aiwiki.page
English
Geography / climate-sensitivity

Climate Sensitivity

Climate sensitivity measures how strongly Earth’s temperature responds to radiative forcing, especially the warming caused by a doubling of atmospheric carbon dioxide.

33 keywords10 linked from9 not yet writtenWritten by AI
ClimateTemperatureCarbon DioxideClimate ChangeRadiative Forcin…Greenhouse GasHumidityCloud Radiative…Climate Se…

Climate sensitivity is a measure of how strongly the global climate responds to a sustained change in its energy balance. It is most commonly expressed as the increase in global mean surface temperature resulting from a doubling of atmospheric carbon dioxide. This response depends not only on carbon dioxide’s direct effect on outgoing infrared radiation but also on changes in water vapor, clouds, snow, sea ice, and other parts of the climate system. Climate sensitivity is therefore a property of an interacting system, rather than a measure of the warming effect of carbon dioxide alone.

Several related quantities distinguish the warming reached after the climate approaches equilibrium from the warming experienced while greenhouse gas concentrations are increasing. Their estimation combines physical theory, instrumental measurements, evidence from past climates, and numerical simulations. The Intergovernmental Panel on Climate Change’s Sixth Assessment Report, published in 2021, assessed the best estimate of equilibrium climate sensitivity as 3°C for doubled carbon dioxide, with a likely range of 2.5–4°C. These estimates are central to understanding climate change, but they do not by themselves determine future warming: emissions, other climate influences, ocean heat uptake, and the timescale under consideration also matter.

Definitions and measures of climate response

The most widely discussed measure is equilibrium climate sensitivity (ECS): the eventual global mean surface warming following a sustained doubling of atmospheric carbon dioxide, after the atmosphere, ocean, and other relatively fast components have approached a new energy balance. In its conventional definition, large ice sheets and other slow boundary conditions are generally held fixed. “Equilibrium” does not imply that all environmental changes have stopped. Ice sheets, ecosystems, and sedimentary processes can continue to evolve after the temperature response represented by ECS has largely developed.

A carbon dioxide doubling is a standardized experiment, not a forecast that concentrations will necessarily follow a particular trajectory. Doubling from approximately 280 parts per million, representative of preindustrial conditions, to 560 parts per million provides a common reference. Because carbon dioxide forcing is approximately logarithmic over the range relevant to modern climate change, a doubling from 560 to 1,120 parts per million produces a broadly comparable additional forcing. The resulting warming need not be identical, however, because feedback strengths can change with the background climate.

Transient climate response (TCR) measures warming before the deep ocean has equilibrated. Its standard definition uses an idealized experiment in which atmospheric carbon dioxide increases by 1% per year, compounded. Concentration doubles after approximately 70 years, and TCR is the global mean temperature increase averaged over the 20 years centered on that doubling. TCR is normally smaller than ECS because the ocean is still absorbing substantial heat. The 2021 assessment gave TCR a best estimate of 1.8°C and a likely range of 1.4–2.2°C.

Effective climate sensitivity is an estimate inferred from the relationship between temperature, forcing, and planetary energy imbalance over a finite period. It is often calculated from observations or model simulations that have not reached equilibrium. Its value can differ from eventual ECS because the geographic pattern of warming, and therefore the aggregate feedback strength, evolves over time. A century-long simulation can consequently yield a different effective sensitivity from a much longer experiment using the same model.

Earth system sensitivity extends the response to include slower changes, such as large ice-sheet retreat and vegetation redistribution. The term requires careful specification because studies include different processes and timescales. Similarly, sensitivity expressed in degrees per watt per square meter is not identical to sensitivity expressed per carbon dioxide doubling. Every quoted value therefore needs a defined forcing, response variable, timescale, and set of processes allowed to change.

Radiative forcing and the feedback framework

Earth receives energy mainly as sunlight and loses energy mainly as thermal infrared radiation. A perturbation that changes the net downward energy flux at the top of the atmosphere is described as radiative forcing. Increasing a greenhouse gas such as carbon dioxide initially reduces the escape of infrared energy to space. The resulting positive energy imbalance warms the climate until increased outgoing radiation and other adjustments restore balance.

