← Back to Part 6

6.3 Climate Feedbacks

Climate feedbacks are processes that amplify or dampen the initial temperature response to a radiative forcing. They are the primary source of uncertainty in climate projections, determining whether the equilibrium climate sensitivity to CO₂ doubling is closer to 2 degrees C or 5 degrees C. Understanding feedback mechanisms is essential for constraining future warming and evaluating the risk of abrupt climate transitions.

Feedback Loop Framework

In control theory, a feedback loop modifies the system response to an external perturbation. For the climate system, an initial radiative forcing $\Delta F$ (e.g., from CO₂ increase) produces a temperature change, which in turn modifies other climate variables (water vapor, ice cover, clouds), which then further modify the radiation balance. The total system gain G is:

$$G = \frac{1}{1 - f} \quad \text{where } f = \sum_i f_i \text{ is the total feedback factor}$$

System gain: f < 0 stabilizes (negative feedback), 0 < f < 1 amplifies, f > 1 is runaway

The feedback factor f represents the fraction of the output that is fed back to the input. If f is positive, the system amplifies perturbations (positive feedback). If f is negative, the system damps perturbations (negative feedback). For stability, we require $f < 1$; if $f \geq 1$, the system enters a runaway state. The individual feedback factors are approximately additive:

$$f = f_{\text{Planck}} + f_{\text{WV}} + f_{\text{LR}} + f_{\text{ice}} + f_{\text{cloud}} + \ldots$$

Total feedback is the sum of individual feedback processes

Climate Sensitivity Parameter

The no-feedback (Planck) climate sensitivity parameter $\lambda_0$ is derived by linearizing the Stefan-Boltzmann law around the equilibrium temperature. If the only response to a forcing is increased blackbody emission:

$$\lambda_0 = \frac{1}{4\sigma T_e^3} \approx \frac{1}{4 \times 5.67 \times 10^{-8} \times 255^3} \approx 0.27 \text{ K/(W/m}^2\text{)}$$

Planck sensitivity: temperature change per unit forcing without feedbacks

Using the emission temperature from TOA rather than the surface, the commonly cited value is $\lambda_0 \approx 0.3$ K/(W/m²). The equilibrium climate sensitivity (ECS) is the eventual surface temperature change after all fast feedbacks have fully responded to a CO₂ doubling:

$$\Delta T_{\text{ECS}} = \frac{\lambda_0}{1 - f} \cdot \Delta F_{2\times\text{CO}_2} = \frac{0.3}{1 - f} \times 3.7 \approx \frac{1.1}{1 - f} \text{ K}$$

ECS: no-feedback warming (1.1 K) amplified by the feedback factor

An alternative formulation uses the total feedback parameter $\lambda$ (in W/m²/K), where each feedback contributes a term $\lambda_i$:

$$\Delta T = \frac{\Delta F}{-\sum_i \lambda_i} = \frac{\Delta F}{\lambda_{\text{Planck}} + \lambda_{\text{WV}} + \lambda_{\text{LR}} + \lambda_{\text{ice}} + \lambda_{\text{cloud}}}$$

Feedback parameter formulation: $\lambda_i$ in W/m²/K (negative stabilizes, positive destabilizes)

ECS (Equilibrium Climate Sensitivity)

Equilibrium warming for 2x CO₂ after all fast feedbacks equilibrate. Timescale: centuries. IPCC AR6 best estimate: 3.0 degrees C (likely range: 2.5-4.0 degrees C). Includes ocean thermal equilibration.

TCR (Transient Climate Response)

Warming at the time of CO₂ doubling under 1%/year CO₂ increase. Timescale: ~70 years. IPCC AR6 best estimate: 1.8 degrees C (likely range: 1.4-2.2 degrees C). Accounts for ocean heat uptake delaying full warming. TCR/ECS ratio is typically 0.5-0.7.

Planck Feedback (Stabilizing Baseline)

The Planck feedback is the fundamental stabilizing mechanism in the climate system. As temperature increases, the outgoing longwave radiation increases according to the Stefan-Boltzmann law. This provides a restoring force that limits warming:

$$\lambda_{\text{Planck}} = -4\sigma T_e^3 \approx -3.3 \text{ W/m}^2\text{/K}$$

Planck feedback parameter: always negative (stabilizing)

In feedback factor notation: $f_{\text{Planck}} = -1$. This is defined as the reference state against which all other feedbacks are measured. Without any other feedbacks, a 3.7 W/m² forcing (CO₂ doubling) would produce a warming of 3.7/3.3 = 1.1 K.

