The global mean surface temperature rises by about 3.3 K from January to July, providing a large, annually repeating, well-observed natural experiment. Tsushima and Manabe (2001) found from ERBE satellite data that the net influence of clouds on this annual variation is near zero, while the general circulation models they examined produced systematically different seasonal cloud responses; they proposed the comparison as a test of model cloud feedback but left open whether — given that the seasonal temperature pattern differs fundamentally from the greenhouse-warming pattern — the observed relationships can be generalized to global warming, how they decompose across latitudes, cloud types, and regions, which parameterization choices cause the model discrepancies, and whether the seasonal cycle quantitatively constrains climate sensitivity. The authors' own 2005 and 2013 follow-ups established multi-model cloud-radiative-forcing biases on the annual cycle but left the solar component observationally unsettled (discrepant CERES data versions) and did not attempt attribution or correction; the pattern-effect and cloud-controlling-factor literature quantified why variability-derived feedbacks need not transfer to forced warming; and 2023 studies demonstrated ensemble-specific links between present-day seasonal cloud behavior and warming cloud feedback or equilibrium climate sensitivity (a HadGEM3 perturbed-physics ensemble; a CMIP6 emergent constraint) that remain ensemble- and dataset-dependent. No validated quantitative transfer from observed seasonal-cycle cloud behavior to warming cloud feedback or climate sensitivity exists, and the attribution of model seasonal cloud biases to specific parameterization choices with demonstrated correction is unresolved. The problem is open.

Background

Earth's equilibrium climate sensitivity (ECS) — the global mean surface warming reached at equilibrium after a doubling of atmospheric CO2_2 — is uncertain mainly because of cloud feedback: the amplifying or damping effect on warming of changes in cloud amount, height, and optical properties. Clouds control the top-of-atmosphere (TOA) radiative budget by reflecting solar radiation and trapping outgoing longwave radiation; the net radiative effect of clouds relative to a hypothetical cloud-free atmosphere is called cloud radiative forcing (CRF, also cloud radiative effect). If CRF changes systematically with temperature, clouds exert a feedback on climate. Strictly, the feedback of CRF is not the cloud feedback proper: clouds also mask the non-cloud feedbacks of water vapor, lapse rate, and surface albedo from the satellite’s view (cloud masking), so the two coincide only after this masking is separated. Because the satellite radiation record is short and the forced warming trend is small compared with natural variability, cloud feedback cannot be measured directly from the secular trend, and it remains the dominant source of intermodel spread in climate sensitivity (Bony et al. 2006).

A complementary strategy is to exploit the annual cycle. The global mean surface temperature rises by about 3.3 K3.3\ \mathrm{K} from January to July — mainly because the Northern Hemisphere, with its large land fraction, has a much stronger seasonal cycle than the ocean-dominated Southern Hemisphere — providing a large, annually repeating, well-observed temperature perturbation. Tsushima and Manabe (2001) regressed monthly, globally averaged TOA fluxes from the Earth Radiation Budget Experiment (ERBE, a multi-satellite program measuring TOA radiative fluxes) against the global mean surface temperature, using annually normalized reflected solar radiation (scaled by the ratio of annual-mean to instantaneous incoming solar radiation) to remove the direct effect of the seasonal change in insolation. They found the net influence of clouds on the annual variation of global mean surface temperature to be quite small: on the global scale, neither the solar reflectivity of clouds nor their effective cloud-top height showed a significant relationship with the annual cycle of surface temperature. In contrast, the three general circulation models (GCMs) they examined — models that explicitly predict cloud microphysical properties — produced both increasing cloud reflectivity and increasing cloud-top height with increasing surface temperature. The authors proposed this comparison as a test of model cloud feedback.

To compare feedbacks across models, this lineage uses the feedback parameter (the change in TOA outgoing radiation per degree of global mean surface temperature change) and the gain factor, a nondimensional measure of the strength of feedbacks of the second kind (those involving water vapor, lapse rate, clouds, and surface albedo, as opposed to the basic Stefan–Boltzmann damping of a vertically uniform temperature perturbation) relative to that basic damping; a CRF gain factor near zero means clouds hardly modify the radiative damping.

The authors’ own follow-ups extended the test without settling it: Tsushima, Abe-Ouchi, and Manabe (2005) found the overall radiative damping of the annual variation to be similar between models and observations while its solar and terrestrial components disagree individually, and Tsushima and Manabe (2013) repeated the comparison against 35 models from CMIP3 and CMIP5 using ERBE and two versions of the CERES dataset (Clouds and the Earth’s Radiant Energy System, the successor satellite radiation program), confirming the near-zero observed cloud-radiative-forcing gain factor but leaving the solar component observationally unsettled because the two CERES products disagree.