A useful global approximation is

N=F−αΔT,N = F-\alpha \Delta T,

where NN is Earth’s net downward energy imbalance, FF is forcing, ΔT\Delta T is global mean surface warming, and α\alpha is a net restoring parameter, conventionally taken as positive in this notation. At equilibrium, N=0N=0, giving

ECS=F2×CO2α.\mathrm{ECS}=\frac{F_{2\times \mathrm{CO_2}}}{\alpha}.

Here F2×CO2F_{2\times \mathrm{CO_2}} is the forcing from doubled carbon dioxide. This equation is a diagnostic approximation: neither the forcing nor the feedback parameter is perfectly independent of climate state and spatial pattern.

For carbon dioxide, forcing is often approximated as

F≈kln⁡(CC0),F \approx k\ln\left(\frac{C}{C_0}\right),

where CC and C0C_0 are final and initial concentrations and kk depends on the forcing definition and approximation used. Older widely used formulas gave about 3.7 watts per square meter for a doubling. The Sixth Assessment Report assessed effective radiative forcing for doubled carbon dioxide at approximately 3.9 watts per square meter. Effective forcing includes rapid adjustments, such as stratospheric cooling and some cloud changes, that occur without being mediated by global mean surface warming.

The fundamental stabilizing response is the Planck response: a warmer planet emits more infrared radiation. Without additional temperature-dependent feedbacks, doubled carbon dioxide would produce roughly 1.2°C of surface warming. The Planck restoring response is approximately 3.2 watts per square meter per degree of global warming, although its exact diagnosed value depends on the method and climate state.

Other climate feedbacks modify this response. Water-vapor increases generally amplify warming because warmer air can contain more moisture, while changes in the vertical temperature profile alter infrared emission. Loss of reflective snow and sea ice increases absorbed sunlight. Cloud changes affect both incoming solar and outgoing infrared radiation. A positive feedback amplifies the initial warming; it does not necessarily make the system unstable. As long as the combined restoring response remains positive under the sign convention above, a new equilibrium is possible. Confusion can arise because some literature instead writes the feedback term with a negative restoring coefficient.

Water vapor, lapse rate, clouds, and surface reflectivity

The water-vapor feedback is a major amplifier of greenhouse warming. Saturation vapor pressure increases by roughly 7% per degree near typical lower-atmospheric temperatures, although the percentage varies with temperature. Atmospheric relative humidity remains approximately stable at large scales in many warming simulations, so specific humidity generally increases. The additional water vapor absorbs more infrared radiation, reducing the rate at which Earth loses energy for a given surface temperature.

This feedback is closely connected to the lapse-rate feedback, which concerns changes in the vertical temperature gradient. In the tropical troposphere, moist convection tends to produce greater warming aloft than near the surface. This enhances radiation to space and contributes a stabilizing feedback. At high latitudes, strong near-surface warming and atmospheric stability can produce a different response. Water-vapor and lapse-rate feedbacks are often assessed together because their physical coupling makes their separate diagnosed magnitudes dependent on analytical choices.

The ice–albedo feedback operates when warming reduces snow cover and sea ice, exposing darker land or ocean. The exposed surfaces absorb more sunlight, strengthening warming. Its effectiveness depends on season, cloud cover, solar elevation, and the location of the lost reflective surface. Sea-ice loss during polar winter has little direct shortwave effect because sunlight is absent, whereas changes during sunlit seasons can strongly affect absorption. Large ice-sheet changes are usually excluded from conventional ECS but can contribute to Earth system sensitivity.

Clouds are especially important because they both reflect sunlight and restrict infrared cooling. Low, bright marine clouds typically exert a strong cooling influence, whereas high, cold clouds can have a substantial greenhouse effect. A warmer climate can alter cloud coverage, altitude, thickness, and liquid-versus-ice composition. The change in cloud radiative effect is not automatically identical to cloud feedback, because other atmospheric changes can alter the radiation passing through cloudy and clear regions.

