Lecture: Radiative Transfer - From Shells to the Schwarzschild Continuum

I. The Departure Point: The Zero-D Radiating Temperature

In our previous lectures, our baseline for understanding any planet has been the conservation of energy. In the simplest case (a 1D or Zero-D model), we assume the planet behaves as a blackbody in equilibrium with incoming solar radiation.

The Energy Balance: $G = \frac{S}{4}(1-\alpha) - \sigma T_{rad}^4 = 0$
The Physical Definition: $T_{rad}$ is the "apparent" temperature of the planet as seen from space—the temperature required to balance the absorbed stellar flux.
The Surface Assumption: In the absence of an atmosphere (like on the Moon), $T_s = T_{rad}$. This simple balance governs the surface environment.

II. The First Jump: The Discrete Shell Model

To introduce the greenhouse effect, we move to a situation where $T_s$ is no longer equal to $T_{rad}$. We model the atmosphere as a single isothermal "shell" with a specific infrared emissivity ($\epsilon$).

The TOA Balance: $\frac{S}{4}(1-\alpha) = \epsilon \sigma T_a^4 + (1-\epsilon)\sigma T_s^4$
The Surface Balance: $\sigma T_s^4 = \frac{S}{4}(1-\alpha) + \epsilon \sigma T_a^4$
The Mathematical Result: By solving these, we find $T_s = T_{rad}(2-\epsilon)^{-1/4}$.

The "Lumped" or "Bulk" Assumption:
Notice what is missing in this model: distance. In the shell model, we treat the entire atmosphere as a single, homogenous block. We assign it a bulk property ($\epsilon$). It is a boundary value problem, not a path-dependent one. We only care about the energy crossing the Top of Atmosphere and the Surface, and we deliberately ignore the internal structure of the atmosphere.

III. Breaking the Bulk: Stacking the Shells

Why must we move beyond a single shell? Because a real atmosphere is not a homogenous block of gas. It has a lapse rate (temperature changes with altitude) and a density gradient (mass drops off exponentially).

To capture this reality, let's take our single atmospheric shell and replace it with a stack of $N$ shells, one on top of the other.

Let's zoom in and look at the energy balance across just one of these thin intermediate shells. Suppose an intensity of radiation $I$ enters the bottom of the shell. What happens to it as it crosses to the top?

The change in intensity ($\Delta I$) across this single shell is driven by two competing processes:

Loss (Absorption): The shell absorbs a fraction of the incoming light. By definition, this fraction is the absorptivity of the shell.
Gain (Emission): The shell itself is warm ($T$), so it emits its own radiation.

Assuming Local Thermodynamic Equilibrium (LTE), Kirchhoff's law tells us that the absorptivity of the shell must exactly equal its emissivity ($\epsilon$). If the blackbody emission for the shell's temperature is the Planck function $B(T)$, we can write the exact algebraic balance for the radiation leaving the shell:

\text{Radiation Out} = \text{Radiation In} - \text{Absorbed} + \text{Emitted}

I_{out} = I - \epsilon I + \epsilon B(T)

Rearranging this gives us the net change in radiation ($\Delta I = I_{out} - I$) across our single shell:

\Delta I = -\epsilon I + \epsilon B(T)

Look closely at this algebraic equation. The first term ($-\epsilon I$) states that the amount of light absorbed is proportional to the intensity of the light entering. This is the algebraic form of the Beer-Bouguer-Lambert Law! The second term ($\epsilon B(T)$) is the thermal source function. We have just derived the core of radiative transfer from the simple algebra of our shell model.

IV. The Microscopic Picture: Defining the Emissivity ($\epsilon$)

To turn our stack of shells into a continuous atmosphere, we need to ask: what actually determines $\epsilon$ for a very thin shell of thickness $\Delta s$?

Imagine a single photon traveling through our slice $\Delta s$ like a dart thrown randomly at a wall. The gas molecules are targets on that wall. The absorption cross-section ($k_\lambda$) is the effective area of the target's bullseye.

