6.2 Greenhouse Effect

The greenhouse effect is the process by which certain atmospheric gases absorb and re-emit infrared radiation, trapping thermal energy within the climate system and warming Earth's surface by approximately 33 K above its effective emission temperature. Without this natural greenhouse effect, Earth's mean surface temperature would be roughly -18 degrees C instead of the observed +15 degrees C, rendering the planet inhospitable to life as we know it.

Physical Mechanism

The greenhouse effect arises from the selective absorption of radiation by atmospheric gases. The atmosphere is largely transparent to incoming shortwave solar radiation (0.2-4 micrometers), allowing sunlight to reach and warm the surface. However, the atmosphere is substantially opaque to the outgoing longwave (thermal infrared) radiation emitted by the surface (4-100 micrometers). Greenhouse gas molecules possess vibrational and rotational modes that resonate at infrared frequencies, enabling them to absorb photons in these bands.

Upon absorbing an infrared photon, a greenhouse gas molecule transitions to an excited vibrational-rotational state. It then re-emits radiation isotropically -- equally in all directions. Crucially, about half of this re-emitted radiation is directed downward back toward the surface, creating additional downwelling longwave radiation that supplements the direct solar heating. This process occurs across multiple atmospheric layers, with each layer absorbing upwelling radiation from below and emitting both upward and downward based on its local temperature.

Key insight: The greenhouse effect does not "trap" heat in the sense of preventing its escape. Rather, it raises the effective emission altitude. Since temperature decreases with altitude (the lapse rate), emission from a higher, colder level is weaker, reducing the OLR and creating a net energy imbalance that warms the surface until a new equilibrium is reached.

Major Greenhouse Gases

The principal greenhouse gases differ in their atmospheric concentrations, absorption band positions, radiative efficiencies, and atmospheric lifetimes. The Global Warming Potential (GWP) measures the integrated radiative forcing of a pulse emission relative to CO₂ over a specified time horizon (typically 100 years):

H₂O (Water Vapor)

Concentration: highly variable, 0-4% by volume. Contributes approximately 60% of the natural greenhouse effect. Absorbs broadly across IR, with strong bands at 6.3 micrometers (bending mode) and the pure rotation band beyond 20 micrometers. Water vapor is a condensable gas -- its concentration is controlled by temperature through the Clausius-Clapeyron relation, making it a powerful feedback rather than a direct forcing agent. Its atmospheric lifetime is only ~10 days.

CO₂ (Carbon Dioxide)

Concentration: ~424 ppm (2024), pre-industrial ~280 ppm. GWP₁₀₀ = 1 (reference). Primary absorption at 15 micrometers (667 cm⁻¹ bending mode) and 4.3 micrometers. Responsible for ~25% of the natural greenhouse effect and is the dominant anthropogenic forcing agent. Well-mixed in the atmosphere with a perturbation lifetime of centuries to millennia. Current radiative forcing: ~2.2 W/m² relative to pre-industrial.

CH₄ (Methane)

Concentration: ~1.92 ppm, pre-industrial ~0.72 ppm. GWP₁₀₀ = 28 (without carbon-cycle feedbacks), GWP₂₀ = 80. Absorbs at 3.3 micrometers and 7.7 micrometers. Sources include wetlands, rice paddies, ruminant livestock, fossil fuel extraction, and biomass burning. Atmospheric lifetime: ~12 years. Radiative forcing: ~0.54 W/m².

N₂O (Nitrous Oxide)

Concentration: ~336 ppb, pre-industrial ~270 ppb. GWP₁₀₀ = 265. Absorbs near 4.5 micrometers and 7.8 micrometers. Sources include agricultural soils (nitrification/denitrification), combustion, and industrial processes. Atmospheric lifetime: ~121 years. Radiative forcing: ~0.21 W/m². Also plays a role in stratospheric ozone destruction.

O₃ (Ozone)

Concentration: variable, ~0.02-10 ppm depending on altitude. Absorbs at 9.6 micrometers (within the atmospheric window, making it especially effective), and at UV wavelengths (0.2-0.3 micrometers) in the stratospheric ozone layer. Tropospheric ozone is a short-lived greenhouse gas and pollutant. Radiative forcing: ~0.40 W/m² (tropospheric), ~-0.02 W/m² (stratospheric depletion).

