Equilibrium Climate Sensitivity¶
The long-term global mean surface warming after atmospheric CO2 doubles from a preindustrial baseline and the planetary energy budget rebalances, excluding ice-sheet feedbacks.
Core Idea¶
Equilibrium climate sensitivity (ECS) names an idealized response of a specified climate system: the long-term change in global mean near-surface air temperature after atmospheric carbon dioxide (CO2) is held at twice its preindustrial reference concentration and the planetary energy budget rebalances. It is reported as a temperature difference in kelvins or degrees Celsius. Under the IPCC Sixth Assessment's definition, ECS excludes feedbacks associated with changing ice sheets. The preindustrial baseline and feedback convention fix the target; model estimates can still differ. Doubling CO2 from a different background state is a state-dependent experimental sensitivity and is not automatically this conventional ECS.[1][2]
An ECS value can be estimated in different ways without changing the quantity being estimated. Rugenstein and colleagues ran coupled climate models for millennia under elevated CO2 and estimated their eventual doubled-CO2 warming. Schmittner and colleagues used Last Glacial Maximum temperature reconstructions, model experiments and forcing assumptions to infer an Earth ECS distribution. A short-run regression of energy imbalance against warming instead gives an effective ECS diagnostic that may differ from the eventual equilibrium response; warming at the doubling point of a rising-CO2 experiment is transient climate response (TCR), a different metric.[2][3][4][1]
Structural Signature¶
- Preindustrial reference and standardized perturbation. State the preindustrial baseline climate and sustained doubling of atmospheric CO2 relative to it. A sensitivity to arbitrary forcing per watt per square metre is related, but not automatically this doubled-CO2 temperature quantity.[1][3]
- Global surface-air temperature response. Compare global mean near-surface air temperature at the reference and rebalanced states. Regional changes and top-of-atmosphere fluxes help diagnose the response but are not its reported temperature endpoint.[1][2]
- Equilibrium endpoint. At the target endpoint, the imposed forcing's planetary energy imbalance has been offset by climate response. Ocean heat uptake makes earlier warming a transient state; model runs may approach rather than literally attain exact equilibrium.[1][2]
- Declared feedback boundary. State which climate responses count as feedbacks. IPCC AR6 excludes ice-sheet feedbacks from ECS even where paleoclimate evidence includes altered continental ice sheets as a boundary condition or forcing.[1]
- Qualified estimate. Specify whether a number comes from a long-run model estimate, short-run effective extrapolation or paleoclimate inference, with assumptions and uncertainty. The estimation route is evidence for ECS, not an extra defining physical step.[2][3][4]
What It Is Not¶
ECS is not TCR. TCR is global warming when CO2 reaches twice its reference concentration around year 70 in the idealized experiment where concentration rises by 1% each year; the ocean is still taking up heat. Nor is a 150-year abrupt-4xCO2 regression an observed rebalanced climate. Such a fit commonly estimates effective sensitivity, extrapolating a relation between energy imbalance and warming. Rugenstein and Zelinka each show why evolving feedbacks can make that extrapolation differ from a longer-run model response.[1][2][4]
ECS is not the raw temperature difference between the Last Glacial Maximum and the preindustrial climate. That contrast involved changes in greenhouse gases, ice sheets, dust and other conditions, and its inference to doubled-CO2 ECS depends on how those influences are assigned and normalized. If the slow response of ice sheets to CO2 is counted as part of the output instead of treated separately, the resulting Earth-system sensitivity has a different scope.[1][3]
Scope of Application¶
LongRunMIP assembled long coupled-model CO2 experiments, many lasting a millennium or more, to assess eventual responses and test shorter extrapolations. The CCSM3 abrupt2x simulation ran for 3,000 years; Rugenstein and colleagues report a best equilibrium-warming estimate of 2.57 K for that model and experiment, compared with 2.35 K from its first 150 years' linear estimate. CESM 1.0.4's 2,500-year abrupt2x run gives a different model-specific best estimate, 3.20 K. These numbers are properties estimated for specified model experiments, not direct observations of Earth's future temperature.[2]