Water Vapor Feedback

The water vapor feedback is the strongest positive feedback in the climate system. As temperature rises, the Clausius-Clapeyron equation dictates that the saturation vapor pressure increases at approximately 7% per kelvin:

$$\frac{1}{e_s}\frac{de_s}{dT} \approx \frac{L_v}{R_v T^2} \approx 0.07 \text{ K}^{-1} \quad \text{(at T = 288 K)}$$

Clausius-Clapeyron scaling: ~7% increase in saturation vapor pressure per kelvin

Observations and models consistently show that relative humidity remains approximately constant under warming, so the absolute water vapor concentration tracks the Clausius-Clapeyron increase. Since water vapor is the dominant greenhouse gas, this increased concentration significantly enhances the greenhouse effect:

$\lambda_{\text{WV}} \approx +1.8$ W/m²/K

Feedback parameter

$f_{\text{WV}} \approx +0.55$

Feedback factor (approximately doubles warming)

Ice-Albedo Feedback

The ice-albedo feedback operates through the contrast in reflectivity between ice-covered and ice-free surfaces. Sea ice and snow have albedos of 0.6-0.9, while open ocean has an albedo of only 0.06-0.10. As warming reduces ice and snow extent, the surface absorbs more solar radiation, amplifying the warming:

$$\lambda_{\text{ice}} = -\frac{S_0}{4} \frac{\partial \alpha}{\partial T} \approx +0.3 \text{ W/m}^2\text{/K}$$

Ice-albedo feedback: positive because albedo decreases as temperature increases

$f_{\text{ice}} \approx +0.10$

Feedback factor

Strongest at high latitudes where ice margin shifts. Arctic amplification (2-3x global mean warming) is partly due to this feedback. This feedback is nonlinear -- it weakens as ice disappears entirely and strengthens near the ice margin.

Lapse Rate Feedback

The lapse rate feedback arises from changes in the vertical temperature profile. In the tropics, the lapse rate is constrained by moist adiabatic processes. As surface temperature increases, the moist adiabatic lapse rate decreases (the atmosphere warms more at upper levels relative to the surface), which increases OLR more efficiently than a uniform warming would:

$$\lambda_{\text{LR}} \approx -0.5 \text{ to } -0.8 \text{ W/m}^2\text{/K}$$

Lapse rate feedback: negative (stabilizing) globally, but varies by latitude

$f_{\text{LR}} \approx -0.5$

Feedback factor (negative, stabilizing)

The lapse rate feedback is anti-correlated with the water vapor feedback. In the tropics, upper-tropospheric warming is strong (negative lapse rate feedback) but water vapor increase is also largest there. The combined WV + LR feedback is more robustly constrained than either individually: $\lambda_{\text{WV+LR}} \approx +1.1$ W/m²/K.

Cloud Feedbacks

Cloud feedbacks remain the largest source of uncertainty in climate sensitivity. Clouds exert two competing radiative effects: a shortwave cooling effect (reflecting sunlight) and a longwave warming effect (trapping IR radiation). The net effect depends on cloud type, altitude, optical depth, and coverage:

Low Clouds (SW dominant)

Warm and thick, strong reflection of sunlight. SW CRE: -47 W/m². LW CRE: +25 W/m². Net: -22 W/m² (cooling). If low cloud cover decreases with warming, this is a positive feedback. Marine stratocumulus breakup is a key mechanism.

High Clouds (LW dominant)

Cold and thin, strong IR trapping. SW CRE: -5 W/m². LW CRE: +30 W/m². Net: +25 W/m² (warming). The Fixed Anvil Temperature (FAT) hypothesis suggests high clouds rise with warming, maintaining their temperature but increasing the greenhouse effect.

$$\lambda_{\text{cloud}} \approx +0.4 \pm 0.4 \text{ W/m}^2\text{/K}$$

Cloud feedback: likely positive but with large uncertainty (IPCC AR6)

$f_{\text{cloud}} \approx +0.12$ (range: -0.1 to +0.3). Recent progress has come from high-resolution LES simulations, satellite observations (CERES, CloudSat, CALIPSO), and process-level understanding. The IPCC AR6 concluded that the net cloud feedback is very likely positive, narrowing uncertainty compared to earlier assessments.