What is a "Quantum-Mechanical Capture Area"?
Instead of a physical wall, think of the molecule as a microscopic radio antenna:

Resonance: Quantum mechanics dictates that a molecule can only vibrate or rotate at very specific frequencies. If the passing photon's frequency perfectly matches one of these "resonant" frequencies (such as the energy required to bend a $CO_2$ bond), the molecule acts like an antenna tuned exactly to that station.
Reaching Out: Because of this electromagnetic resonance, the molecule interacts with the photon's electric field over a volume of space much larger than the physical atoms themselves.
The Cross-Section: This electromagnetically inflated target size is the absorption cross-section ($k_\lambda$). At specific resonant wavelengths, $k_\lambda$ is massive. At non-resonant wavelengths, the photon passes right through ($k_\lambda \approx 0$).

From Micro to Macro:
When we multiply $k_\lambda$ (target area per unit mass, $m^2/kg$) by the macroscopic gas density $\rho$ (mass of targets per unit volume, $kg/m^3$), we get $k_\lambda \rho$. This represents the total effective target area blocking the photon per meter of travel.

Therefore, for a very thin shell of thickness $\Delta s$, the fraction of light blocked (the emissivity/absorptivity $\epsilon$) is simply the target area per meter multiplied by the thickness of the shell:

\epsilon = k_\lambda \rho \Delta s

V. The Mathematical Framework: The Schwarzschild Equations

Now, we take the final leap. We substitute our microscopic definition of emissivity ($\epsilon = k_\lambda \rho \Delta s$) back into our shell model's algebraic equation:

\Delta I_\lambda = - (k_\lambda \rho \Delta s) I_\lambda + (k_\lambda \rho \Delta s) B_\lambda(T)

If we stack an infinite number of these shells, making each shell infinitesimally thin, the discrete thickness $\Delta s$ becomes the differential path length $ds$, and the discrete change $\Delta I$ becomes the differential $dI$:

dI_\lambda = - I_\lambda k_\lambda \rho ds + B_\lambda(T) k_\lambda \rho ds

Dividing the entire equation by $k_\lambda \rho ds$, we arrive exactly at the fundamental differential equation for radiative transfer along any arbitrary path—the Schwarzschild Equation:

\frac{dI_\lambda}{k_\lambda \rho ds} = -I_\lambda + B_\lambda(T)

By treating the atmosphere as an infinite stack of thin shells, we have seamlessly translated the algebraic energy balances of our simple climate models into the rigorous calculus required to simulate a real planetary atmosphere.

VI. The Natural Coordinate: Why Use Optical Depth ($\tau$)?

To clean up this equation, we define a new coordinate system. Instead of tracking the geometric height ($z$) in meters, we want to track how much "target area" the radiation has passed through.

Let's set our origin at the planetary surface. We define optical thickness ($\tau_\lambda$) such that $\tau_\lambda = 0$ at the surface, and it increases as we move upward into the atmosphere. For a small upward step $dz$, the change in optical depth is:

d\tau_\lambda = k_\lambda \rho dz

Why introduce $\tau$?

Radiation doesn't care about meters: A photon only cares about how many absorbing molecules it encounters. 1 km of travel at the surface absorbs vastly more radiation than 1 km in the stratosphere.
Mathematical Normalization: By bundling the messy, highly variable properties of the gas ($k_\lambda$ and $\rho$) into $d\tau_\lambda$, we strip the differential equations down to their pure, fundamental radiative physics. In $\tau$-space, the probability of a photon interacting with the atmosphere is uniform everywhere in the vertical column.

VII. Upwelling and Downwelling Radiation

Because a real atmosphere emits radiation in all directions, we must split our radiation field into an upwelling component ($I^+$) and a downwelling component ($I^-$).

Since we anchored our coordinate system at the surface ($\tau = 0$) and set it to increase upward to the Top of Atmosphere ($\tau = \tau_\infty$), we must pay careful attention to the path ($ds$) the photons are taking. (Note: We will drop the $\lambda$ subscript on $\tau$ from here on for visual simplicity).

1. Upwelling Radiation ($I^+$)

For a photon traveling upward from the surface to space, its path $ds$ is exactly in the $+dz$ direction. Therefore, $k_\lambda \rho ds = k_\lambda \rho dz = d\tau$. Substituting this into our general Schwarzschild equation:

The Differential Form:

\frac{dI^+_\lambda}{d\tau} = -I^+_\lambda + B_\lambda(T)

To find the total upwelling radiation reaching an arbitrary altitude $\tau$, we rearrange to $\frac{dI^+}{d\tau} + I^+ = B$, multiply by the integrating factor $e^\tau$, and integrate from the surface ($\tau = 0$) up to our observation level $\tau$.

