Lecture: Radiative Equilibrium in a Gray Atmosphere

1. Introduction: From Shell Models to the Continuum

In previous lectures, we relied on discrete "shell" models to understand the greenhouse effect. While conceptually useful, planetary atmospheres are continuous fluids. To transition from a discrete shell model to a continuous vertical profile, we must integrate the radiative fluxes across the atmospheric column.

Our tool for this analytical step is the Gray Gas Approximation.

The "Gray" Assumption: We assume the absorption coefficient $\kappa_\lambda$ is independent of wavelength $\lambda$. While real gases have complex absorption bands, the gray gas model yields exact analytical solutions that expose the fundamental thermodynamic scaling of planetary climates.
Coordinate System: We define optical depth $\tau$ as measured upward from the surface.
- $\tau = 0$ is the surface boundary.
- $\tau = \tau_{\infty}$ is the Top of the Atmosphere (TOA).

By treating radiation as a one-dimensional, two-stream problem (upwelling $I^+$ and downwelling $I^-$), our governing differential equations become:

\frac{dI^+}{d\tau} = B(\tau) - I^+(\tau)

\frac{dI^-}{d\tau} = I^-(\tau) - B(\tau)

Where $B(\tau) = \sigma T(\tau)^4$ is the local Planck emission. Moving upward ($d\tau > 0$), the upwelling stream $I^+$ is depleted by absorption ($-I^+$) and augmented by local thermal emission ($+B$). The signs are reversed for the downwelling stream because it propagates in the $-d\tau$ direction.

2. Integral Solutions and the Heating Rate

To understand how an atmosphere approaches equilibrium, we must first define the local thermal forcing: the Radiative Heating Rate ($\mathcal{H}$). This is defined as the negative divergence of the net flux ($F_{net} = I^+ - I^-$).

Taking the derivative of $F_{net}$ with respect to $\tau$ and substituting our two-stream equations:

\mathcal{H}(\tau) = -\frac{dF_{net}}{d\tau} = -\left(\frac{dI^+}{d\tau} - \frac{dI^-}{d\tau}\right) = -(B - I^+ - I^- + B) = I^+(\tau) + I^-(\tau) - 2B(\tau)

This equation has profound physical meaning. A local layer of atmosphere cools by emitting radiation both upward and downward ($-2B$). It heats by absorbing the ambient radiation field ($I^+ + I^-$).

A. The Formal Integral Solutions

To evaluate $\mathcal{H}$, we solve the differential equations using integrating factors.

Upwelling Stream ($I^+$): $I^+(\tau) = I^+(0)e^{-\tau} + \int_0^\tau B(\tau') e^{-(\tau - \tau')} d\tau'$
Physical Meaning: The upwelling flux at level $\tau$ is the surface emission $I^+(0)$ attenuated by the optical depth $e^{-\tau}$, plus the cumulative emission of all atmospheric layers below $\tau'$, attenuated by the intervening distance $(\tau - \tau')$.
Downwelling Stream ($I^-$):
Assuming space emits no infrared radiation ($I^-(\tau_\infty) = 0$):
$I^-(\tau) = \int_\tau^{\tau_\infty} B(\tau') e^{-(\tau' - \tau)} d\tau'$
Physical Meaning: The downwelling flux is strictly the integrated emission of the atmospheric layers above $\tau$.

B. Two Limiting Regimes of Radiative Transfer (Detailed Derivations)

Using these equations, we can rigorously derive the behavior of an atmosphere in two extreme physical limits.

1. The Optically Thin Limit ($\tau_\infty \ll 1$):

In a highly tenuous atmosphere (e.g., modern Mars), the total optical depth is very small. We can apply a Taylor expansion to the exponential attenuation terms: $e^{-x} \approx 1 - x$.

Let's evaluate the integrals assuming the atmosphere is roughly isothermal at temperature $T$, meaning $B(\tau') \approx B = \sigma T^4$.