Combined Feedback Budget

The total feedback determines the equilibrium climate sensitivity. The following table summarizes the major feedback contributions based on IPCC AR6 and recent literature:

Feedback$\lambda_i$ (W/m²/K)$f_i$SignConfidence
Planck-3.3-1.00NegativeVery high
Water vapor+1.8+0.55PositiveVery high
Lapse rate-0.6-0.18NegativeHigh
Ice-albedo+0.3+0.10PositiveHigh
Cloud+0.4+0.12PositiveMedium
Total-1.4-0.41Net negative-

$$\text{ECS} = \frac{\Delta F_{2\times\text{CO}_2}}{-\sum_i \lambda_i} = \frac{3.7}{1.4} \approx 2.6 \text{ K}$$

Central estimate consistent with IPCC AR6 likely range of 2.5-4.0 degrees C

Tipping Points and Abrupt Transitions

Beyond the linear feedback framework, the climate system contains potential tipping elements -- subsystems that may undergo rapid, self-reinforcing, and potentially irreversible transitions once a critical threshold is crossed. These represent highly nonlinear feedbacks that could fundamentally alter the climate trajectory:

AMOC Collapse

The Atlantic Meridional Overturning Circulation transports ~1.3 PW of heat northward. Freshwater from Greenland ice melt could weaken or shut down the AMOC, causing regional cooling in NW Europe (up to 5-10 K) while the Southern Hemisphere warms. Threshold estimated at 1.5-4 degrees C global warming. Recovery could take centuries.

Ice Sheet Instability

The West Antarctic Ice Sheet (WAIS) rests on bedrock below sea level, making it susceptible to Marine Ice Sheet Instability (MISI) and Marine Ice Cliff Instability (MICI). Complete WAIS collapse would raise global sea level by ~3.3 m. The Greenland Ice Sheet (GIS) has a threshold of ~1.5-2 degrees C for irreversible loss (~7 m sea level rise over millennia).

Permafrost Carbon Release

Arctic permafrost stores approximately 1,500 Gt of organic carbon -- nearly twice the current atmospheric carbon pool. Thawing permafrost releases CO₂ and CH₄ through microbial decomposition. This constitutes a positive feedback: warming thaws permafrost, releasing greenhouse gases, causing more warming. Estimates suggest 5-15% of permafrost carbon could be released by 2100 under high-emission scenarios.

Amazon Dieback

The Amazon rainforest generates ~25-50% of its own rainfall through evapotranspiration recycling. Deforestation and warming could push the system past a critical threshold where the forest can no longer sustain itself, transitioning to savanna. This would release approximately 50-90 Gt C and eliminate a major carbon sink. The threshold may lie at 20-25% deforestation combined with 3-4 degrees C warming.

Fortran: Time-Dependent Feedback Model

This Fortran program implements a time-dependent energy balance model with explicit feedback processes, showing how the surface temperature evolves toward equilibrium after a step-function CO₂ doubling:

program feedback_model
  ! Time-dependent energy balance model with climate feedbacks
  ! Tracks temperature evolution after CO2 doubling
  implicit none

  integer, parameter :: nyr = 500          ! Integration length (years)
  real(8), parameter :: dt = 3.1557d7      ! Time step = 1 year (seconds)
  real(8), parameter :: sigma = 5.670d-8   ! Stefan-Boltzmann constant
  real(8), parameter :: C_ocean = 4.2d8    ! Ocean mixed layer heat capacity (J/m^2/K)
  real(8), parameter :: C_deep = 1.0d10    ! Deep ocean heat capacity (J/m^2/K)
  real(8), parameter :: gamma = 0.8d0      ! Ocean heat exchange coeff (W/m^2/K)

  real(8), parameter :: S0 = 1361.0d0      ! Solar constant (W/m^2)
  real(8), parameter :: alpha0 = 0.30d0    ! Base planetary albedo
  real(8), parameter :: T_e = 255.0d0      ! Effective emission temperature (K)
  real(8), parameter :: dF_2xCO2 = 3.7d0   ! CO2 doubling forcing (W/m^2)

  ! Feedback parameters (W/m^2/K)
  real(8), parameter :: lam_planck = -3.3d0
  real(8), parameter :: lam_wv = 1.8d0
  real(8), parameter :: lam_lr = -0.6d0
  real(8), parameter :: lam_ice = 0.3d0
  real(8), parameter :: lam_cloud = 0.4d0

  real(8) :: T_s(0:nyr), T_deep(0:nyr)
  real(8) :: dT, dT_deep
  real(8) :: forcing, feedback_total, net_flux
  real(8) :: lambda_total, ECS_predicted
  real(8) :: fb_wv, fb_lr, fb_ice, fb_cloud
  integer :: yr

  ! Total feedback parameter
  lambda_total = lam_planck + lam_wv + lam_lr + lam_ice + lam_cloud
  ECS_predicted = dF_2xCO2 / (-lambda_total)

  write(*,'(A)') '=============================================='
  write(*,'(A)') '  Time-Dependent Feedback Model'
  write(*,'(A)') '=============================================='
  write(*,'(A,F8.3,A)') '  Total feedback parameter: ', lambda_total, ' W/m^2/K'
  write(*,'(A,F8.2,A)') '  Predicted ECS:            ', ECS_predicted, ' K'
  write(*,'(A,F8.2,A)') '  Ocean mixed layer C:      ', C_ocean/1d6, ' MJ/m^2/K'
  write(*,'(A)')         '=============================================='