The Integral Form:

I^+_\lambda(\tau) = I^+_\lambda(0) e^{-\tau} + \int_0^\tau B_\lambda(T(\tau')) e^{-(\tau - \tau')} d\tau'

This equation tells us that the upwelling radiation at level $\tau$ is the sum of two parts:

The Surface Term: The radiation originally emitted by the surface ($I^+(0)$), mathematically attenuated by the transmission function ($e^{-\tau}$) of the entire layer below us.
The Atmospheric Term: The sum of the thermal emissions ($B_\lambda$) from every layer of gas below us ($\tau'$), attenuated by the optical distance remaining between that emitting layer and our observation level ($e^{-(\tau - \tau')}$).

2. Downwelling Radiation ($I^-$)

For a photon traveling downward from space toward the surface, its path $ds$ is exactly opposite to our vertical coordinate. So, $ds = -dz$, which means $k_\lambda \rho ds = -d\tau$. Substituting this in:

The Differential Form:

-\frac{dI^-_\lambda}{d\tau} = -I^-_\lambda + B_\lambda(T) \implies \frac{dI^-_\lambda}{d\tau} = I^-_\lambda - B_\lambda(T)

To find the total downwelling radiation, we rearrange to $\frac{dI^-}{d\tau} - I^- = -B$, multiply by the integrating factor $e^{-\tau}$, and integrate from our observation level $\tau$ upward to the Top of Atmosphere ($\tau_\infty$).

The Integral Form:

I^-_\lambda(\tau) = I^-_\lambda(\tau_\infty) e^{-(\tau_\infty - \tau)} + \int_\tau^{\tau_\infty} B_\lambda(T(\tau')) e^{-(\tau' - \tau)} d\tau'

Similarly, this tells us the downwelling radiation is the Top of Atmosphere input ($I^-(\tau_\infty)$, which is zero for planetary infrared but non-zero for solar), attenuated by the distance down to our level, plus the sum of all atmospheric emissions from the layers above us.

VIII. Synthesis and Rigor: The Radiating Level

Looking at our upwelling integral equation, we can ask a profound physical question: If we point a satellite at the planet from space (where $\tau = \tau_\infty$), what specific layer of the atmosphere is that satellite actually "seeing"?

Evaluating our upwelling integral at the Top of Atmosphere ($\tau = \tau_\infty$):

I_\lambda^+(\tau_\infty) = I_\lambda(0) e^{-\tau_\infty} + \int_0^{\tau_\infty} B_\lambda(T(\tau')) e^{-(\tau_\infty - \tau')} d\tau'

If the atmosphere is optically thick (like Earth in the center of the $CO_2$ band), the total optical depth $\tau_\infty$ is very large. The surface term is completely absorbed ($e^{-\tau_\infty} \approx 0$). The radiation escaping to space is entirely governed by the atmospheric term:

I_\lambda^+(\tau_\infty) \approx \int_0^{\tau_\infty} B_\lambda(T(\tau')) e^{-(\tau_\infty - \tau')} d\tau'

To make this intuitive, let's define the "optical depth from space" as $\tau_{space} = \tau_\infty - \tau'$. This variable is $0$ at the satellite and increases as it looks down into the atmosphere. The transmission term simply becomes $e^{-\tau_{space}}$.

Our rule of thumb states that space "sees" down into the atmosphere to an optical depth of $\tau_{space} \approx 1$. We can prove this rigorously.

Proof 1: Maximizing the Weighting Function

The integrand above measures emission per unit optical depth. But $d\tau$ does not represent a constant physical thickness in meters. To find the physical altitude the satellite sees, we must evaluate emission per physical altitude ($dz$).

Deriving the Scale Height ($H$):
Atmospheric density is governed by Hydrostatic Balance ($\frac{dp}{dz} = -\rho g$) and the Ideal Gas Law ($p = \rho R_s T$). Combining these gives an exponential decay for density:

\rho(z) = \rho_0 e^{-z/H}

Here, $H = \frac{R_s T}{g}$ is the Scale Height (roughly 8 km for Earth).