For the upwelling stream, the integral domain $[0, \tau]$ is extremely small, so the integral evaluates to approximately $B \cdot \tau$:

I^+(\tau) \approx \sigma T_g^4 (1-\tau) + B\tau

For the downwelling stream, the integral over $[\tau, \tau_\infty]$ evaluates to:

I^-(\tau) \approx B(\tau_\infty - \tau)

Now, substitute these approximations back into the heating rate equation $\mathcal{H} = I^+ + I^- - 2B$:

\mathcal{H}(\tau) \approx [\sigma T_g^4 (1-\tau) + B\tau] + [B(\tau_\infty - \tau)] - 2B

Grouping the $B$ terms together:

\mathcal{H}(\tau) \approx \sigma T_g^4 (1-\tau) + B(\tau + \tau_\infty - \tau - 2) = \sigma T_g^4 (1-\tau) + B(\tau_\infty - 2)

Because $\tau_\infty \ll 1$, the term $(\tau_\infty - 2) \approx -2$. Thus, the heating rate simplifies to:

\mathcal{H}(\tau) \approx \sigma T_g^4 (1-\tau) - 2\sigma T^4

Physical Result: In the absolute thin limit ($\tau \to 0$), equilibrium ($\mathcal{H}=0$) requires $2\sigma T^4 = \sigma T_g^4$, meaning the atmosphere's temperature is purely slaved to the ground: $T \approx 0.84 T_g$. It is a non-local regime dominated entirely by boundary conditions.

2. The Optically Thick / Diffusive Limit ($\tau \gg 1$ and $\tau_\infty - \tau \gg 1$):

In a dense atmosphere (e.g., Venus), deep layers are entirely shielded from the boundaries. Pierrehumbert derives this limit rigorously using the integral solutions. Because the exponential term $e^{-|\tau - \tau'|}$ decays extremely rapidly as photons move away from their source, a layer at $\tau$ only "sees" radiation from its immediate surroundings ($\tau' \approx \tau$).

We can therefore approximate the emission source $B(\tau')$ inside the integrals by performing a Taylor series expansion centered around $\tau$:

B(\tau') \approx B(\tau) + (\tau' - \tau)\frac{dB}{d\tau}

Substitute this into the integral for $I^+(\tau)$. Since deep in the atmosphere the surface contribution $e^{-\tau}$ is nearly zero, we drop it:

I^+(\tau) \approx \int_0^\tau \left[ B(\tau) + (\tau' - \tau)\frac{dB}{d\tau} \right] e^{-(\tau - \tau')} d\tau'

Let $x = \tau - \tau'$, which means $dx = -d\tau'$. The integration limits change from $[0, \tau]$ to $[\tau, 0]$, but absorbing the negative sign from $dx$ flips them back to $[0, \tau]$. Furthermore, because we are in an optically thick regime ($\tau \gg 1$), we can extend the upper bound of the integral to $\infty$ without losing accuracy (the exponential makes distant contributions vanish anyway):

I^+(\tau) \approx \int_0^\infty \left[ B(\tau) - x\frac{dB}{d\tau} \right] e^{-x} dx

Using standard exponential integrals ($\int_0^\infty e^{-x} dx = 1$ and $\int_0^\infty x e^{-x} dx = 1$), this evaluates precisely to:

I^+(\tau) \approx B(\tau) - \frac{dB}{d\tau}

Following a perfectly symmetric procedure for $I^-(\tau)$ using the substitution $x = \tau' - \tau$, we get:

I^-(\tau) \approx B(\tau) + \frac{dB}{d\tau}

The net flux $F_{net}$ in this deeply buried layer is therefore:

F_{net} = I^+ - I^- = \left(B - \frac{dB}{d\tau}\right) - \left(B + \frac{dB}{d\tau}\right) = -2\frac{dB}{d\tau}

Physical Result: In an optically thick medium, photons perform a random walk. Radiation becomes a purely local diffusive process, mathematically identical to Fourier's Law of heat conduction. The net flux is strictly proportional to the negative local gradient of the temperature profile (represented by the Planck function).

3. The Condition for Radiative Equilibrium

A climate is in pure Radiative Equilibrium when the heating rate is exactly zero at every vertical level. There is no net convergence or divergence of energy:

\mathcal{H}(\tau) = 0 \implies \frac{dF_{net}}{d\tau} = 0 \implies F_{net} = \text{constant}

By definition of planetary energy balance, the constant net outgoing infrared flux at the TOA must equal the absorbed solar radiation:

F_{net}(\tau_\infty) = \frac{S_0}{4}(1 - \alpha) = \sigma T_{e}^4

Where $T_e$ is the planet's effective emission temperature. Since $F_{net}$ is constant, $F_{net}(\tau) = \sigma T_e^4$ everywhere.