The Last Glacial Maximum offers a different evidence setting: Earth was near equilibrium in a climate with lower CO2, larger ice sheets and higher dust loading. Schmittner and colleagues combined land and sea temperature proxies with a range of model runs to infer doubled-CO2 ECS. Their conditional analysis reported a median 2.3 K and a 66% range of 1.7–2.6 K; they explicitly note that alternate assumptions or data subsets can widen the range. The paper's interval is one study's result, not a universal fixed ECS or a replacement for the assessment of all evidence.[3][1]
Clarity¶
Every reported sensitivity should answer whether CO2 was doubled from the preindustrial baseline, which temperature was compared, whether equilibrium was reached or inferred, and which feedbacks counted. Rugenstein's CCSM3 long run addresses an eventual model response after sustained doubling. Schmittner's LGM analysis infers the doubled-CO2 target from a different real-world forcing mix and proxies; the raw LGM temperature contrast is an input to inference rather than the output value.[2][3]
Use “effective ECS” when a short abrupt-forcing run supplies a regression-based extrapolation. Zelinka and colleagues regress top-of-atmosphere energy imbalance against global temperature in 150-year abrupt-4xCO2 simulations, scaling the result to a doubling. They state that such estimates generally understate the eventual equilibrium warming because feedbacks can evolve with the warming pattern and state.[4]
Manages Complexity¶
A single long-term temperature difference summarizes many coupled processes: CO2 radiative forcing, heat storage, clouds, water vapour, lapse rate and surface albedo. The ECS definition fixes a comparison target so different model and observational lines can be discussed together. It does not erase why estimates differ: response strength can depend on background climate, spatial warming pattern, feedback evolution and treatment of non-CO2 forcing.[1][2][4]
The fixed target also disciplines paleoclimate comparisons. An LGM proxy record supplies partial evidence of a large, near-equilibrium climate contrast; model and forcing assumptions translate that evidence to a doubled-CO2 quantity. The translation should remain visible with its uncertainty rather than being hidden behind one reported number.[3][1]
Abstract Reasoning¶
Specify the preindustrial reference climate. Imagine holding atmospheric CO2 at twice its preindustrial concentration. Allow the modeled or inferred climate to adjust until the planetary energy budget is rebalanced under the declared feedback convention. Compare global mean near-surface air temperature at that endpoint with the reference. When the endpoint is not observed, identify the evidence and assumptions that support its estimate.[1][2]
The boundary tests change one role at a time. A response while CO2 is still increasing and ocean heat uptake persists is TCR under its defined ramp protocol. A regional temperature difference is not the global ECS endpoint. An LGM-to-modern cooling number has no doubled-CO2 meaning until other forcings, state changes and ice-sheet scope are addressed.[1][3]
Knowledge Transfer¶
The reusable map is preindustrial reference climate → sustained doubled CO2 → rebalanced global surface temperature → difference under a feedback convention. A long model run can approximate that endpoint directly within its model. A glacial reconstruction can constrain it indirectly by combining quasi-equilibrium temperature evidence with model and forcing assumptions. An early regression can estimate an effective proxy for the endpoint while preserving the distinction between the proxy and the target.[2][3][4]
The word “sensitivity” travels to many input-output ratios. Here the CO2 doubling, global temperature endpoint and energy-balance condition make the measure specifically climatic. Feedback is among the processes affecting its value, not a synonym for ECS itself.
Examples¶
Millennial model equilibration¶
In Rugenstein and colleagues' LongRunMIP analysis, CCSM3 is simulated under an abrupt sustained CO2 doubling for 3,000 years. The authors' best estimate of its eventual global mean surface warming is 2.57 K. The first-150-year extrapolation gives 2.35 K, illustrating that a near-term effective estimate and the longer-run equilibrium estimate can diverge within the same model.[2]
Mapped back: reference → model preindustrial control; perturbation → abrupt2x CO2; response → modeled global mean surface-air warming; endpoint → long-run estimate of restored energy balance; boundary → model-specific result under its represented feedbacks, rather than measured Earth ECS.
