The DOI matches its Crossref record; 72 of the 72 references it cites were found in a registry or an index; 16 of the 29 sources its claims cite were checked for support.
Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections
Open accessAbstract
Documenting the uncertainty of climate change projections is a fundamental objective of the inter-comparison exercises organized to feed into the Intergovernmental Panel on Climate Change (IPCC) reports. Usually, each modeling center contributes to these exercises with one or two configurations of its climate model, corresponding to a particular choice of “free parameter” values, resulting from a long and often tedious “model tuning” phase. How much uncertainty is omitted by this selection and how might readers of IPCC reports and users of climate projections be misled by its omission? We show here how recent machine learning approaches can transform the way climate model tuning is approached, opening the way to a simultaneous acceleration of model improvement and parametric uncertainty quantification. We show how an automatic selection of model configurations defined by different values of free parameters can produce different “warming worlds,” all consistent with present-day observations of the climate system.
Overview
CoChat Summary
TL;DR
The paper applies history matching, a machine-assisted tuning approach, to the IPSL-CM6A-LR climate model and derives four coupled configurations with contrasting equilibrium climate sensitivity (ECS) that remain broadly compatible with present-day observations. It argues that CMIP-style single-configuration contributions omit parametric uncertainty, potentially underestimating uncertainty in climate change projections.
Key findings
- History matching produced four coupled configurations with ECS values ranging from 3.7 to 5.4 K (abstract; Results reports 3.7 to 5.3 K), compared with 4.6 K for the standard IPSL-6A configuration.
- Two of the four configurations, Exp 1 and Exp 4, showed present-day climate skill similar to IPSL-6A and could have been selected instead of IPSL-6A for CMIP6; Exp 2 underestimated cloud radiative forcing, and Exp 3 had a very weak Atlantic Meridional Overturning Circulation.
- With only five configurations (IPSL-6A plus the four new ones), the authors cover about half of the CMIP multi-model ECS range, suggesting that CMIP ensembles may underestimate parametric uncertainty in ECS.
- Despite the wide ECS range, the simulated climate changes at a given global temperature change were more similar than different across configurations, leading the authors to advocate separate uncertainty quantification for ECS and for climate change at a given warming level.
Methods
- The authors translated the by-hand tuning targets of IPSL-6A into 14 scalar metrics (11 radiative, 3 rainfall) and treated 18 atmospheric free parameters as unknown, using prior ranges set by expert judgment without knowledge of IPSL-6A values.
- History matching generated parameter vectors by random sampling, ran 2-year stand-alone atmospheric simulations forced by observed SSTs, computed metrics on the second year, and fitted Gaussian process emulators to rule out implausible parameter vectors over iterative waves (250, 200, and 180 simulations in waves 1–3).
- ECS was estimated from abrupt4×CO2 coupled simulations using linear regression over the first 150 years, with a clim+4K atmosphere-only proxy used to select configurations with contrasted ECS before running full coupled experiments.
Limitations
- The paper states that the exploration is “very partial,” that the authors lack expertise to ensure reasonable tuning of sea ice cover or AMOC intensity, and that structural uncertainty is generally not known a priori, so history matching cannot fully guarantee against overtuning.
- The paper notes that about 800 2-year-long atmospheric simulations were needed to extract nine acceptable configurations, and that no historical simulations were run for Exp 1–4, so most evaluation compared piControl simulations with present-day observations.