CFCs & Halocarbons

Concentrations: ppt levels. GWP₁₀₀ ranges from 1,000 to 23,000. Absorb in the atmospheric window region (8-13 micrometers), where few other gases absorb, making them disproportionately effective per molecule. CFC-12 has a GWP of 10,200 and lifetime of 100 years. Regulated under the Montreal Protocol due to ozone depletion. Combined radiative forcing: ~0.34 W/m².

Absorption Spectra and Atmospheric Windows

The infrared absorption spectrum of the atmosphere is a complex superposition of millions of individual spectral lines from all greenhouse gases. Key absorption bands and windows include:

Major Absorption Bands

CO₂ 15 $\mu$m band: 13-17 $\mu$m (strongest IR absorber)

H₂O rotation band: >20 $\mu$m

H₂O 6.3 $\mu$m bending mode

O₃ 9.6 $\mu$m band

CH₄ 7.7 $\mu$m band

Atmospheric Windows

8-13 $\mu$m: Primary thermal IR window

3.5-4.0 $\mu$m: Near-IR window

0.3-0.7 $\mu$m: Visible window

CFCs and HFCs absorb within the 8-13 $\mu$m window, explaining their extreme GWP values

Radiative Forcing from CO₂

The radiative forcing from increased CO₂ follows a logarithmic relationship with concentration. This logarithmic dependence arises because the center of the 15 micrometer absorption band is already saturated (optically thick), and additional CO₂ primarily broadens the absorption into the wings of the band where it is not yet saturated:

$$\Delta F = 5.35 \ln\left(\frac{C}{C_0}\right) \text{ W/m}^2$$

Myhre et al. (1998) simplified expression for CO₂ radiative forcing

For a doubling of CO₂ (C/C₀ = 2):

$$\Delta F_{2\times\text{CO}_2} = 5.35 \ln(2) = 5.35 \times 0.693 \approx 3.7 \text{ W/m}^2$$

Radiative forcing for CO₂ doubling from 280 to 560 ppm

For other greenhouse gases, the forcing relationships differ. Methane and nitrous oxide have approximately square-root dependencies because their absorption bands are less saturated:

$$\Delta F_{\text{CH}_4} = 0.036 \left(\sqrt{M} - \sqrt{M_0}\right)$$

M in ppb; includes overlap corrections with N₂O

$$\Delta F_{\text{N}_2\text{O}} = 0.12 \left(\sqrt{N} - \sqrt{N_0}\right)$$

N in ppb; includes overlap corrections with CH₄

Simple Layer Greenhouse Models

The greenhouse effect can be understood through a hierarchy of simple models. In the simplest single-layer model, the atmosphere is treated as a single slab that is transparent to shortwave radiation but absorbs all longwave radiation (emissivity $\varepsilon = 1$). The atmosphere radiates both upward to space and downward to the surface:

$$\text{TOA balance: } \frac{S_0(1-\alpha)}{4} = \sigma T_a^4$$

$$\text{Surface balance: } \frac{S_0(1-\alpha)}{4} + \sigma T_a^4 = \sigma T_s^4$$

$$\Rightarrow T_s = 2^{1/4} \, T_e \approx 1.189 \, T_e \approx 303 \text{ K}$$

Single-layer greenhouse model: surface is warmed by atmospheric back-radiation

Extending to N layers, each fully absorbing and emitting in the IR, the surface temperature becomes:

$$T_s = (N+1)^{1/4} \, T_e$$

N-layer greenhouse model: each layer adds to the thermal blanketing effect

N = 0

T_s = 255 K

No atmosphere

N = 1

T_s = 303 K

Single layer

N = 2

T_s = 335 K

Two layers

Spectral Line Shapes and HITRAN

Real absorption spectra consist of millions of individual spectral lines, each with a characteristic shape determined by the broadening mechanism. The HITRAN database (High-Resolution Transmission molecular absorption) catalogs the positions, intensities, and broadening parameters of over 7.4 million spectral lines for 55 molecules.