The 2001 paper left the central issue open: the geographical pattern of the seasonal temperature anomaly (hemispherically antisymmetric, annually periodic, land-dominated) differs fundamentally from the pattern of greenhouse-gas-induced warming, so it is unknown whether the near-zero seasonal cloud feedback says anything quantitative about cloud feedback under global warming; the global averaging leaves the latitude, region, and cloud-type dependence undetermined; the responsible model deficiencies were not identified; and no constraint on climate sensitivity was derived. Later work sharpened both sides of the question. The pattern effect — the dependence of radiative feedbacks on the spatial pattern of surface temperature change — is now recognized as a fundamental obstacle to transferring variability-derived feedbacks to forced warming (Myers, Zelinka, and Klein 2023). Conversely, the cloud-controlling factor paradigm treats observed sensitivities of clouds to their large-scale environmental controls (surface temperature, lower-tropospheric static stability, and others) as invariant across time scales and uses them to constrain warming cloud feedback statistically (Klein et al. 2017; Ceppi and Nowack 2021). A perturbed-physics ensemble (a set of model variants in which uncertain parameters of the subgrid parameterizations are systematically varied) of the HadGEM3 model and the CMIP archive (the Coupled Model Intercomparison Project, which provides the multi-model output used in IPCC assessments) have both been used to test whether present-day seasonal cloud behavior predicts warming cloud feedback, with partial but ensemble-dependent success (Furtado et al. 2023; Jiang et al. 2023).

Problem Statement

Can the observed relationships between clouds and the annual/seasonal cycle of surface temperature be used as a quantitative, observationally grounded test of the cloud feedback that will operate under CO2-induced global warming — and, where models fail this test, to identify and correct the responsible cloud parameterization deficiencies?

The source study (Tsushima and Manabe 2001) established the two ingredients that make this question sharp: the observed net cloud influence on the 3.3 K annual cycle of global mean surface temperature is near zero (no significant global-scale relationship between cloud solar reflectivity or effective cloud-top height and the annual cycle of surface temperature, from ERBE data), whereas the general circulation models it examined produce systematically different seasonal cloud responses. The unresolved question, common to the four linked aspects below, is whether this annually repeating natural experiment contains usable information about cloud feedback under global warming:

  1. Transfer: establish, quantitatively and with uncertainty, whether and how much the seasonal-cycle cloud–temperature sensitivities — globally or in specific regions and cloud types — predict the cloud feedback that operates under CO2-driven warming, resolving the pattern dependence of feedbacks (the seasonal anomaly is hemispherically antisymmetric and annually periodic, unlike greenhouse warming) and the separation of cloud feedback from cloud masking in cloud-radiative-forcing regressions.
  2. Representativeness: determine how the observed relationships decompose across latitudes, regions, and cloud types, beyond the global averaging used in the source analysis.
  3. Attribution and correction: identify which cloud parameterization choices cause the documented model–observation discrepancies in the seasonal cloud response, and demonstrate physically justified corrections that reduce them without degrading other observational constraints.
  4. Constraint on sensitivity: quantify what the observed seasonal-cycle cloud behavior does and does not constrain about equilibrium climate sensitivity.

These aspects jointly serve one objective: turning the observed annual cycle into a validated constraint on warming cloud feedback and an actionable diagnostic of model cloud parameterizations. The question admits either a positive or a negative answer about the transfer: the seasonal-cycle cloud–temperature relationships may or may not predict the warming cloud feedback beyond stated bounds, and attributing a model–observation discrepancy to specific parameterization choices is part of the attribution-and-correction aspect whenever such a discrepancy is claimed.