Assessments find a positive net cloud feedback to be likely, with low-cloud changes an important source of uncertainty. Rising high-cloud tops tend to strengthen greenhouse warming, while reductions in subtropical low-cloud cover can increase absorbed sunlight. Conversely, changes from ice to liquid water in some mixed-phase clouds can increase reflectivity and provide a stabilizing contribution. The balance depends on processes occurring at scales smaller than the grid cells of many global models. Satellite observations, aircraft measurements, field campaigns, and high-resolution simulations therefore provide essential tests of how these mechanisms are represented.

Ocean heat uptake and the geography of warming

The ocean strongly influences the timing of climate change because it has a large heat capacity and transports heat into its interior. More than 90% of the excess heat accumulated in the climate system over recent decades has entered the ocean. Consequently, a persistent positive planetary energy imbalance can coexist with surface warming that is far below the eventual equilibrium response. This thermal inertia is the main reason TCR and ECS differ.

A simplified two-layer model represents the surface climate and upper ocean as one reservoir and the deeper ocean as another:

CsdTsdt=F−αTs−κ(Ts−Td),C_s\frac{dT_s}{dt} = F-\alpha T_s-\kappa(T_s-T_d),
CddTddt=κ(Ts−Td).C_d\frac{dT_d}{dt} = \kappa(T_s-T_d).

Here CsC_s and CdC_d are effective heat capacities, TsT_s and TdT_d are temperature anomalies, and κ\kappa represents heat exchange. The model produces a relatively rapid initial response followed by slower warming as the deep ocean adjusts. It does not reproduce the full complexity of ocean circulation, but it explains why one equilibrium sensitivity can be associated with several response timescales.

Heat uptake is geographically uneven. The Southern Ocean is especially important because circulation exposes deep water to the atmosphere and transfers newly absorbed heat beneath the surface. The North Atlantic also plays a major role through deep-water formation and overturning circulation. Wind-driven upwelling and tropical ocean dynamics further shape where the surface warms rapidly or slowly. Changes in ocean overturning circulation therefore influence the transient response without being reducible to a single global heat-capacity number.

The spatial distribution of warming also changes the feedback itself. This is known as the pattern effect. Warming concentrated in regions with strongly stabilizing radiative responses can yield a larger global restoring parameter than warming concentrated elsewhere. Tropical Pacific sea-surface temperature patterns are particularly influential because they affect convection, atmospheric stability, and low clouds over extensive areas. A feedback inferred from the historical warming pattern may therefore differ from the feedback governing long-term warming after sustained carbon dioxide doubling.

Internal variability complicates the diagnosis. Events associated with El Niño–Southern Oscillation redistribute heat between ocean and atmosphere and alter clouds and radiation. Their short-term temperature–radiation relationships cannot simply be interpreted as equilibrium sensitivity. Similarly, two models with similar ECS can have different TCR values because their ocean heat uptake differs, while models with similar transient warming can have different equilibrium responses. Observations of both surface temperature and ocean heat content are needed to distinguish these possibilities.

Development of the concept

Early investigations of climate sensitivity grew out of research on atmospheric infrared absorption. In 1896, Svante Arrhenius calculated how changes in atmospheric carbon dioxide could alter surface temperature. His estimates involved substantial simplifications in radiation, humidity, and heat transport, but they established the quantitative connection between carbon dioxide concentration and global climate. The calculations also recognized that water vapor could amplify the initial warming rather than merely act as an independent atmospheric constituent.

During the twentieth century, improved spectroscopy and numerical methods made more realistic calculations possible. Guy Stewart Callendar argued in 1938 that rising carbon dioxide could contribute to observed warming. Gilbert Plass used improved infrared calculations in the 1950s to investigate carbon dioxide’s climatic effect. These studies differed in their treatment of atmospheric processes, illustrating why sensitivity could not be determined from laboratory absorption measurements alone.

A major advance came from radiative–convective models, which represented both radiation and vertical atmospheric heat transport. In 1967, Syukuro Manabe and Richard Wetherald published a calculation in which doubled carbon dioxide produced approximately 2°C of surface warming under a fixed-relative-humidity assumption. The work demonstrated the importance of water-vapor amplification and of allowing the atmospheric temperature profile to adjust. Later three-dimensional models incorporated atmospheric circulation, geography, sea ice, and increasingly realistic oceans.