The three fundamental line shape profiles are:

Lorentz Profile (Pressure Broadening)

$$f_L(\nu) = \frac{1}{\pi} \frac{\gamma_L}{(\nu - \nu_0)^2 + \gamma_L^2}$$

Dominates in the lower atmosphere (below ~30 km). Half-width $\gamma_L \propto p/\sqrt{T}$. At sea level, typical half-widths are 0.05-0.10 cm⁻¹. Broad Lorentzian wings are important for continuum absorption.

Doppler Profile (Thermal Broadening)

$$f_D(\nu) = \frac{1}{\gamma_D \sqrt{\pi}} \exp\left[-\left(\frac{\nu - \nu_0}{\gamma_D}\right)^2\right]$$

Dominates in the upper atmosphere (above ~50 km) where pressure is low. Half-width $\gamma_D = (\nu_0/c)\sqrt{2k_BT/m}$. Gaussian shape with narrower core than Lorentz profile.

Voigt Profile (Combined)

$$f_V(\nu) = \int_{-\infty}^{\infty} f_L(\nu') \, f_D(\nu - \nu') \, d\nu'$$

Convolution of Lorentz and Doppler profiles. Used in the 20-50 km altitude range where both mechanisms are significant. Computationally expensive; approximations like the Humlicek algorithm are employed.

Equivalent Width and Curve of Growth

The equivalent width W of a spectral line is a measure of the total absorption integrated across the line, defined as:

$$W = \int_0^{\infty} \left(1 - e^{-\tau(\nu)}\right) d\nu$$

Equivalent width: total absorption expressed as width of a perfectly absorbing line

The curve of growth describes how W increases with absorber amount u. It has three distinct regimes:

Weak Line Regime

$W \propto u$

Linear growth: line is optically thin everywhere. Absorption proportional to amount of absorber.

Strong Line Regime

$W \propto \sqrt{u}$

Square-root growth: line center is saturated, only Lorentz wings contribute. This is the regime for CO₂ at 15 $\mu$m (explains logarithmic forcing).

Very Strong Regime

$W \propto \ln(u)$

Logarithmic growth: line and near wings are saturated. Additional absorber has diminishing effect.

Radiative-Convective Equilibrium

A purely radiative equilibrium atmosphere would have an extremely steep temperature gradient near the surface (lapse rate exceeding 10 K/km), which is convectively unstable. In reality, convection rapidly redistributes heat vertically, constraining the lapse rate to approximately the moist adiabatic value (~6.5 K/km in the troposphere). The radiative-convective equilibrium (RCE) model adjusts the temperature profile so that:

$$\frac{\partial T}{\partial z}\bigg|_{\text{rad}} < -\Gamma_c \implies \text{Convective adjustment applied}$$

$$\Gamma = -\frac{dT}{dz} = \begin{cases} \Gamma_{\text{rad}} & \text{if } \Gamma_{\text{rad}} \leq \Gamma_c \\ \Gamma_c \approx 6.5 \text{ K/km} & \text{if } \Gamma_{\text{rad}} > \Gamma_c \end{cases}$$

Radiative-convective equilibrium: convection limits the lapse rate

The classic Manabe & Strickler (1964) RCE calculation showed that with observed CO₂ and H₂O concentrations and a critical lapse rate of 6.5 K/km, the model produces a surface temperature of approximately 288 K, matching observations. Doubling CO₂ in this model yields a surface warming of approximately 2.3 K (before feedbacks).

Water Vapor Feedback and Clausius-Clapeyron

As the surface warms, the saturation vapor pressure increases exponentially according to the Clausius-Clapeyron equation:

$$\frac{de_s}{dT} = \frac{L_v \, e_s}{R_v \, T^2} \quad \Rightarrow \quad e_s(T) \approx e_s(T_0) \exp\left[\frac{L_v}{R_v}\left(\frac{1}{T_0} - \frac{1}{T}\right)\right]$$

This gives approximately 7% increase in saturation vapor pressure per kelvin of warming. If relative humidity remains roughly constant (as observations and models suggest), the absolute amount of water vapor increases, amplifying the initial warming. This is the strongest positive feedback in the climate system, roughly doubling the warming from CO₂ alone.

Radiative Transfer Methods

Computing radiative fluxes through the atmosphere requires integrating absorption and emission over the full spectrum. Three main approaches exist, in order of decreasing accuracy and computational cost:

Line-by-Line (LBL)

Resolves every spectral line individually at extremely fine spectral resolution (~0.001 cm⁻¹). Uses HITRAN line parameters. Serves as benchmark truth for other methods. Computationally expensive: millions of monochromatic calculations per profile. Used for validation, not for GCMs.