  ! Initial conditions (pre-industrial equilibrium)
  T_s(0) = 0.0d0        ! Anomaly from equilibrium
  T_deep(0) = 0.0d0     ! Deep ocean anomaly

  ! Time integration (forward Euler)
  do yr = 1, nyr
    dT = T_s(yr-1)
    dT_deep = T_deep(yr-1)

    ! Step-function forcing (CO2 doubles at year 0)
    forcing = dF_2xCO2

    ! Compute individual feedback contributions
    fb_wv = lam_wv * dT        ! Water vapor feedback
    fb_lr = lam_lr * dT        ! Lapse rate feedback
    fb_ice = lam_ice * dT      ! Ice-albedo feedback
    fb_cloud = lam_cloud * dT  ! Cloud feedback

    ! Total feedback (including Planck)
    feedback_total = lambda_total * dT

    ! Net energy flux into mixed layer
    net_flux = forcing + feedback_total - gamma * (dT - dT_deep)

    ! Update temperatures
    T_s(yr) = T_s(yr-1) + (dt / C_ocean) * net_flux
    T_deep(yr) = T_deep(yr-1) + (dt / C_deep) * gamma * (dT - dT_deep)
  end do

  ! Output time series
  write(*,'(/,A)') '  Year  |  dT_sfc  | dT_deep | Net Flux | % of ECS'
  write(*,'(A)')   '  ------|----------|---------|----------|--------'
  do yr = 0, nyr, 25
    net_flux = dF_2xCO2 + lambda_total * T_s(yr) - &
               gamma * (T_s(yr) - T_deep(yr))
    write(*,'(I6, A, F8.3, A, F7.3, A, F8.3, A, F7.1, A)') &
          yr, ' |', T_s(yr), ' |', T_deep(yr), ' |', &
          net_flux, ' |', 100.0d0 * T_s(yr) / ECS_predicted, '%'
  end do

  ! Report characteristic timescales
  write(*,'(/,A)') '  Characteristic timescales:'
  write(*,'(A,F8.1,A)') '    Fast response (mixed layer):  ', &
        C_ocean / (-lambda_total) / 3.1557d7, ' years'
  write(*,'(A,F8.1,A)') '    Slow response (deep ocean):   ', &
        C_deep / gamma / 3.1557d7, ' years'
  write(*,'(A,F8.3,A)') '    Final dT_surface (yr 500):    ', T_s(nyr), ' K'
  write(*,'(A,F8.3,A)') '    Predicted ECS:                ', ECS_predicted, ' K'
  write(*,'(A,F8.1,A)') '    Fraction of ECS realized:     ', &
        100.0d0 * T_s(nyr) / ECS_predicted, '%'

  ! TCR estimate (warming at year 70 with 1%/yr CO2 increase)
  write(*,'(/,A)') '  --- TCR Estimate ---'
  ! For 1%/yr CO2 increase, forcing increases linearly: F(t) = (dF/70) * t
  T_s(0) = 0.0d0
  T_deep(0) = 0.0d0
  do yr = 1, 70
    dT = T_s(yr-1)
    dT_deep = T_deep(yr-1)
    forcing = dF_2xCO2 * dble(yr) / 70.0d0  ! Linear ramp
    feedback_total = lambda_total * dT
    net_flux = forcing + feedback_total - gamma * (dT - dT_deep)
    T_s(yr) = T_s(yr-1) + (dt / C_ocean) * net_flux
    T_deep(yr) = T_deep(yr-1) + (dt / C_deep) * gamma * (dT - dT_deep)
  end do
  write(*,'(A,F8.3,A)') '    TCR (warming at doubling):    ', T_s(70), ' K'
  write(*,'(A,F8.3)')    '    TCR/ECS ratio:                ', T_s(70) / ECS_predicted

end program feedback_model

Interactive Simulation: Climate Sensitivity Analysis

Python

Calculate equilibrium climate sensitivity for different feedback combinations. Shows water vapor (+), lapse rate (-), cloud (+/-), and ice-albedo (+) feedbacks. Plots temperature change vs CO2 for various feedback strengths.

climate_sensitivity.py102 lines

Click Run to execute the Python code

Code will be executed with Python 3 on the server

← 6.2 Greenhouse Effect6.4 Climate Models →