- Evident from the text: the study is limited to one model (IPSL-CM6A-LR), uses a small number of selected configurations, and relies on an imperfect clim+4K ECS proxy for selecting contrasted cases before coupled confirmation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
![Fig. 1. Illustration of the iterative procedure for history matching. Both graphs display the latitudinal variations of the zonally averaged short-wave (SW) cloud radiative effect (CRE) computed as the difference between the total and clear-sky TOA SW radiation. The CERES-EBAF L3b observations are shown in black with error bars of ±4 W/m2 (46). (A) SW CRE computed for the second year of an ensemble of 2-year forced-by-SST simulations for the three successive waves of the history matching procedure with 250, 200, and 180 simulations for waves 1, 2, and 3, respectively. The nine “BEST” simulations are shown in red and the gray line corresponds to a control simulation (CTRL) run with the parameters of the IPSL-6A configuration. (B) Ten-year average SW-CRE obtained in coupled ocean-atmosphere simulations in the multi-model ensemble of CMIP5 (yellow) and CMIP6 (orange) and in the control and four experiments retained from the history matching because of both their good behavior in present-day conditions [taken among the nine “BEST” configurations of (A)] and of their contrasted ECS (see the main text). For these last four experiments, we show the SW CRE obtained in both the coupled (thick line) and stand-alone (thin, 1-year average) atmospheric simulations that are almost superimposed.](https://figures.cochat.ai/paper-figures/66b752ef94bf4417b4c49912e5a122b5/toward-machine-assisted-tuning-avoiding-the-underestimation-of-uncertainty-in-figure-1.jpg)

![Fig. 3. Metrics computed on the CTRL and Exp 1 to 4 piControl coupled simulations. The dispersion across those five configurations is compared with the multi-model spread for CMIP5 (yellow) and CMIP6 (orange) also computed from piControl simulations. It is compared as well with the multi-decadal variability in the IPSL-6A piControl simulation (one of the orange dots in the CMIP6 column), computed from 32 successive 30-year-long periods starting in 1850 (“IPSL-6A pi” column), and with the inter-member dispersion in the 33-member historical ensemble run with IPSL-6A for CMIP6 (“IPSL-6A Ens” column, brown). The period retained for the comparison of the historical simulations with observations is 1979–2005. The comparison of the IPSL pi and historical runs gives an estimate of the error, which is made by comparing with present-day observations simulations run in preindustrial rather than present-day conditions. (A) Global mean SST (°C). (B) Root mean square error computed on the mean seasonal cycle of the SST, between 65°N and 65°S to avoid sea ice dispersion, and after removing the mean bias (RMSC stands for centered root mean square). The mean bias is retrieved to minimize the impact of the global warming between 1850 and present day. (C) December-January-February-Marchmean sea ice extent. (D) Maximum intensity of the Atlantic Meridional Overturning Circulation (AMOC, Sv). (E to H) Root mean square error (RMSE) on the mean seasonal cycle of the Outgoing SW radiation [OSR, W/m2 (E)], Outgoing LW radiation [OLR, W/m2 (F)], total SW + LW Outgoing radiation [TOT RAD, W/m2 (G)], and rainfall [Precip, mm/day (H)].](https://figures.cochat.ai/paper-figures/50a2b358123f436285ed17c17528dcd2/toward-machine-assisted-tuning-avoiding-the-underestimation-of-uncertainty-in-figure-3.jpg)


Verification
Claims the paper makes
13 Valid · 1 Not valid · 11 Not checked · 74 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“note that this global mean SST fluctuates by about 0.2 K on centennial time scales in IPSL-6A (39)”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > RESULTS > Translating “by-hand tuning” into numbers
What the sources say
- Partly supportedcites [39] R. Bonnet, D. Swingedouw, G. Gastineau, O. Boucher, J. Deshayes, F. Hourdin, J. Mignot, J. Servonnat, A. Sima, Increased risk of near term global warming due to a recent AMOC weakening. Nat. Commun. 12, 6108 (2021). 34671020 PMC8528826 sourceExperimental — accuracy not yet measured
The Results and discussion section on low-frequency internal climate variability (page unavailable) says IPSL-CM6A-LR has multi-centennial variability in global near-surface air temperature (GSAT). It does not establish that global mean sea surface temperature (SST) fluctuates by about 0.2 K; searches for SST and 0.2 K did not find that result. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“However, 5 among the 33 members of an ensemble of historical simulations with this IPSL-6A version (that differed only by their initial state) showed climate trends over the 20th century compatible with observations”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > INTRODUCTION
What the sources say