Band Models

Divide the spectrum into spectral bands (10-50 cm⁻¹ wide) and compute average transmission for each band. Statistical approaches like the Goody random model and Malkmus model treat line positions and intensities statistically. RRTM (Rapid Radiative Transfer Model) uses the correlated-k method for efficiency.

Correlated-k Method

Groups spectral points by absorption coefficient strength rather than wavelength. Reduces spectral integration to ~16 quadrature points per band. Used in most modern GCM radiation schemes (RRTMG, ecRad). Accurate to within 1 W/m² of LBL for broadband fluxes.

Fortran: Radiative Forcing Calculator

This Fortran program computes the radiative forcing for various CO₂ concentrations, including historical and projected future scenarios, using the logarithmic forcing relationship and simplified expressions for other greenhouse gases:

program radiative_forcing
  ! Compute radiative forcing for various greenhouse gas concentrations
  ! Uses Myhre et al. (1998) simplified expressions
  implicit none

  real(8), parameter :: CO2_preind = 280.0d0   ! Pre-industrial CO2 (ppm)
  real(8), parameter :: CH4_preind = 722.0d0   ! Pre-industrial CH4 (ppb)
  real(8), parameter :: N2O_preind = 270.0d0   ! Pre-industrial N2O (ppb)
  real(8), parameter :: sigma = 5.670d-8       ! Stefan-Boltzmann constant
  real(8), parameter :: lambda0 = 0.31d0       ! No-feedback sensitivity (K/(W/m^2))

  real(8) :: CO2_vals(8), dF_CO2, dF_CH4, dF_N2O, dF_total
  real(8) :: dT_nofb, dT_withfb
  real(8) :: CH4_current, N2O_current
  real(8) :: feedback_factor
  integer :: i, j

  ! CO2 scenarios (ppm)
  CO2_vals = (/ 280.0d0, 350.0d0, 400.0d0, 420.0d0, &
                500.0d0, 560.0d0, 700.0d0, 1000.0d0 /)

  ! Current CH4 and N2O
  CH4_current = 1920.0d0   ! ppb (2024 value)
  N2O_current = 336.0d0    ! ppb (2024 value)

  ! Header
  write(*,'(A)') '=============================================='
  write(*,'(A)') '  Radiative Forcing from CO2 Concentration'
  write(*,'(A)') '=============================================='
  write(*,'(A)') '  CO2    |  dF_CO2  | dT(no fb) | dT(with fb)'
  write(*,'(A)') '  (ppm)  | (W/m^2)  |    (K)    |    (K)'
  write(*,'(A)') '----------------------------------------------'

  feedback_factor = 2.8d0   ! Typical feedback amplification

  do i = 1, 8
    ! CO2 forcing: dF = 5.35 * ln(C/C0)
    dF_CO2 = 5.35d0 * log(CO2_vals(i) / CO2_preind)

    ! Temperature change without feedbacks
    dT_nofb = lambda0 * dF_CO2

    ! Temperature change with feedbacks (amplified)
    dT_withfb = dT_nofb * feedback_factor

    write(*,'(F7.0, A, F8.2, A, F9.2, A, F9.2)') &
          CO2_vals(i), ' |', dF_CO2, '  |', dT_nofb, '  |', dT_withfb
  end do

  ! --- Combined forcing from all GHGs ---
  write(*,'(/,A)') '=============================================='
  write(*,'(A)')   '  Combined GHG Forcing (current vs pre-ind)'
  write(*,'(A)')   '=============================================='

  ! CO2 forcing
  dF_CO2 = 5.35d0 * log(420.0d0 / CO2_preind)

  ! CH4 forcing (simplified Myhre formula)
  dF_CH4 = 0.036d0 * (sqrt(CH4_current) - sqrt(CH4_preind))