Glacial evidence for an Earth estimate¶
Schmittner and colleagues compare Last Glacial Maximum land and sea temperature reconstructions with model simulations spanning alternative sensitivities. Their inference links a quasi-equilibrium Earth temperature contrast to a doubled-CO2 response while representing lower greenhouse gases, larger ice sheets and dust. The resulting distribution has conditional median 2.3 K; its width changes under alternate data and assumptions.[3]
Mapped back: reference → preindustrial Earth; perturbation evidence → LGM forcing mix, translated to the CO2-doubling target; response → proxy-constrained global temperature; endpoint → inferred near-equilibrium comparison; boundary → ice-sheet and non-CO2 forcing treatment, model structure and proxy uncertainty. The raw LGM cooling is not itself ECS.[3][1]
Structural Tensions¶
The sources establish an estimate-versus-target tension. A short experiment is computationally feasible and can yield an effective ECS proxy, but feedbacks and the spatial pattern of warming can evolve over the centuries needed to approach equilibrium. A long model run gets closer to the endpoint at greater computational cost, yet remains model-dependent. Paleoclimate evidence samples an almost equilibrated real climate, but it requires uncertain proxy and forcing translation. These are evidential differences, not one universal numerical correction from effective to equilibrium ECS.[2][3][4]
Diagnostic: Does the reported temperature value describe the target equilibrium state, a short-run extrapolation, or a proxy-constrained inference, and are the feedback and forcing assumptions stated?
Structural–Framed Character¶
ECS is structural within climate science. Evaluative weight: larger values imply stronger warming under the defined experiment, but the quantity does not itself prescribe policy. Human-practice dependence: the doubled-CO2 and feedback conventions are chosen for comparability while climate response is physical. Institutional origin: the measure appears across assessments, model studies and paleoclimate work, not in one software package. Vocabulary travel: generic “sensitivity” can mean other input-output relations, so the full doubled-CO2 equilibrium roles must be checked. Import versus recognition: a new estimate must map its forcing and feedback convention to the target rather than reuse the label alone. The portable balanced-state role belongs to the live Equilibrium Prime; a broader reference → perturbation → balanced-response comparison pattern remains a separate future-Prime question. Its character: a specific climate-response quantity whose preindustrial doubled-CO2 temperature target and evidential limits keep it domain-specific.[1][2][3]
Structural Core vs. Domain Accent¶
The core is a temperature difference under sustained CO2 doubling from the preindustrial reference at planetary energy rebalance, with stated feedback scope. CCSM3, CESM 1.0.4, the Last Glacial Maximum, the 2.57 K and 2.3 K study estimates, and the particular regression or proxy models are accents. Replace one climate model with another and the target remains ECS even if its value changes. Remove the equilibrium endpoint or preindustrial doubled-CO2 reference and the measure changes identity.[2][3][1]
Feedback processes such as clouds or sea-ice albedo affect the result; no single one is ECS. A climate ensemble may compare many ECS estimates but is a collection of simulations, not this scalar property. The portable energy-balance endpoint is captured by the live Equilibrium Prime. A general reference → perturbation → balanced-response pattern across substrates, if warranted, is a future-Prime question; it is not a second asserted parent for this entry. The named ECS identity remains climate-specific because its preindustrial CO2 doubling, global surface-air temperature and ice-sheet feedback boundary have no independently evidenced cross-domain instance.
Instantiates / Related Primes¶
This entry presupposes Equilibrium.
The graph records one strict composition/presupposes edge to the live Equilibrium Prime. Every ECS target uses a climate state in which the sustained doubled-CO2 perturbation has been balanced by the system's radiative response on the specified long-run timescale. Incoming and outgoing planetary energy fluxes balance at that endpoint. An estimate may infer it without directly attaining it, but the endpoint remains constitutive. Equilibrium can describe many nonclimate systems without ECS; ECS is a temperature difference relying on a balanced state, not itself a subtype of state.[1][2]
The live Measurement Prime describes a procedure mapping an attribute through an instrument and scale; ECS is the response quantity that different procedures aim to estimate. The live Measure Prime is a set function and Metric is a mathematical distance function. The live Feedback Prime captures processes that contribute to ECS but is not its genus. Climate Ensemble, Cloud Feedback and Ice–Albedo Feedback are relevant collections or components rather than parents.