The True Weighting Function:
Optical depth from space ($\tau_{space}$) is the integral of absorption from the top of the atmosphere down to altitude $z$:

\tau_{space}(z) = \int_{z}^{\infty} k_\lambda \rho_0 e^{-z'/H} dz' = k_\lambda H \rho_0 e^{-z/H} = k_\lambda H \rho(z)

Taking the derivative of this with respect to altitude ($z$):

\frac{d\tau_{space}}{dz} = -k_\lambda \rho_0 e^{-z/H} = -\frac{\tau_{space}}{H} \implies d\tau_{space} = \frac{\tau_{space}}{H} dz \quad \text{(magnitude)}

Substituting this geometric scaling into our original integral gives us the true Weighting Function ($W$) that determines which physical altitude emits the most escaping radiation:

W(\tau_{space}) \propto \tau_{space} e^{-\tau_{space}}

Taking the derivative $\frac{d}{d\tau_{space}}(\tau_{space} e^{-\tau_{space}})$ and setting it to zero yields exactly $\mathbf{\tau_{space} = 1}$. This is the perfect physical compromise: deep enough to have sufficient gas density to emit photons, but high enough that those photons can actually escape to space without being re-absorbed.

An Intuitive Heuristic: The Proportionality of Optical Depth
While the hydrostatic derivation above is mathematically rigorous, we can also arrive at this exact same geometric scaling using a much simpler, intuitive argument.

By definition, the optical thickness of a very thin physical slice $dz$ is $d\tau_{space} = k_\lambda \rho dz$. This means the local change in optical depth ($d\tau_{space}$) is directly proportional to the local gas density $\rho$.

Because an atmosphere is a compressible gas resting under its own weight, the local density of the gas at any altitude is directly proportional to the total mass of all the gas pressing down on it from above.

Since the total optical depth from space ($\tau_{space}$) is just a measure of that total mass of gas above us, it follows naturally that the local density ($\rho$) is directly proportional to the total optical depth from space ($\tau_{space}$).

Therefore, the local change in optical depth ($d\tau_{space}$) must be directly proportional to the total optical depth accumulated from space ($\tau_{space}$). We can write this geometric scaling mathematically as:

d\tau_{space} = \frac{\tau_{space}}{H} dz \quad \text{(magnitude)}

Here, $H$ is just the proportionality constant, representing a characteristic length scale of the atmosphere (known as the scale height). This physical shortcut immediately yields the $d\tau \propto \tau dz$ relationship necessary to derive the $\tau_{space}=1$ peak in the weighting function.

Proof 2: The Eddington-Barbier Approximation

If we assume the temperature profile (and thus the Planck function) varies roughly linearly with optical depth near the top of the atmosphere ($B_\lambda(\tau_{space}) \approx B_0 + B_1\tau_{space}$), we can solve the exact integral:

I_\lambda^+(\text{space}) \approx \int_0^{\infty} (B_0 + B_1 \tau_{space}) e^{-\tau_{space}} d\tau_{space}

Because $\int_0^{\infty} e^{-x} dx = 1$ and $\int_0^{\infty} x e^{-x} dx = 1$, the integral perfectly evaluates to:

I_\lambda^+(\text{space}) \approx B_0 + B_1

Evaluating our linear approximation explicitly at the layer exactly 1 optical thickness down from space yields the exact same result:

B_\lambda(\tau_{space}=1) = B_0 + B_1(1) = B_0 + B_1

Therefore:

I_\lambda^+(\text{space}) \approx B_\lambda(\tau_{space} = 1)

IX. Conclusion

The mathematical jump from algebra to calculus, and the shift from altitude to optical depth, rigorously proves what our simple Zero-D model only hinted at: The radiating temperature of a planet ($T_{rad}$) is the literal, physical temperature of the atmospheric layer sitting exactly one optical thickness down from space ($\tau_{space} = 1$).

As we add greenhouse gases to the atmosphere, we increase the absorption cross-section ($k_\lambda$). This pushes the $\tau_{space}=1$ radiating level to a higher, colder physical altitude. Because the planet is now radiating from a colder layer, it emits less energy to space, creating an energy imbalance that drives the surface temperature ($T_s$) up until equilibrium is restored.