- Partly supportedcites [39] R. Bonnet, D. Swingedouw, G. Gastineau, O. Boucher, J. Deshayes, F. Hourdin, J. Mignot, J. Servonnat, A. Sima, Increased risk of near term global warming due to a recent AMOC weakening. Nat. Commun. 12, 6108 (2021). 34671020 PMC8528826 sourceExperimental — accuracy not yet measured
The Results and discussion section says some IPSL-CM6A-LR ensemble members reproduce observed historical warming trends, and Methods says the historical simulations differed in their initial conditions. But the paper identifies a subset of six members consistent with observed GSAT and AMOC trends, not five, and the passages do not establish an ensemble size of 33. (LLM, conf=0.86)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“This strategy was already applied successfully to the by-hand tuning of the IPSL-6A configuration (36)”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > RESULTS > Translating “by-hand tuning” into numbers
What the sources say
- Supportedcites [36] J. Mignot, F. Hourdin, J. Deshayes, O. Boucher, G. Gastineau, I. Musat, M. Vancoppenolle, J. Servonnat, A. Caubel, F. Chéruy, S. Denvil, J.-L. Dufresne, C. Ethé, L. Fairhead, M.-A. Foujols, J.-Y. Grandpeix, G. Levavasseur, O. Marti, M. Menary, C. Rio, C. Rousset, Y. Silvy, The tuning strategy of IPSL-CM6A LR. J. Adv. Model. Earth Syst. 13, e2020MS002340 (2021). source
Section 5 (Discussion and Conclusion; page not provided) describes a tuning strategy developed to achieve IPSL-CM6A-LR and says its final stage involved three years of back-and-forth adjustments. This supports the claim that the configuration was tuned through an iterative, by-hand strategy. (LLM, conf=0.86)
plainstates a causeSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“This success of the automated procedure benefited from the expertise gained in the team from decades of practice of by-hand tuning of the IPSL model (14, 17, 18, 37).”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > DISCUSSION
What the sources say
- Not checked — no full text of the sourcecites [14] F. Hourdin, I. Musat, S. Bony, P. Braconnot, F. Codron, J.-L. Dufresne, L. Fairhead, M.-A. Filiberti, P. Friedlingstein, J.-Y. Grandpeix, G. Krinner, P. Levan, Z.-X. Li, F. Lott, The LMDZ4 general circulation model: Climate performance and sensitivity to parametrized physics with emphasis on tropical convection. Climate Dynam. 27, 787– 813 (2006). source
- Partly supportedcites [17] F. Hourdin, M.-A. Foujols, F. Codron, V. Guemas, J.-L. Dufresne, S. Bony, S. Denvil, L. Guez, F. Lott, J. Ghattas, P. Braconnot, O. Marti, Y. Meurdesoif, L. Bopp, Impact of the LMDZ atmospheric grid configuration on the climate and sensitivity of the IPSL-CM5A coupled model. Climate Dynam. 40, 2167– 2192 (2013). sourceExperimental — accuracy not yet measured
Section 3.3, “Need for tuning” (page unavailable), describes tuning IPSL model parameters and reports its effects, supporting the claim that the team had experience tuning the model. It does not say that this practice spanned decades or that the expertise contributed to the success of an automated history-matching procedure. (LLM, conf=0.82)
- Partly supportedcites [18] F. Hourdin, J.-Y. Grandpeix, C. Rio, S. Bony, A. Jam, F. Cheruy, N. Rochetin, L. Fairhead, A. Idelkadi, I. Musat, J.-L. Dufresne, A. Lahellec, M.-P. Lefebvre, R. Roehrig, LMDZ5B: The atmospheric component of the IPSL climate model with revisited parameterizations for clouds and convection. Climate Dynam. 40, 2193– 2222 (2013). sourceExperimental — accuracy not yet measured
Section 3.4 (page unavailable) says tuning the IPSL model remained somewhat manual, and describes a two-year iterative tuning process. It does not establish decades of by-hand tuning or say that the team's experience contributed to the success of an automated history-matching procedure. (LLM, conf=0.82)
- Partly supportedcites [37] F. Hourdin, C. Rio, J.-Y. Grandpeix, J.-B. Madeleine, F. Cheruy, N. Rochetin, A. Jam, I. Musat, A. Idelkadi, L. Fairhead, M.-A. Foujols, L. Mellul, A.-K. Traore, J.-L. Dufresne, O. Boucher, M.-P. Lefebvre, E. Millour, E. Vignon, J. Jouhaud, F. B. Diallo, F. Lott, G. Gastineau, A. Caubel, Y. Meurdesoif, J. Ghattas, LMDZ6A: The atmospheric component of the IPSL climate model with improved and better tuned physics. J. Adv. Model. Earth Syst. 12, e2019MS001892 (2020). sourceExperimental — accuracy not yet measured
Section 4.6 describes extensive tuning of the IPSL coupled model by many scientists, including retuning the atmosphere for each of 15 versions over two years. The paper also says the team had only recently started using history matching. It does not say that decades of by-hand tuning contributed to the success of an automated procedure; page numbers are unavailable in our copy. (LLM, conf=0.78)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Every source the paper cites for this claim was checked and supports it.