  ! N2O forcing (simplified Myhre formula)
  dF_N2O = 0.12d0 * (sqrt(N2O_current) - sqrt(N2O_preind))

  dF_total = dF_CO2 + dF_CH4 + dF_N2O

  write(*,'(A, F8.3, A)') '  CO2 forcing:   ', dF_CO2, ' W/m^2'
  write(*,'(A, F8.3, A)') '  CH4 forcing:   ', dF_CH4, ' W/m^2'
  write(*,'(A, F8.3, A)') '  N2O forcing:   ', dF_N2O, ' W/m^2'
  write(*,'(A)')           '  ----------------------------'
  write(*,'(A, F8.3, A)') '  Total forcing: ', dF_total, ' W/m^2'
  write(*,'(A, F8.3, A)') '  dT (no fb):    ', lambda0 * dF_total, ' K'
  write(*,'(A, F8.3, A)') '  dT (with fb):  ', &
        lambda0 * dF_total * feedback_factor, ' K'

  ! --- Time series: CO2 trajectory ---
  write(*,'(/,A)') '=============================================='
  write(*,'(A)')   '  CO2 Trajectory and Forcing (1750-2100)'
  write(*,'(A)')   '=============================================='
  write(*,'(A)')   '  Year  | CO2(ppm) | dF(W/m^2) | dT_eq(K)'
  write(*,'(A)')   '----------------------------------------------'

  ! Simplified CO2 trajectory
  do j = 1750, 2100, 25
    if (j <= 1850) then
      dF_CO2 = 5.35d0 * log((280.0d0) / CO2_preind)
    else if (j <= 2024) then
      ! Approximate exponential growth
      dF_CO2 = 5.35d0 * log((280.0d0 + &
               140.0d0 * ((dble(j) - 1850.0d0) / 174.0d0)**2) / CO2_preind)
    else
      ! SSP2-4.5 projection (approximate)
      dF_CO2 = 5.35d0 * log((420.0d0 + &
               2.5d0 * (dble(j) - 2024.0d0)) / CO2_preind)
    end if
    dT_withfb = lambda0 * dF_CO2 * feedback_factor
    write(*,'(I6, A, F8.1, A, F9.3, A, F8.2)') &
          j, '  |', CO2_preind * exp(dF_CO2 / 5.35d0), &
          ' |', dF_CO2, '  |', dT_withfb
  end do

end program radiative_forcing

Interactive Simulation: Simple Layer Greenhouse Model

Python

Model Earth with 0, 1, 2, ... N atmospheric layers. Calculates surface temperature for each case using T_s = (N+1)^(1/4) * T_e. Also shows how varying emissivity affects surface temperature.

greenhouse_layer_model.py101 lines

import numpy as np
import matplotlib.pyplot as plt

# Physical constants
sigma = 5.670e-8    # Stefan-Boltzmann constant (W/m^2/K^4)
S0 = 1361.0         # Solar constant (W/m^2)
alpha = 0.30        # Planetary albedo
F_solar = S0 * (1 - alpha) / 4  # Absorbed solar flux = 238.2 W/m^2
T_e = (F_solar / sigma)**0.25    # Effective temperature = 255 K

# ---- Model 1: N fully-absorbing layers ----
N_layers = np.arange(0, 21)
T_sfc_full = (N_layers + 1)**0.25 * T_e

# ---- Model 2: Partial absorber (varying emissivity) ----
def greenhouse_emissivity(N, eps):
    """Iterative solution for N-layer model with emissivity eps."""
    if N == 0:
        return T_e
    T = np.full(N + 1, T_e)  # T[0]=surface, T[1..N]=layers
    for iteration in range(1000):
        T_old = T.copy()
        # Surface: absorbs solar + downwelling from layer 1
        T[0] = ((F_solar + eps * sigma * T[1]**4) / sigma)**0.25
        # Interior layers
        for k in range(1, N):
            flux_in = eps * sigma * (T[k-1]**4 + T[k+1]**4)
            T[k] = (flux_in / (2 * eps * sigma))**0.25
        # Top layer
        T[N] = ((eps * sigma * T[N-1]**4) / (2 * eps * sigma))**0.25
        if np.max(np.abs(T - T_old)) < 0.001:
            break
    return T[0]

emissivities = np.linspace(0.05, 1.0, 40)
T_sfc_eps = np.array([greenhouse_emissivity(2, e) for e in emissivities])

