Radiative forcing is a change in the net flow of electromagnetic radiation into Earth’s climate system caused by a specified disturbance, such as an increase in atmospheric carbon dioxide, a volcanic eruption, or a change in solar output. It is usually expressed as a global annual average in watts per square metre, . Positive forcing tends to warm the surface and lower atmosphere; negative forcing tends to cool them. The term “forcing” denotes a perturbation to an energy balance, not a mechanical force.
The concept allows physically different drivers of climate change to be compared on a common energetic basis. Its precise meaning depends on which atmospheric and surface adjustments are included before the flux change is calculated. Modern assessments distinguish instantaneous radiative forcing, stratospherically adjusted radiative forcing, and effective radiative forcing. These quantities describe the initiating disturbance rather than the full temperature response, which also depends on climate feedbacks, ocean heat uptake, and the geographical pattern of the forcing.
Physical basis in Earth’s radiation budget
The Earth’s energy budget is governed primarily by absorbed sunlight and outgoing thermal radiation. Averaged over the planet, incoming solar radiation is approximately . Roughly 30% is reflected by clouds, atmospheric particles, and the surface, leaving about to be absorbed. A climate in long-term energetic equilibrium emits the same amount of energy to space as thermal infrared radiation. These rounded values describe the background flows; radiative forcing measures a change in their balance, often only a few watts per square metre.
At the top of the atmosphere, the net downward radiative flux can be written as
where is total solar irradiance at Earth’s orbital distance, is planetary albedo, and OLR is outgoing longwave radiation. The factor of four arises because Earth intercepts sunlight over a disk but distributes the global average over its spherical surface. With downward flux taken as positive, a reduction in reflected sunlight or outgoing infrared radiation produces a positive change in . Increasing reflection or infrared emission produces a negative change.
The physical mechanisms depend strongly on wavelength and altitude. Most solar energy arrives at visible and near-infrared wavelengths, whereas Earth emits predominantly in the thermal infrared. Greenhouse gases absorb and emit within particular spectral bands, altering the levels from which radiation escapes to space. Clouds both reflect sunlight and interact with thermal infrared radiation. Airborne particles can scatter sunlight, absorb it, and change cloud properties. A single disturbance can therefore produce competing shortwave and longwave contributions whose net effect cannot be inferred from either component alone.
Radiative forcing must be distinguished from the energy imbalance actually present after the climate has begun responding. An imposed positive forcing initially increases retained energy, but subsequent warming increases thermal emission and activates other feedbacks. The observed imbalance is the remaining difference between incoming and outgoing energy after those responses have occurred. It is also distinct from surface radiation: an absorbing atmospheric particle can reduce sunlight reaching the ground while warming the atmosphere and producing positive forcing at the top of the atmosphere. Specifying the level and sign convention is therefore essential.
Definitions and the treatment of rapid adjustments
Instantaneous radiative forcing is the radiative flux change immediately after a forcing agent is altered, with temperatures, clouds, water vapour, and other relevant properties held fixed. It is a useful diagnostic of the direct radiative interaction. For example, increasing carbon dioxide while retaining an unchanged atmospheric profile immediately changes infrared absorption and emission. This quantity does not yet include the atmospheric temperature changes that follow from the altered heating rates.
Stratospherically adjusted radiative forcing, conventionally called radiative forcing in many earlier assessments, allows temperatures in the stratosphere to reach radiative equilibrium while holding the surface and tropospheric state fixed. This adjustment matters because greenhouse gases do not affect all atmospheric layers identically. Increased carbon dioxide generally cools the stratosphere, changing its infrared emission and thus the net flux transmitted to the rest of the climate system. Historically, the adjusted flux was evaluated at the tropopause; after stratospheric radiative adjustment, the corresponding flux perturbation can also be diagnosed at the top of the atmosphere.
Effective radiative forcing, or ERF, additionally includes adjustments in atmospheric temperature, water vapour, clouds, and aspects of the land surface that are not mediated by global surface warming. These are called rapid adjustments. “Rapid” is a physical distinction rather than a rigid time cutoff: the important question is whether a change belongs to the response to the forcing agent itself or to the feedback response associated with changing global surface temperature. An aerosol-induced cloud change, for example, can form part of ERF even though clouds also participate in temperature-dependent feedbacks.