In the paper
“In particular, the rather good score in simulating SST in IPSL-6A comes from the fact that we targeted on purpose some biases in the atmospheric fluxes that directly impact persistent temperature biases as the East-Tropical-Ocean (53) and Circum Antartica warm biases.”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > DISCUSSION
What the sources say
- Supportedcites [53] F. Hourdin, C. Rio, A. Jam, A.-K. Traore, I. Musat, Convective boundary layer control of the sea surface temperature in the tropics. J. Adv. Model. Earth Syst. 12, e2019MS001988 (2020). source
The citation is attached to the East-Tropical-Ocean warm bias, not the separate Circum-Antarctic example. Section 3.5 reports that IPSL-CM6A reduced the east-tropical-ocean SST bias relative to earlier versions and links SST biases to atmospheric surface fluxes. The paper also explains that increased evaporation helps cool the sea surface in those regions. (LLM, conf=0.88)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
47 more claims · 3 with their cited sources checked
Create a free account to see all 99 claims and to verify the 6 references still open.
- Probably not a substantial claim. 27
- It may state two separate claims as one. 26
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > INTRODUCTION
“We thus derive four coupled configurations with ECS values ranging from 3.7 to 5.4 K.”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > RESULTS > Coupled simulation with contrasted ECS
“The ECS estimates vary from 3.7 to 5.3 K for a value of 4.6 K for the IPSL-6A and CTRL configuration.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > RESULTS > Coupled simulation with contrasted ECS
“The precipitation change in the three configurations looks also very similar at first glance for a temperature target of +5.5 K (left column in Fig. 5).”
§ Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections > RESULTS > Coupled simulation with contrasted ECS
“Figure: Fig. 5. Surface evaporation (mm/day) and surface net solar radiation change (W/m2) in coupled simulations for a global temperature change of 5.5 K.”
Citations
Cites
Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer, R. J., & Taylor, K. E. (2016) · Geoscientific Model Development · doi:10.5194/gmd-9-1937-2016 · Cited by 11,981
Huffman, G. J., Bolvin, D. T., Nelkin, E. J., Wolff, D. B., Adler, R. F., Gu, G., Hong, Y., Bowman, K. P., & Stocker, E. F. (2007) · Journal of Hydrometeorology · doi:10.1175/jhm560.1 · Cited by 6,767 · cited 1×
“For rainfall, the three metrics retained are as follows: (i) the variability over ocean around the maritime continent (underestimated in the IPSL-6A version), with a target from The Tropical Rainfall Measuring Mission (TRMM) daily rainfall product (47) and a relative uncertainty of 10%; (ii) the annual mean rainfall o…”
BackgroundA dynamic global vegetation model for studies of the coupled atmosphere‐biosphere system
Krinner, G., Viovy, N., de Noblet‐Ducoudré, N., Ogée, J., Polcher, J., Friedlingstein, P., Ciais, P., Sitch, S., & Prentice, I. C. (2005) · Global Biogeochemical Cycles · doi:10.1029/2003gb002199 · Cited by 2,496
Presentation and Evaluation of the IPSL‐CM6A‐LR Climate Model
Boucher, O., Servonnat, J., Albright, A. L., Aumont, O., Balkanski, Y., Bastrikov, V., Bekki, S., Bonnet, R., Bony, S., Bopp, L., et al. (2020) · Journal of Advances in Modeling Earth Systems · doi:10.1029/2019ms002010 · Cited by 1,694 · cited 2×
“The IPSL coupled model IPSL-CM6A-LR The tuning exercise and ECS exploration are done with IPSLCM6A-LR (called IPSL-6A), the standard configuration of the IPSL coupled model designed for CMIP6 (35).”