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14, 6))
fig.patch.set_facecolor('#0f172a')
for ax in [ax1, ax2]:
    ax.set_facecolor('#1e293b')
    ax.tick_params(colors='#cbd5e1')
    for spine in ax.spines.values():
        spine.set_color('#334155')
    ax.grid(True, color='#334155', alpha=0.4)

# Panel 1: Surface temperature vs number of layers
ax1.plot(N_layers, T_sfc_full, 'o-', color='#f97316', lw=2.5, ms=6,
         label='Full absorber (\u03b5 = 1)')
ax1.axhline(y=288, color='#2dd4bf', ls='--', lw=1.5, label='Observed T_s = 288 K')
ax1.axhline(y=T_e, color='#60a5fa', ls='--', lw=1.5, label=f'T_e = {T_e:.0f} K (no atmosphere)')
# Mark key points
ax1.plot(0, T_e, 'o', color='#60a5fa', ms=10, zorder=5)
ax1.plot(1, (2)**0.25 * T_e, 's', color='#fbbf24', ms=10, zorder=5)
ax1.annotate(f'N=0: {T_e:.0f} K', (0, T_e), xytext=(1.5, T_e-10),
             color='#cbd5e1', fontsize=9)
ax1.annotate(f'N=1: {(2)**0.25*T_e:.0f} K', (1, (2)**0.25*T_e),
             xytext=(2.5, (2)**0.25*T_e+5), color='#cbd5e1', fontsize=9)

ax1.set_xlabel('Number of Atmospheric Layers N', color='#cbd5e1', fontsize=12)
ax1.set_ylabel('Surface Temperature (K)', color='#cbd5e1', fontsize=12)
ax1.set_title('N-Layer Greenhouse: T_s = (N+1)$^{1/4}$ T_e', color='#cbd5e1', fontsize=13)
ax1.legend(fontsize=9, facecolor='#1e293b', edgecolor='#334155', labelcolor='#cbd5e1')
ax1.set_xlim(0, 20)
ax1.set_ylim(240, 400)

# Panel 2: Surface temperature vs emissivity (for N=1,2,3)
for N, c, ls in [(1, '#38bdf8', '-'), (2, '#a78bfa', '--'), (3, '#f87171', ':')]:
    T_arr = np.array([greenhouse_emissivity(N, e) for e in emissivities])
    ax2.plot(emissivities, T_arr, color=c, lw=2.5, ls=ls, label=f'N = {N} layers')

ax2.axhline(y=288, color='#2dd4bf', ls='--', lw=1.5, label='Observed T_s = 288 K')
ax2.axhline(y=T_e, color='#60a5fa', ls='--', lw=1.5, alpha=0.5)
ax2.set_xlabel('Layer Emissivity \u03b5', color='#cbd5e1', fontsize=12)
ax2.set_ylabel('Surface Temperature (K)', color='#cbd5e1', fontsize=12)
ax2.set_title('Effect of Emissivity on Greenhouse Warming', color='#cbd5e1', fontsize=13)
ax2.legend(fontsize=9, facecolor='#1e293b', edgecolor='#334155', labelcolor='#cbd5e1')
ax2.set_xlim(0, 1)
ax2.set_ylim(250, 360)

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor=fig.get_facecolor())

print("Simple Layer Greenhouse Model")
print("=" * 50)
print(f"Absorbed solar flux: {F_solar:.1f} W/m^2")
print(f"Effective temperature T_e: {T_e:.1f} K ({T_e - 273.15:.1f} C)")
print()
print("Full absorber (eps = 1):")
print(f"  N=0 (no atmosphere): T_s = {T_e:.1f} K")
for N in [1, 2, 3, 5, 10]:
    Ts = (N + 1)**0.25 * T_e
    print(f"  N={N}: T_s = {Ts:.1f} K ({Ts - 273.15:.1f} C), warming = +{Ts - T_e:.1f} K")
print()
print("Partial absorber (N=2):")
for eps in [0.2, 0.4, 0.6, 0.8, 1.0]:
    Ts = greenhouse_emissivity(2, eps)
    print(f"  eps={eps:.1f}: T_s = {Ts:.1f} K ({Ts - 273.15:.1f} C)")
print()
print("Earth needs ~1-2 layers at eps~0.8 to match observed 288 K.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

← 6.1 Radiation Budget 6.3 Climate Feedbacks →