Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Butterfly effect and a self-modulating El Niño response to global warming

An Addendum to this article was published on 18 February 2021

A Publisher Correction to this article was published on 17 November 2020

This article has been updated

Abstract

El Niño and La Niña, collectively referred to as the El Niño–Southern Oscillation (ENSO), are not only highly consequential1,2,3,4,5,6 but also strongly nonlinear7,8,9,10,11,12,13,14. For example, the maximum warm anomalies of El Niño, which occur in the equatorial eastern Pacific Ocean, are larger than the maximum cold anomalies of La Niña, which are centred in the equatorial central Pacific Ocean7,8,9. The associated atmospheric nonlinear thermal damping cools the equatorial Pacific during El Niño but warms it during La Niña15,16. Under greenhouse warming, climate models project an increase in the frequency of strong El Niño and La Niña events, but the change differs vastly across models17, which is partially attributed to internal variability18,19,20,21,22,23. Here we show that like a butterfly effect24, an infinitesimal random perturbation to identical initial conditions induces vastly different initial ENSO variability, which systematically affects its response to greenhouse warming a century later. In experiments with higher initial variability, a greater cumulative oceanic heat loss from ENSO thermal damping reduces stratification of the upper equatorial Pacific Ocean, leading to a smaller increase in ENSO variability under subsequent greenhouse warming. This self-modulating mechanism operates in two large ensembles generated using two different models, each commencing from identical initial conditions but with a butterfly perturbation24,25; it also operates in a large ensemble generated with another model commencing from different initial conditions25,26 and across climate models participating in the Coupled Model Intercomparison Project27,28. Thus, if the greenhouse-warming-induced increase in ENSO variability29 is initially suppressed by internal variability, future ENSO variability is likely to be enhanced, and vice versa. This self-modulation linking ENSO variability across time presents a different perspective for understanding the dynamics of ENSO variability on multiple timescales in a changing climate.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Butterfly effect on ENSO variability.
Fig. 2: Impact on equatorial Pacific Ocean heat balance arising from the butterfly effect.
Fig. 3: Self-modulating mechanism of the ENSO response to greenhouse warming.
Fig. 4: Robustness of ENSO self-modulation in large ensembles generated with other models.
Fig. 5: Self-modulating mechanism of ENSO response in CMIP5 and CMIP6 models.

Data availability

Data related to the paper can be downloaded from the following: ORA-S5, https://www.ecmwf.int/en/research/climate-reanalysis/ocean-reanalysis; HadISST v1.1, https://www.metoffice.gov.uk/hadobs/hadisst/; ERSST v5, https://www.ncdc.noaa.gov/data-access/marineocean-data/extended-reconstructed-sea-surface-temperature-ersst-v5/; NCEP/NCAR reanalysis, https://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.derived.surfaceflux.html/; ERA5, https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5; CMIP5, https://esgf-node.llnl.gov/projects/cmip5/; CMIP6, https://esgf-node.llnl.gov/projects/cmip6/; CESM-LENS, http://www.cesm.ucar.edu/projects/community-projects/LENS/data-sets.html; GFDL-CM3, http://www.cesm.ucar.edu/projects/community-projects/MMLEA/; GFDL-ESM2M, http://www.cesm.ucar.edu/projects/community-projects/MMLEA/.

Code availability

The codes for calculating EOF and the parameter |αD| can be downloaded from https://drive.google.com/open?id=1d2R8wKpFNW-vMIfoJsbqIGPIBd9Z_8rj. All codes are available upon request.

Change history

References

  1. 1.

    Ropelewski, C. F. & Halpert, M. S. Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation. Mon. Weath. Rev. 115, 1606–1626 (1987).

    ADS  Google Scholar 

  2. 2.

    McPhaden, M. J., Zebiak, S. E. & Glantz, M. H. ENSO as an integrating concept in Earth science. Science 314, 1740–1745 (2006).

    ADS  CAS  PubMed  Google Scholar 

  3. 3.

    Jonkman, S. N. Global perspectives on loss of human life caused by floods. Nat. Hazards 34, 151–175 (2005).

    Google Scholar 

  4. 4.

    Kunii, O., Nakamura, S., Abdur, R. & Wakai, S. The impact on health and risk factors of the diarrhoea epidemics in the 1998 Bangladesh floods. Public Health 116, 68–74 (2002).

    CAS  PubMed  Google Scholar 

  5. 5.

    del Ninno, C. & Dorosh, P. A. Averting a food crisis: private imports and public targeted distribution in Bangladesh after the 1998 flood. Agric. Econ. 25, 337–346 (2001).

    Google Scholar 

  6. 6.

    Cai, W. et al. More extreme swings of the South Pacific convergence zone due to greenhouse warming. Nature 488, 365–369 (2012).

    ADS  CAS  PubMed  Google Scholar 

  7. 7.

    Frauen, C. & Dommenget, D. El Niño and La Niña amplitude asymmetry caused by atmospheric feedbacks. Geophys. Res. Lett. 37, L18801 (2010).

    ADS  Google Scholar 

  8. 8.

    Takahashi, K., Montecinos, A., Goubanova, K. & Dewitte, B. ENSO regimes: reinterpreting the canonical and Modoki El Niño. Geophys. Res. Lett. 38, L10704 (2011).