Climate models commonly estimate ERF by comparing simulations with identical prescribed sea-surface temperatures and sea ice but different forcing-agent concentrations or emissions. The atmosphere and land are allowed to adjust. Because land temperatures can change even when ocean temperatures are fixed, some protocols apply corrections for the resulting surface-temperature response. Another approach uses coupled-model experiments and extrapolates the relationship between top-of-atmosphere imbalance and global temperature change back to zero warming. These methods need not give identical results when feedbacks vary with time or warming pattern.
ERF has become a preferred comparison metric because it generally predicts global temperature responses more consistently across different forcing agents than instantaneous forcing does. Nevertheless, it is not an agent-independent conversion into temperature. Two disturbances with equal global ERF can differ in geographical distribution, vertical heating, precipitation effects, and evolving feedbacks. A forcing estimate is therefore incomplete unless its definition, reference state, averaging period, and included adjustments are identified.
Greenhouse gases and their spectral effects
Greenhouse-gas forcing follows from the interaction between molecular absorption spectra and the atmosphere’s vertical temperature structure. In much of the atmosphere, temperature decreases upward through the troposphere. Increasing infrared opacity shifts the effective emission to space toward higher, colder levels, initially reducing outgoing radiation in the affected spectral regions. The surface and lower atmosphere then tend to warm until the enhanced emission associated with the warmer climate helps restore balance. The actual calculation requires radiative transfer across many wavelengths, accounting for overlapping gas absorption, clouds, pressure, and temperature.
For carbon dioxide, a widely used approximation for stratospherically adjusted forcing is
where and are the perturbed and reference concentrations. This expression gives approximately for a doubling. It is an approximation for a specified forcing definition and concentration range, not a universal identity. The Intergovernmental Panel on Climate Change’s Sixth Assessment Report, published in 2021, assessed the ERF from doubled carbon dioxide as . The difference reflects updated calculations and the inclusion of rapid adjustments, rather than a contradiction between the two quantities.
The approximately logarithmic concentration dependence arises because additional absorption progressively affects spectral regions beyond the strongest, already optically thick portions of carbon dioxide bands. Saturation near a band centre therefore does not make further increases ineffective. Absorption in band wings, changes in emission altitude, and conditions in the thinner upper atmosphere remain important. Calculations depend on laboratory and atmospheric spectroscopy, including the effects of pressure broadening and temperature on molecular absorption lines.
Methane, nitrous oxide, ozone, and halogenated gases have different absorption spectra and chemical behaviour. Their forcings cannot all be represented by the carbon dioxide logarithmic formula. Methane and nitrous oxide have overlapping absorption features, so their combined forcing is not exactly the sum of two independently calculated contributions. Some halogenated gases absorb strongly in spectral regions where the atmosphere is relatively transparent, producing substantial forcing per molecule even at low concentrations. Methane also influences ozone and stratospheric water vapour through atmospheric chemistry.
Water vapour requires a distinction between a forcing agent and a feedback. Most tropospheric water vapour is controlled by temperature and the water cycle; its increase during warming is ordinarily treated as a feedback rather than an independent forcing. By contrast, stratospheric water-vapour changes caused by methane oxidation can be assigned a forcing contribution. The same radiatively active substance can thus occupy different roles depending on the process causing its concentration to change.
Aerosols, clouds, and surface modification
Atmospheric aerosols are suspended solid or liquid particles with diverse compositions, sizes, and optical properties. Sulfate particles generally scatter sunlight and produce negative forcing, while black carbon strongly absorbs sunlight and can produce positive forcing. Mineral dust both scatters and absorbs, with its net effect depending on particle size, mineral composition, altitude, and the underlying surface. Unlike well-mixed greenhouse gases, many tropospheric aerosols remain airborne for only days to weeks. Their burdens and forcing patterns are consequently concentrated around emission regions and transport pathways.
Aerosol–radiation interactions include direct scattering and absorption together with associated atmospheric adjustments. Absorbing particles heat the layer in which they occur, potentially changing stability, circulation, humidity, and clouds. Black carbon above a bright cloud deck can absorb radiation that would otherwise escape to space, producing strong positive forcing. The same aerosol over a dark surface can have a different net effect. A reduction in solar radiation at the ground is therefore not sufficient evidence that an aerosol cools the planet as a whole.
Aerosol–cloud interactions provide another major pathway. Some particles act as cloud condensation nuclei, allowing water droplets to form. With cloud liquid water held fixed, increasing droplet number generally produces smaller droplets and a more reflective cloud. This response is often called the Twomey effect, after Sean Twomey’s work on cloud reflectivity. Subsequent changes in evaporation, precipitation, entrainment, cloud water, and cloud coverage are less uniform. Smaller droplets do not inevitably create longer-lived clouds, because competing processes depend on cloud type and meteorological conditions.