MethodSelf-citationAn Assessment of Earth's Climate Sensitivity Using Multiple Lines of Evidence
Sherwood, S. C., Webb, M. J., Annan, J. D., Armour, K. C., Forster, P. M., Hargreaves, J. C., Hegerl, G., Klein, S. A., Marvel, K. D., Rohling, E. J., et al. (2020) · Reviews of Geophysics · doi:10.1029/2019rg000678 · Cited by 1,254
AMIP: The Atmospheric Model Intercomparison Project
Gates, W. L. (1992) · Bulletin of the American Meteorological Society · doi:10.1175/1520-0477(1992)073<1962:atamip>2.0.co;2 · Cited by 1,124
A new method for diagnosing radiative forcing and climate sensitivity
Gregory, J. M., Ingram, W. J., Palmer, M. A., Jones, G. S., Stott, P. A., Thorpe, R. B., Lowe, J. A., Johns, T. C., & Williams, K. D. (2004) · Geophysical Research Letters · doi:10.1029/2003gl018747 · Cited by 1,075
Toward Optimal Closure of the Earth's Top-of-Atmosphere Radiation Budget
Loeb, N. G., Wielicki, B. A., Doelling, D. R., Smith, G. L., Keyes, D. F., Kato, S., Manalo-Smith, N., & Wong, T. (2009) · Journal of Climate · doi:10.1175/2008jcli2637.1 · Cited by 960 · cited 2×
“All the radiative metrics but one were given a 1σ tolerance to error of ±5 W/m(2) for an observational uncertainty of about 4 W/ m(2) for the CERES-EBAF L3b satellite product used as target (46).”
MethodHourdin, F., Musat, I., Bony, S., Braconnot, P., Codron, F., Dufresne, J.-L., Fairhead, L., Filiberti, M.-A., Friedlingstein, P., Grandpeix, J.-Y., et al. (2006) · Climate Dynamics · doi:10.1007/s00382-006-0158-0 · Cited by 844
Self-citationMeehl, G. A., Senior, C. A., Eyring, V., Flato, G., Lamarque, J.-F., Stouffer, R. J., Taylor, K. E., & Schlund, M. (2020) · Science Advances · doi:10.1126/sciadv.aba1981 · Cited by 823
Research works printed but not linked: 2
- P. Forster T. Storelvmo K. Armour W. Collins J.-L. Dufresne D. Frame D. Lunt T. Mauritsen M. Palmer M. Watanabe M. Wild H. Zhang Climate Change 2021: The Scientific Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change V. Masson-Delmotte P. Zhai A. Pirani S. L. Connors C. Péan S. Berger N. Caud Y. Chen L. Goldfarb M. I. Gomis M. Huang K. Leitzell E. Lonnoy J. B. R. Matthews T. K. Maycock T. Waterfield O. Yelekçi R. Yu and B. Zhou Eds. (Cambridge Univ. Press 2021) chap. 7 pp. 923–1054.
- P. N. Edwards A Vast Machine: Computer Models Climate Data and the Politics of Global Warming (MIT Press 2010).