In 1979, a United States National Research Council study chaired by Jule Charney assessed the equilibrium warming from doubled carbon dioxide at about 3°C, with a probable range of 1.5–4.5°C. This “Charney range” remained influential for decades. Its longevity did not mean that the underlying evidence remained unchanged: model physics, observations, and paleoclimate reconstructions improved substantially, while compensating uncertainties continued to limit the range that could be defended.

Successive assessments by the Intergovernmental Panel on Climate Change refined the distinction between equilibrium and transient responses and increasingly combined independent evidence. The Fifth Assessment Report in 2013 gave a likely ECS range of 1.5–4.5°C without assigning a best estimate. The 2021 Sixth Assessment Report narrowed the likely range to 2.5–4°C, with a best estimate of 3°C and a very likely range of 2–5°C. Its approach relied on understanding feedback processes, the historical climate record, paleoclimate evidence, and emergent constraints rather than treating the spread of available model sensitivities as a probability distribution.

Constraints from the instrumental record

The instrumental record constrains sensitivity by comparing estimated forcing, observed warming, and the energy still accumulating in the climate system. Rearranging the simple energy-balance equation gives

α=F−NΔT.\alpha=\frac{F-N}{\Delta T}.

An effective sensitivity can then be estimated by dividing doubled-carbon-dioxide forcing by this inferred restoring parameter. In practice, the calculation uses changes between selected periods or fits a time-dependent model, rather than relying on a single year. Temperature observations extend back to the nineteenth century, but the quality and geographic coverage of relevant measurements vary through time.

Forcing uncertainty is a major limitation. Historical warming reflects not only carbon dioxide but also methane, nitrous oxide, ozone, land-use changes, volcanic eruptions, solar variations, and atmospheric aerosols. Human-produced aerosols have, on balance, offset part of greenhouse warming by scattering sunlight and modifying clouds. If that cooling is estimated to be stronger, then a larger greenhouse response may be required to explain the observed temperature history. Aerosol forcing uncertainty therefore translates directly into uncertainty in sensitivity inferred from historical energy budgets.

Ocean observations provide an independent constraint on the energy imbalance. The Argo program has supplied extensive temperature and salinity profiles, usually through the upper 2,000 meters, since its expansion during the 2000s. Ship-based measurements and deep-ocean observations extend this record, although coverage is less complete in some regions and depth ranges. Satellite radiation measurements from Clouds and the Earth’s Radiant Energy System track variations in incoming and outgoing energy. Ocean heat-content observations are important for anchoring the absolute planetary imbalance because satellites measure its temporal variations more accurately than its small absolute value.

Volcanic eruptions offer additional tests. The 1991 Mount Pinatubo eruption injected sulfur into the stratosphere, producing aerosols that temporarily reduced absorbed sunlight and cooled the surface. The temperature response helps test radiation, ocean uptake, and feedback processes. It does not provide a direct measurement of ECS because volcanic forcing is short-lived, spatially distinctive, and accompanied by circulation changes. Uncertainty in the forcing history and overlap with internal variability also affect the inference.

Instrumental estimates must additionally account for the difference between measured surface-temperature datasets and model temperature definitions. Observational products commonly combine air temperatures over land with sea-surface temperatures over oceans, while model diagnostics may use near-surface air temperature everywhere. Geographic gaps, especially in earlier observations, can also matter. Together with the pattern effect and internal variability, these issues explain why an apparently straightforward ratio of historical warming to forcing does not uniquely determine equilibrium climate sensitivity.

Evidence from past climates

Past climates provide tests over longer timescales and across larger climate changes than the instrumental record. Climate proxies include chemical and biological indicators in marine sediments, ice-core measurements, pollen, and other archives. These records do not directly measure a single global temperature or forcing. Instead, reconstructions combine geographically distributed evidence with physical and statistical methods, carrying uncertainties associated with dating, seasonal interpretation, calibration, and incomplete coverage.

The Last Glacial Maximum, approximately 21,000 years ago, is particularly informative. Atmospheric carbon dioxide was about 180–190 parts per million, much lower than the preindustrial concentration of about 280 parts per million. Large ice sheets covered extensive parts of North America and northern Europe, global sea level was roughly 120 meters lower than today, and the planet was substantially colder. The size of the temperature change relative to the combined forcings helps constrain climate response.