Surface modification also changes radiative balance. Converting dark forest to brighter cropland or pasture can increase reflected sunlight, especially where snow becomes more exposed. Conversely, black-carbon deposition on snow and ice reduces reflectivity and produces positive forcing. Land-use change additionally affects carbon storage, aerodynamic roughness, and evapotranspiration. These effects should not be conflated: a local surface-temperature change caused by altered evaporation is not necessarily equivalent to a top-of-atmosphere radiative forcing.
In its 2021 assessment, the IPCC estimated total anthropogenic aerosol ERF for 2019 relative to 1750 at about , with an assessed range of approximately to . Most of the assessed negative contribution came from aerosol–cloud interactions. The broad range reflects difficulty in reconstructing preindustrial particle populations and determining cloud responses, not uncertainty about whether particles interact with radiation. Aerosols have partially offset greenhouse-gas forcing globally, but their short lifetimes and uneven distribution make them physically different from persistent carbon dioxide accumulation.
Solar, volcanic, and orbital influences
Changes in the Sun alter the energy entering Earth’s climate system. To first order, a small change in total solar irradiance produces a global forcing of
before allowing for atmospheric adjustments. The roughly 11-year solar cycle changes total irradiance by about 0.1% from minimum to maximum, corresponding to an order of in globally averaged absorbed sunlight. Variations at individual wavelengths, especially ultraviolet wavelengths, can be proportionally larger and affect stratospheric ozone and heating differently from a uniform change across the solar spectrum.
Large explosive volcanic eruptions can inject sulfur dioxide into the stratosphere, where chemical reactions form sulfuric-acid aerosol droplets. These particles reflect sunlight and usually produce negative global forcing lasting a few years. The climatic effect depends more directly on the amount of sulfur reaching the stratosphere and its subsequent distribution than on the eruption’s total ash volume. Most volcanic ash settles relatively quickly, whereas stratospheric sulfate can spread widely because removal by rainfall is absent there.
The 1991 eruption of Mount Pinatubo in the Philippines provided a well-observed example. Its aerosol cloud produced a global forcing of several negative watts per square metre at its strongest and contributed to global surface cooling of roughly half a degree Celsius over the following one to two years. The eruption also warmed the lower stratosphere through aerosol absorption. This combination of surface cooling and stratospheric heating illustrates why a single global forcing value cannot describe every atmospheric response. Volcanic eruptions rich in water vapour, or with unusual injection heights and compositions, can have additional effects that require separate assessment.
Changes in Earth’s orbit and axial orientation redistribute incoming sunlight over latitude and season. These variations, commonly called Milankovitch cycles, operate on timescales of tens to hundreds of thousands of years. They are central to explanations of glacial–interglacial cycles because summer sunlight at high northern latitudes affects whether winter snow survives and ice sheets grow. Their global annual-mean forcing can be small compared with their regional seasonal effects. A global-mean scalar therefore captures orbital influences less directly than it captures a broadly distributed greenhouse-gas perturbation.
The classification of an influence as forcing also depends on the system boundary and timescale. Ice-sheet albedo changes may be feedbacks in a long-term simulation of glacial cycles but prescribed forcings in an experiment studying the atmosphere’s response to an imposed ice sheet. Similarly, solar variability is external to the terrestrial climate system, whereas greenhouse-gas changes may be externally prescribed in one experiment and generated by carbon-cycle feedbacks in another. “External” is thus relative to the processes included in the analysis.
Development of the concept
The foundations of radiative-forcing analysis developed before the modern terminology. In the 1820s, Joseph Fourier examined how Earth’s temperature depends on exchanges of heat with its surroundings and on atmospheric effects. In 1856, Eunice Newton Foote reported experiments comparing the heating of gases, including carbon dioxide, and connected atmospheric composition with climate. In 1859, John Tyndall began laboratory investigations that established the selective absorption of infrared radiation by gases such as water vapour and carbon dioxide. These studies helped show that the atmosphere’s climatic role depends on composition, not simply on its mass.
In 1896, Svante Arrhenius published calculations connecting changes in atmospheric carbon dioxide with surface temperature. His treatment differed substantially from modern models, particularly in its representation of atmospheric absorption and feedbacks, but it made quantitative greenhouse-climate calculations a sustained scientific problem. Guy Stewart Callendar’s 1938 work brought together evidence for rising carbon dioxide and temperature change. During the mid-twentieth century, improved infrared measurements and calculations by researchers including Gilbert Plass strengthened the physical basis for estimating changes in radiative flux.