Relationships to Other Abstractions¶
Current abstraction Equilibrium Climate Sensitivity Domain-specific
Parents (1) — more general patterns this builds on
-
Equilibrium Climate Sensitivity presupposes Equilibrium Prime
The doubled-CO2 temperature endpoint presupposes a planetary state with balanced energy fluxes.Every admitted ECS target is a global surface-temperature difference between the preindustrial reference and the climate state after sustained doubled CO2 has restored planetary energy balance. The endpoint therefore presupposes Equilibrium in the live Prime sense: opposing incoming and outgoing planetary energy fluxes have zero net imbalance on the specified long-run scale. An estimate may infer this state without directly attaining it, but the defined target still requires it. Equilibrium occurs in many nonclimate systems without CO2 sensitivity, and ECS as a scalar difference is not itself a balanced state; composition/presupposes, not subsumption, expresses the strict dependency.
Hierarchy path (1) — routes to 1 parentless root
- Equilibrium Climate Sensitivity → Equilibrium → Fixed Point
Neighborhood in Abstraction Space¶
Equilibrium Climate Sensitivity sits in a sparse region of the domain-specific corpus (98th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (2551 abstractions)
Nearest neighbors
- Cloud feedback — 0.78
- Brunt–Väisälä Frequency — 0.77
- Reference Evapotranspiration Estimation — 0.76
- Albedo — 0.76
- Diffusive–Thermal Instability — 0.76
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
TCR: global warming at CO2 doubling during a defined rising-CO2 experiment, before equilibrium. Effective ECS: an extrapolated diagnostic from a finite run that may differ from eventual equilibrium. Climate sensitivity parameter: a temperature-per-unit-forcing relation whose normalization and conventions need checking. Earth-system sensitivity: a broader target when slow ice-sheet response is counted as feedback. Raw LGM cooling: a multi-forcing temperature contrast requiring translation. Cloud or ice-albedo feedback: components of response, not the full doubled-CO2 quantity. Mathematical Metric: a distance function, not this climate usage. Other-baseline doubled-CO2 sensitivity: a state-dependent experimental comparison requiring its own reference and translation before being called conventional ECS.[1][2][3][4]
References¶
[1] Intergovernmental Panel on Climate Change, “The Earth's Energy Budget, Climate Feedbacks, and Climate Sensitivity”, in Climate Change 2021, The Physical Science Basis, Working Group I, Chapter 7, DOI 10.1017/9781009157896.009, Box 7.1 and §§7.5, 7.5.3 and 7.5.5. Official assessment; definitions of ECS and TCR, effective-ECS conversion and paleoclimate forcing/ice-sheet limits. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r
[2] M. Rugenstein, J. Bloch-Johnson, J. Gregory, T. Andrews and T. Mauritsen, “Equilibrium Climate Sensitivity Estimated by Equilibrating Climate Models”, Geophysical Research Letters 47 (2020), DOI 10.1029/2019GL083898, §§1–3 and Table 1. Original full article; the quoted long-run and early-regression values belong to specified model experiments. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q
[3] A. Schmittner, N. M. Urban, J. D. Shakun et al., “Climate Sensitivity Estimated from Temperature Reconstructions of the Last Glacial Maximum”, Science 334 (2011): 1385–1388, DOI 10.1126/science.1203513, abstract, pp. 1385–1386 and Fig. 3. Original article; its reported posterior depends on model, proxy and forcing assumptions. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o
[4] M. D. Zelinka, T. A. Myers, D. T. McCoy et al., “Causes of Higher Climate Sensitivity in CMIP6 Models”, Geophysical Research Letters 47 (2020), DOI 10.1029/2019GL085782, §§1–2. Original full article; 150-year abrupt-4xCO2 regression diagnoses effective, rather than fully attained equilibrium, sensitivity. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h