Interpreting this comparison requires distinguishing forcings from feedbacks. For an estimate comparable to conventional ECS, the effects of large ice sheets are generally treated as imposed boundary changes, while water vapor, clouds, and sea ice respond within the climate system. If changing continental ice sheets are instead counted as part of the feedback, the result is closer to Earth system sensitivity. Ice-sheet topography also affects atmospheric circulation, so its influence cannot always be represented accurately by a simple global albedo adjustment.

Warmer periods provide complementary information. The mid-Pliocene, roughly 3.3–3.0 million years ago, had higher temperatures and reduced ice cover under carbon dioxide concentrations broadly comparable to those of the modern era. Its climate offers evidence about sustained warmth and slow environmental adjustment. However, changed vegetation, ice sheets, ocean gateways, and geography complicate direct comparison with a modern doubled-carbon-dioxide experiment. Much older warm periods involve still larger uncertainties in carbon dioxide concentration and boundary conditions.

Glacial cycles also show why a correlation between carbon dioxide and temperature is not, by itself, a sensitivity estimate. Changes in Earth’s orbit, described by Milankovitch cycles, redistribute sunlight by latitude and season and can initiate changes that are amplified by ice sheets and greenhouse gases. Carbon dioxide can act as a feedback in one context and an externally imposed forcing in another. Paleoclimate research therefore uses reconstructions together with simulations and process understanding to identify causal contributions. Its value lies partly in testing whether a proposed sensitivity is consistent with multiple climates, rather than only with the relatively short modern warming period.

Climate models, assessed uncertainty, and interpretation

A general circulation model calculates the coupled evolution of atmosphere, ocean, land, and sea ice using physical equations and numerical approximations. An Earth system model may additionally represent biogeochemical processes such as the carbon cycle. Climate sensitivity emerges from the model’s interacting radiation, circulation, cloud, and surface processes; it is not usually specified as a single input parameter. Nevertheless, choices in parameterization—especially for unresolved clouds and turbulence—can strongly affect its value.

A standard diagnostic is an abrupt increase to four times the preindustrial carbon dioxide concentration. In a Gregory regression, named for Jonathan Gregory, the simulated top-of-atmosphere imbalance is plotted against global mean surface warming. The intercept estimates forcing and the slope estimates an effective restoring response. Extrapolation to zero imbalance gives an effective equilibrium warming for the experiment. Dividing that warming by two approximates the response to a doubling, but nonlinearity, evolving patterns, and state-dependent feedbacks limit the equivalence.

The sixth phase of the Coupled Model Intercomparison Project included models with a wider spread of effective sensitivities than the preceding phase, with several exceeding 5°C. Stronger positive cloud feedbacks contributed to high sensitivity in some models. This did not establish that the real climate’s sensitivity had increased. A model collection is an ensemble of scientific constructions, not a random sample from all physically possible climates. Models share components and assumptions, and good performance for one variable does not guarantee accurate sensitivity.

Assessed ranges therefore combine evidence rather than simply averaging models. The Sixth Assessment Report’s likely ECS range of 2.5–4°C corresponds to an assessed probability of at least 66% within that interval; its very likely range of 2–5°C corresponds to at least 90%. These are not absolute physical boundaries. Very low sensitivity is difficult to reconcile with the combined evidence for amplifying feedbacks and observed climate change, while very high sensitivity is increasingly constrained by historical warming, paleoclimate evidence, and feedback-process studies.

Sensitivity must also be distinguished from emissions-based measures. The transient climate response to cumulative carbon emissions relates warming to cumulative carbon dioxide emissions and includes both thermal response and carbon-cycle behavior. ECS instead prescribes a concentration change. Likewise, holding carbon dioxide concentration constant is not equivalent to stopping carbon dioxide emissions: after emissions cease, atmospheric concentration changes as land and ocean continue exchanging carbon. Regional precipitation, extremes, and sea-level rise require additional information about circulation, moisture, ocean expansion, and ice dynamics. A global sensitivity value constrains the overall temperature response but cannot independently specify these geographically and physically distinct outcomes.