The emergence of numerical atmospheric models made it possible to separate a radiative perturbation from the resulting climate response more systematically. Syukuro Manabe and Richard Wetherald’s 1967 radiative–convective calculations included water-vapour feedback and demonstrated how changing carbon dioxide could alter an atmospheric temperature profile. Such work required explicit treatment of infrared transfer, solar heating, convection, and humidity. Later three-dimensional models added atmospheric circulation, ocean interactions, spatially varying clouds, and other processes that determine how a globally imposed perturbation becomes a geographically structured climate response.
Radiative forcing became increasingly standardized through the assessments of the Intergovernmental Panel on Climate Change, beginning with its First Assessment Report in 1990. Common baseline years and consistent units made it possible to compare contributions from greenhouse gases, aerosols, solar changes, and surface albedo. The conventional stratospherically adjusted definition provided a useful predictor of global temperature response for many agents. It also exposed cases in which equivalent conventional forcings produced different responses because of atmospheric and cloud adjustments.
Research in the 2000s and 2010s increasingly emphasized effective radiative forcing and the distinction between adjustments and feedbacks. The IPCC’s Fifth Assessment Report in 2013 used ERF prominently, and the Sixth Assessment Report in 2021 further developed its quantitative assessment. This evolution did not replace the underlying energy-budget principle. It refined which changes belong in the initial perturbation so that comparisons among carbon dioxide, aerosols, ozone, and other agents better reflect their influence on the coupled climate system.
Calculation, observation, and uncertainty
The most detailed calculations begin with wavelength-resolved radiative transfer. Line-by-line models use spectroscopic databases to calculate absorption and emission across large numbers of molecular transitions. They account for line strengths, pressure broadening, temperature dependence, spectral overlap, and atmospheric profiles. Because such calculations are computationally expensive, weather and climate models usually employ faster band or correlated- methods tested against detailed calculations. Cloud geometry, aerosol optical properties, and surface reflectance must also be specified; molecular spectroscopy alone does not determine the forcing of the whole atmosphere.
For greenhouse gases, controlled calculations can compare radiation with and without a concentration change while holding the background atmosphere fixed. Aerosol forcing requires additional information about particle size, composition, mixing state, vertical distribution, and cloud interactions. ERF calculations usually need ensembles or long simulations because natural variations in clouds and circulation can obscure a small forced change in the global energy budget. Model differences may arise from both radiative-transfer approximations and physically different rapid adjustments.
Satellite instruments observe reflected solar radiation and emitted thermal radiation at the top of the atmosphere. The Clouds and the Earth’s Radiant Energy System, or CERES, has provided major measurements of these fluxes since the late 1990s. Spectrally resolved infrared instruments offer complementary evidence by identifying changes in the wavelength regions associated with particular gases. Ground-based measurements also detect changes in downward infrared radiation. Such observations test the mechanisms behind forcing, although no single instrument directly measures every component of historical anthropogenic ERF.
A central difficulty is the counterfactual baseline. Satellites observe the atmosphere that exists, not an otherwise identical atmosphere without industrial-era emissions. Clouds, water vapour, surface temperature, and circulation change simultaneously with forcing agents. Separating their effects requires radiative calculations, statistical analysis, model experiments, and independent constraints on atmospheric composition. The preindustrial aerosol state is particularly difficult to reconstruct because systematic measurements were sparse and natural particle sources are variable.
Measurement uncertainty also differs from uncertainty in the attribution of a flux change. Instruments require stable calibration, and small long-term changes must be extracted from much larger background flows. Ocean heat-content measurements constrain Earth’s energy accumulation and help anchor estimates of the absolute imbalance. They do not directly identify which forcing agent supplied the energy. Assessment therefore combines observations and process understanding, preserving distinctions among instrumental error, uncertain historical emissions, uncertain adjustments, and internal climate variability. Uncertainties among individual forcing components can also be correlated, so their ranges should not automatically be added as independent errors.
From forcing to temperature and precipitation
A simple global energy-balance relation connects effective forcing to warming:
where is the downward energy imbalance, is ERF, is global surface-temperature change, and is a positive radiative restoring coefficient in this sign convention. The equation expresses that warming increases the net loss of energy to space and thereby opposes the imposed forcing. At equilibrium, , giving . This is a useful approximation, not an exact law with a universally constant .