Cited by
Opinion: Optimizing climate models with process knowledge, resolution, and artificial intelligence
Schneider, T., Leung, L. R., & Wills, R. C. J. (2024) · Atmospheric Chemistry and Physics · doi:10.5194/acp-24-7041-2024 · Cited by 53
Jones, C., Adloff, F., Booth, B. B. B., Cox, P. M., Eyring, V., Friedlingstein, P., Frieler, K., Hewitt, H. T., Jeffery, H. A., Joussaume, S., et al. (2024) · Earth System Dynamics · doi:10.5194/esd-15-1319-2024 · Cited by 38
Using Machine Learning to Generate a GISS ModelE Calibrated Physics Ensemble (CPE)
Elsaesser, G. S., Elsaesser, G. S., van Lier‐Walqui, M., Yang, Q., Kelley, M., Ackerman, A. S., Fridlind, A. M., Cesana, G., Schmidt, G. A., Wu, J., et al. (2025) · Journal of Advances in Modeling Earth Systems · doi:10.1029/2024ms004713 · Cited by 50
“…the body of work on parameter calibration (Gregoire et al., 2011; Hourdin et al., 2017, 2020; Järvinen et al., 2010; Ma et al., 2022; Mauritsen et al., 2012; Mignot et al., 2021), focusing on automated calibration frameworks (Hourdin et al., 2023; Watson‐Parris et al., 2021; Yarger et al., 2024).”
Raoult, N., Douglas, N., MacBean, N., Kolassa, J., Quaife, T. L., Roberts, A. G., Fisher, R. A., Fer, I., Bacour, C., Dagon, K., et al. (2025) · Journal of Advances in Modeling Earth Systems · doi:10.1029/2024ms004733 · Cited by 61
“In the field of uncertainty quantification, one key method that can be used in combination with emulators and that is gaining traction for LSM parameter estimation is the so‐called HM method (Hourdin et al., 2023).”
Exploration of diverse solutions for the calibration of imperfect climate models
Peatier, S., Sanderson, B. M., & Terray, L. (2024) · Earth System Dynamics · doi:10.5194/esd-15-987-2024 · Cited by 15
“The method, however, makes some significant assumptions about the ability of a statistical model to predict the parametric response of high-order modes and does not allow an exploration of structural trade-offs between different variables, such as those found by McNeall et al.”
Metadata
- Authors
- Hourdin, F., Ferster, B., Deshayes, J., Mignot, J., Musat, I., & Williamson, D.
- Affiliations
- Centre National de la Recherche Scientifique (FR)
- University of Exeter (GB)
- Corresponding author
- Hourdin, Frédéric
- Venue
- Science Advances
- Published
- 2023-07-21
- Volume
- 9
- Issue
- 29
- Pages
- eadf2758-eadf2758
- ISSN
- 2375-2548 (ISSN-L)
- PMC
- PMC10355829
- OpenAlex
- W4384818614
- Cited by
- 49
- References
- 58
- Citations by year
- 2026: 11 · 2025: 14 · 2024: 20 · 2023: 4
- FWCI
- 5.49
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by-nc
- Locations
- Science Advances (publishedVersion, cc-by-nc)
- PubMed Central (publishedVersion, cc-by-nc)
- PubMed (publishedVersion)
- HAL (submittedVersion, cc-by-nc)
- Funders
- Indiana Corn Marketing Council
- Agence Nationale de la Recherche (ANR-16-CE01-0010, ANR-18-MPGA-0001)
- Partnership for Advanced Computing in Europe AISBL
- Centre National d’Etudes Spatiales
- Centre National de la Recherche Scientifique
- Grand Équipement National De Calcul Intensif (2021-A0100107732)
- Sorbonne Université
- Engineering and Physical Sciences Research Council (N510129, EP/Y005597/1, EP/N510129/, EP/N510129/1)
- Biotechnology and Biological Sciences Research Council (BB/V011588/1)
- Natural Environment Research Council (NE/N018486/1)
- Topics
- Physical Sciences › Environmental Science › Global and Planetary Change › Climate variability and models
- Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Meteorological Phenomena and Simulations
- Physical Sciences › Environmental Science › Global and Planetary Change › Climate Change and Geoengineering
- Keywords
- climate model tuning, machine learning, uncertainty quantification, climate change projections, model parameter selection, free parameters, model intercomparison, IPCC
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections. https://app.cochat.ai/papers/10.1126/sciadv.adf2758
- The paper
- Hourdin, F., Ferster, B., Deshayes, J., Mignot, J., Musat, I., & Williamson, D. (2023). Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections. Science Advances. https://doi.org/10.1126/sciadv.adf2758
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.