4. Analytical Derivation of the Temperature Profile

We now solve for the exact equilibrium temperature profile $T(\tau)$ using Pierrehumbert's "Sum and Difference" method.

Step 1: The Difference Equation

Subtract the two-stream equations:

\frac{d}{d\tau}(I^+ - I^-) = (B - I^+) - (I^- - B) = 2B - (I^+ + I^-)

Because we are in equilibrium, $\frac{d}{d\tau}(I^+ - I^-) = \frac{dF_{net}}{d\tau} = 0$. Therefore:

B(\tau) = \frac{1}{2}(I^+ + I^-)

Result: In radiative equilibrium, the local emission function perfectly averages the ambient upwelling and downwelling radiation streams.

Step 2: The Sum Equation

Add the two-stream equations:

\frac{d}{d\tau}(I^+ + I^-) = (B - I^+) + (I^- - B) = -(I^+ - I^-) = -F_{net}

Since $F_{net}$ is constant, we easily integrate this from $\tau=0$:

I^+ + I^- = -F_{net} \cdot \tau + C

Step 3: Applying TOA Boundary Conditions

To find the constant of integration $C$, we evaluate at the Top of the Atmosphere ($\tau = \tau_\infty$).

We know two things at TOA:

$I^-(\tau_\infty) = 0$ (No IR from space).
$F_{net} = I^+(\tau_\infty) - I^-(\tau_\infty) \implies I^+(\tau_\infty) = F_{net}$.

Therefore, the sum at the TOA is simply: $I^+(\tau_\infty) + I^-(\tau_\infty) = F_{net}$.

Plugging this into our integrated Sum Equation at $\tau = \tau_\infty$:

F_{net} = -F_{net}\tau_\infty + C \implies C = F_{net}(1 + \tau_\infty)

Step 4: The Final Profile

Substitute $C$ back into the Sum Equation for any level $\tau$:

I^+(\tau) + I^-(\tau) = F_{net}(1 + \tau_\infty - \tau)

Recall from Step 1 that $B(\tau) = \frac{1}{2}(I^+ + I^-)$. Therefore:

B(\tau) = \frac{F_{net}}{2}(1 + \tau_\infty - \tau)

Substituting $B(\tau) = \sigma T(\tau)^4$ and $F_{net} = \sigma T_e^4$:

T(\tau) = T_e \left[ \frac{1 + \tau_{\infty} - \tau}{2} \right]^{1/4}

This is the fundamental analytical solution for a gray radiative equilibrium atmosphere.

5. Physical Implications of the Equilibrium State

The analytical profile we just derived forces profound boundary anomalies and steep thermal gradients that are critical to understanding real planetary climates.

I. The Skin Temperature ($T_{skin}$)

What is the temperature of the air at the absolute edge of space? Evaluate the profile at $\tau = \tau_\infty$:

T(\tau_\infty) = T_e \left[ \frac{1 + \tau_{\infty} - \tau_{\infty}}{2} \right]^{1/4} = T_e \left( \frac{1}{2} \right)^{1/4} \approx 0.84 T_e

The Physics: The uppermost layer of the atmosphere radiates both up and down ($2\sigma T^4$), but only receives radiation from below ($\sigma T_e^4$). To balance energy, $2T_{skin}^4 = T_e^4$. This fundamental limit explains why pure radiative planetary stratospheres are highly isothermal; they cannot cool below this boundary condition.

II. The Surface Discontinuity ($T_g$ vs $T_{air}$)

Let's evaluate the air temperature immediately adjacent to the ground ($\tau = 0$):

T_{air}(0) = T_e \left[ \frac{1 + \tau_{\infty}}{2} \right]^{1/4}

Now, we must find the actual temperature of the solid ground, $T_g$. The ground emission is $I^+(0) = \sigma T_g^4$.