References

  1. Tsushima, Yoko, and Syukuro Manabe (2001). Influence of cloud feedback on annual variation of global mean surface temperature. Journal of Geophysical Research: Atmospheres 106(D19), 22635–22646. DOI: 10.1029/2000jd000235. https://doi.org/10.1029/2000jd000235
  2. Tsushima, Y., A. Abe-Ouchi, and S. Manabe (2005). Radiative damping of annual variation in global mean surface temperature: comparison between observed and simulated feedback. Climate Dynamics 24, 591–597. DOI: 10.1007/s00382-005-0002-y. https://doi.org/10.1007/s00382-005-0002-y
  3. Bony, Sandrine, Robert Colman, Vladimir M. Kattsov, Richard P. Allan, Christopher S. Bretherton, Jean-Louis Dufresne, Alex Hall, Stephane Hallegatte, Marika M. Holland, William Ingram, David A. Randall, Brian J. Soden, George Tselioudis, and Mark J. Webb (2006). How Well Do We Understand and Evaluate Climate Change Feedback Processes? Journal of Climate 19, 3445–3482. DOI: 10.1175/JCLI3819.1. https://doi.org/10.1175/JCLI3819.1
  4. Colman, Robert A., and Scott B. Power (2010). Atmospheric radiative feedbacks associated with transient climate change and climate variability. Climate Dynamics 34, 919–933. DOI: 10.1007/s00382-009-0541-8. https://doi.org/10.1007/s00382-009-0541-8
  5. Tsushima, Yoko, and Syukuro Manabe (2013). Assessment of radiative feedback in climate models using satellite observations of annual flux variation. Proceedings of the National Academy of Sciences 110(19), 7568–7573. DOI: 10.1073/pnas.1216174110. https://doi.org/10.1073/pnas.1216174110 (open full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC3651433/)
  6. Klein, Stephen A., Alex Hall, Joel R. Norris, and Robert Pincus (2017). Low-Cloud Feedbacks from Cloud-Controlling Factors: A Review. Surveys in Geophysics 38, 1307–1329. DOI: 10.1007/s10712-017-9433-3. https://doi.org/10.1007/s10712-017-9433-3
  7. Dessler, A. E., and P. M. Forster (2018). An Estimate of Equilibrium Climate Sensitivity From Interannual Variability. Journal of Geophysical Research: Atmospheres 123, 8634–8645. DOI: 10.1029/2018JD028481. https://doi.org/10.1029/2018JD028481
  8. Sherwood, S. C., M. J. Webb, J. D. Annan, K. C. Armour, P. M. Forster, J. C. Hargreaves, G. Hegerl, S. A. Klein, K. D. Marvel, E. J. Rohling, M. Watanabe, T. Andrews, P. Braconnot, C. S. Bretherton, G. L. Foster, Z. Hausfather, A. S. von der Heydt, R. Knutti, T. Mauritsen, J. R. Norris, C. Proistosescu, M. Rugenstein, G. A. Schmidt, K. B. Tokarska, and M. D. Zelinka (2020). An Assessment of Earth's Climate Sensitivity Using Multiple Lines of Evidence. Reviews of Geophysics 58(4), e2019RG000678. DOI: 10.1029/2019RG000678. https://doi.org/10.1029/2019RG000678
  9. Zelinka, Mark D., Timothy A. Myers, Daniel T. McCoy, Stephen Po-Chedley, Peter M. Caldwell, Paulo Ceppi, Stephen A. Klein, and Karl E. Taylor (2020). Causes of Higher Climate Sensitivity in CMIP6 Models. Geophysical Research Letters 47, e2019GL085782. DOI: 10.1029/2019GL085782. https://doi.org/10.1029/2019GL085782
  10. Ceppi, Paulo, and Peer Nowack (2021). Observational evidence that cloud feedback amplifies global warming. Proceedings of the National Academy of Sciences 118(30), e2026290118. DOI: 10.1073/pnas.2026290118. https://doi.org/10.1073/pnas.2026290118
  11. Myers, Timothy A., Mark D. Zelinka, and Stephen A. Klein (2023). Observational Constraints on the Cloud Feedback Pattern Effect. Journal of Climate 36(18), 6533–6545. DOI: 10.1175/JCLI-D-22-0862.1. https://doi.org/10.1175/JCLI-D-22-0862.1
  12. Furtado, K., Y. Tsushima, P. R. Field, J. Rostron, and D. Sexton (2023). The Relationship Between the Present-Day Seasonal Cycles of Clouds in the Mid-Latitudes and Cloud-Radiative Feedback. Geophysical Research Letters 50, e2023GL103902. DOI: 10.1029/2023gl103902. https://doi.org/10.1029/2023gl103902
  13. Jiang, Xianan, Hui Su, Jonathan H. Jiang, J. David Neelin, Longtao Wu, Yoko Tsushima, and Gregory Elsaesser (2023). Muted extratropical low cloud seasonal cycle is closely linked to underestimated climate sensitivity in models. Nature Communications 14, 5586. DOI: 10.1038/s41467-023-41360-0. https://doi.org/10.1038/s41467-023-41360-0
  14. Roemer, Florian E., Stefan A. Buehler, Manfred Brath, Lukas Kluft, and Viju O. John (2023). Direct observation of Earth's spectral long-wave feedback parameter. Nature Geoscience 16, 416–421. DOI: 10.1038/s41561-023-01175-6. https://doi.org/10.1038/s41561-023-01175-6