The restoring coefficient includes several feedback processes. Increased thermal emission from a warmer planet provides the basic stabilizing Planck response, related to blackbody radiation. Water-vapour increases generally amplify warming, while the loss of reflective snow and ice produces a positive albedo feedback. Changes in the atmospheric lapse rate and in clouds alter the response further. Their combined effect determines climate sensitivity, conventionally expressed as the eventual global warming from doubled carbon dioxide. The IPCC’s 2021 assessment gave a best estimate of , with a likely range of –.
The ocean delays equilibration because of its large heat capacity and the transport of heat below the surface. Consequently, forcing and global temperature need not rise in fixed proportion over short periods. In a simple transient model, an effective heat capacity multiplies the rate of temperature change; more elaborate models distinguish a surface layer from the deep ocean. Changes in circulation and the geographical pattern of ocean warming can alter the effective feedback strength over decades to centuries. The response to a sustained forcing therefore differs from the response to a short volcanic pulse.
Equal global forcing does not guarantee equal regional warming. The response depends on where energy is added or removed, atmospheric circulation, surface properties, and coupling to the ocean. Forcing efficacy describes the temperature response per unit forcing relative to a reference agent, usually carbon dioxide. ERF reduces some differences in efficacy by incorporating rapid adjustments, but it does not remove all dependence on spatial pattern. This matters particularly for geographically concentrated aerosols, snow-darkening agents, and disturbances acting at different atmospheric heights.
Precipitation introduces a further energetic constraint. Globally averaged rainfall is linked to atmospheric radiative cooling and surface sensible-heat exchange, rather than only to the moisture-holding capacity of warmer air. Increased carbon dioxide can rapidly reduce atmospheric radiative cooling and suppress precipitation before much surface warming occurs; subsequent warming generally increases precipitation. Solar forcing and absorbing aerosols produce different atmospheric heating profiles and therefore different rapid rainfall adjustments. The frequently cited increase of atmospheric water-vapour capacity by roughly 7% per degree does not imply that global mean precipitation increases at the same rate.
Historical forcing budgets and attribution
A historical forcing budget compares the climate system at two specified times using consistent definitions. The IPCC Sixth Assessment Report estimated total anthropogenic ERF in 2019 relative to 1750 at , with an assessed range of –. This positive total represents greenhouse-gas and other warming influences partly offset by aerosol and surface-albedo cooling. It is a difference from a reference state, not the total greenhouse effect of the atmosphere and not the observed planetary energy imbalance in 2019.
In that assessment, concentration changes in carbon dioxide contributed approximately , methane , nitrous oxide , and halogenated gases collectively . Ozone changes supplied an additional positive contribution. Aerosols supplied the largest negative contribution and much of the uncertainty in the total. The assessed solar contribution over the same 1750–2019 comparison was small relative to the anthropogenic total. All such values are tied to the stated dates and assessment methods; they are not timeless constants.
Budgets can be organized either by changes in atmospheric constituents or by the emissions that caused them. In a concentration-based account, methane forcing describes the radiative effect of the methane abundance change itself. In an emissions-based account, methane emissions can also be assigned their effects on ozone, stratospheric water vapour, and other chemically mediated changes. Similarly, nitrogen-oxide emissions affect several constituents with opposing radiative effects. Mixing the two accounting frameworks without adjustment creates double counting or omissions. The relevant question is whether the budget describes what changed in the atmosphere or which emissions caused those changes.
Radiative forcing supports attribution of observed climate change but does not by itself complete it. Attribution studies combine forcing histories with modelled responses, observations, and estimates of internal variability. Spatial and vertical temperature patterns provide information beyond the global mean: greenhouse-gas increases, solar changes, and volcanic aerosols do not produce identical atmospheric signatures. Natural variations can redistribute heat and alter temperatures over years or decades without representing a sustained external forcing. Separating these processes requires both energetic and dynamical evidence.
For comparing emissions over time, forcing is also the basis of global warming potential, which integrates the forcing following an emission pulse over a chosen time horizon and compares it with an equal mass of carbon dioxide. This metric depends on atmospheric lifetimes, radiative efficiencies, chemical effects, and the chosen horizon, commonly 20 or 100 years. It is not identical to the temperature at the end of that period. Carbon dioxide also lacks a single atmospheric lifetime: uptake by land and ocean removes different fractions on different timescales, and part of an emitted pulse persists for millennia. A forcing comparison must therefore distinguish a sustained concentration change, a sustained emission rate, and a one-time emission pulse.