We can solve for $I^+(0)$ algebraically by combining the definitions of the Sum and Difference of the fluxes at the surface:

$I^+(0) - I^-(0) = F_{net}$
$I^+(0) + I^-(0) = F_{net}(1 + \tau_\infty)$

Adding these two equations yields $2I^+(0) = F_{net}(2 + \tau_\infty)$, or:

I^+(0) = F_{net} \left( 1 + \frac{\tau_\infty}{2} \right) \implies T_g = T_e \left( 1 + \frac{\tau_\infty}{2} \right)^{1/4}

The Physics: Compare $T_g$ to $T_{air}(0)$. Because $(1 + \tau_\infty/2)$ is strictly greater than $(0.5 + \tau_\infty/2)$, the ground is always strictly hotter than the air immediately above it in pure radiative equilibrium.

Why? The ground must heat up enough to radiate away both the incoming solar radiation and the downwelling atmospheric back-radiation ($I^-$). The air layer immediately above it only has to balance IR exchange.

III. The Radiative Lapse Rate and Convective Instability

To understand whether this atmosphere is dynamically stable, we must examine the vertical temperature gradient, or the lapse rate $\Gamma_{rad} = -dT/dz$.

From our equilibrium profile $T(\tau)$, we differentiate $T^4$ with respect to $\tau$:

4T^3 \frac{dT}{d\tau} = -\frac{T_e^4}{2} \implies \frac{dT}{d\tau} = -\frac{T_e^4}{8T^3}

To find the physical lapse rate $\Gamma_{rad}$, we use the chain rule: $dT/dz = (dT/d\tau) \cdot (d\tau/dz)$. Since $\tau$ is measured upward from the surface, $d\tau/dz = \kappa \rho$.

\Gamma_{rad} = -\frac{dT}{dz} = \left( \frac{T_e^4}{8T^3} \right) \kappa \rho

Eliminating Temperature: As derived, this equation is unsatisfying because $T$ is in the denominator. To reveal the true physical bounds, we must eliminate $T$ entirely.

First, substitute density $\rho$ using the Ideal Gas Law ($\rho = p / RT$):

\Gamma_{rad} = \frac{\kappa p T_e^4}{8 R T^4}

Now, substitute our exact analytical solution for $T^4 = T_e^4 \left[ \frac{1 + \tau_\infty - \tau}{2} \right]$ into the denominator:

\Gamma_{rad} = \frac{\kappa p T_e^4}{8 R \left( T_e^4 \left[ \frac{1 + \tau_\infty - \tau}{2} \right] \right)} = \frac{\kappa p}{4 R (1 + \tau_\infty - \tau)}

Let's define $\tau^* = \tau_\infty - \tau$ as the optical depth measured from space downward. We can relate $\tau^*$ to pressure $p$ via hydrostatic balance ($dp = -\rho g dz$). Since $d\tau^* = -\kappa \rho dz$, it follows that $d\tau^* = (\kappa/g) dp$. Assuming the absorber is well-mixed ($\kappa$ is constant), integrating from space downwards yields $\tau^* = \kappa p / g$, or $p = g \tau^* / \kappa$.

Substituting $p$ into our equation perfectly eliminates $\kappa$:

\Gamma_{rad} = \frac{\kappa \left(\frac{g \tau^*}{\kappa}\right)}{4 R (1 + \tau^*)} = \frac{g}{4 R} \left( \frac{\tau^*}{1 + \tau^*} \right)

Physical Interpretation:

This is a beautiful, temperature-independent result that reveals three profound insights:

The Deep Atmosphere Limit: Deep in an optically thick atmosphere ($\tau^* \to \infty$), the term $\frac{\tau^*}{1 + \tau^*} \to 1$. The radiative lapse rate hits a hard asymptotic ceiling: $\Gamma_{rad} \to \frac{g}{4R}$.
The Condition for Convection: An atmosphere becomes unstable and convects if $\Gamma_{rad}$ exceeds the dry adiabatic lapse rate $\Gamma_a = g/c_p$. Therefore, a purely gray atmosphere will only convect if $\frac{g}{4R} > \frac{g}{c_p}$, which simplifies to the condition: $c_p > 4R$.
The Earth Paradox: For a diatomic gas like Earth's atmosphere ($N_2/O_2$), $c_p \approx 3.5 R$. Because $3.5 R$ is less than $4 R$, a well-mixed, perfectly gray Earth atmosphere would actually never convect! This proves that real-world convection requires one of two things: either a polyatomic gas with a very high heat capacity (like a pure $CO_2$ atmosphere), or "non-gray" properties where opacity $\kappa$ increases strongly near the surface (like water vapor scaling with pressure/temperature).