $I_\nu/\nu^3$ is proportional to the photon phase-space density — the number of photons per unit volume of position–momentum space — which is a Lorentz invariant and, by Liouville's theorem, is conserved along a ray in the absence of interactions. In static vacuum the frequency of a photon does not change, so the $\nu^3$ factor is constant and $I_\nu$ itself is conserved: that is ordinary surface-brightness invariance. But in a gravitational field or an expanding universe the frequency shifts, $\nu\to\nu/(1+z)$ , and then $I_\nu$ alone is not conserved — it is the ratio $I_\nu/\nu^3$ that stays fixed. Integrating over frequency, the bolometric surface brightness then dims as $(1+z)^{-4}$ , the celebrated Tolman signature of an expanding cosmos.

Do these definitions still hold for absorbing or emitting media?

The moment definitions always hold — $J_\nu$ , $F_\nu$ , $u_\nu$ , and $P_\nu$ are angular integrals of whatever $I_\nu$ happens to be, in any medium. What changes is the invariance of $I_\nu$ along a ray. In vacuum $I_\nu$ is constant; inside a medium it obeys the equation of radiative transfer, $\mathrm{d}I_\nu/\mathrm{d}s = -\alpha_\nu I_\nu + j_\nu$ , where $\alpha_\nu$ removes energy by absorption and scattering and $j_\nu$ adds it by emission. The intensity then relaxes toward the source function $S_\nu = j_\nu/\alpha_\nu$ , and in thermal equilibrium $S_\nu = B_\nu(T)$ , so $I_\nu$ drives toward the Planck function. The moments and their closures — especially $P=u/3$ deep inside where the field is nearly isotropic — remain the tools that turn this ray-by-ray equation into the tractable moment equations of stellar structure.