    ADS  Google Scholar 

  9. 9.

    Choi, K.-Y., Vecchi, G. A. & Wittenberg, A. T. ENSO transition, duration, and amplitude asymmetries: role of the nonlinear wind stress coupling in a conceptual model. J. Clim. 26, 9462–9476 (2013).

    ADS  Google Scholar 

  10. 10.

    Dommenget, D., Bayr, T. & Frauen, C. Analysis of the non-linearity in the pattern and time evolution of El Niño Southern Oscillation. Clim. Dyn. 40, 2825–2847 (2013).

    Google Scholar 

  11. 11.

    Takahashi, K. & Dewitte, B. Strong and moderate nonlinear El Niño regimes. Clim. Dyn. 46, 1627–1645 (2016).

    Google Scholar 

  12. 12.

    Cai, W. et al. ENSO and greenhouse warming. Nat. Clim. Change 5, 849–859 (2015).

    ADS  Google Scholar 

  13. 13.

    Karamperidou, C., Jin, F.-F. & Conroy, J. L. The importance of ENSO nonlinearities in tropical Pacific response to external forcing. Clim. Dyn. 49, 2695–2704 (2017).

    Google Scholar 

  14. 14.

    Geng, T., Cai, W., Wu, L. & Yang, Y. Atmospheric convection dominates genesis of ENSO asymmetry. Geophys. Res. Lett. 46, 8387–8396 (2019).

    ADS  Google Scholar 

  15. 15.

    Sun, D.-Z. et al. Radiative and dynamical feedbacks over the equatorial cold tongue: results from nine atmospheric GCMs. J. Clim. 19, 4059–4074 (2006).

    ADS  Google Scholar 

  16. 16.

    Lloyd, J., Guilyardi, E., Weller, H. & Slingo, J. The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmos. Sci. Lett. 10, 170–176 (2009).

    ADS  Google Scholar 

  17. 17.

    Cai, W. et al. Increased variability of eastern Pacific El Niño under greenhouse warming. Nature 564, 201–206 (2018).

    ADS  CAS  PubMed  Google Scholar 

  18. 18.

    Wittenberg, A. T. Are historical records sufficient to constrain ENSO simulations? Geophys. Res. Lett. 36, L12702 (2009).

    ADS  Google Scholar 

  19. 19.

    Stevenson, S., Fox-Kemper, B., Jochum, M., Rajagopalan, B. & Yeager, S. G. ENSO model validation using wavelet probability analysis. J. Clim. 23, 5540–5547 (2010).

    ADS  Google Scholar 

  20. 20.

    Stevenson, S. L. Significant changes to ENSO strength and impacts in the twenty-first century: Results from CMIP5. Geophys. Res. Lett. 39, L17703 (2012).

    ADS  Google Scholar 

  21. 21.

    Cobb, K. M. et al. Highly variable El Niño–Southern Oscillation throughout the Holocene. Science 339, 67–70 (2013).

    ADS  CAS  PubMed  Google Scholar 

  22. 22.

    Maher, N., Matei, D., Milinski, S. & Marotzke, J. ENSO change in climate projections: forced response or internal variability? Geophys. Res. Lett. 45, 11390–11398 (2018).

    ADS  Google Scholar 

  23. 23.

    Zheng, X. T., Hui, C. & Yeh, S. W. Response of ENSO amplitude to global warming in CESM large ensemble: uncertainty due to internal variability. Clim. Dyn. 50, 4019–4035 (2018).

    Google Scholar 

  24. 24.

    Lorenz, E. N. The predictability of a flow which possesses many scales of motion. Tellus 21, 289–307 (1969).

    ADS  Google Scholar 

  25. 25.

    Deser, C. et al. Insights from Earth system model initial-condition large ensembles and future prospects. Nat. Clim. Change 10, 277–286 (2020); correction https://doi.org/10.1038/s41558-020-0854-5 (2020).

    ADS  Article  Google Scholar 

  26. 26.

    Rodgers, K. B., Lin, J. & Frölicher, T. L. Emergence of multiple ocean ecosystem drivers in a large ensemble suite with an Earth system model. Biogeosciences 12, 3301–3320 (2015).

    ADS  Google Scholar 

  27. 27.

    Taylor, K. E., Stouffer, R. J. & Meehl, G. A. An overview of CMIP5 and the experiment design. Bull. Am. Meteorol. Soc. 93, 485–498 (2012).

    ADS  Google Scholar 

  28. 28.

    Eyring, V. et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev. 9, 1937–1958 (2016).

    ADS  Google Scholar 

  29. 29.

    Cai, W. et al. Increasing frequency of extreme El Niño events due to greenhouse warming. Nat. Clim. Change 4, 111–116 (2014).

    ADS  Google Scholar 

  30. 30.

    Kug, J.-S., Jin, F.-F. & An, S.-I. Two types of El Niño events: cold tongue El Niño and warm pool El Niño. J. Clim. 22, 1499–1515 (2009).

    ADS  Google Scholar 

  31. 31.

    Kao, H. Y. & Yu, J.-Y. Contrasting Eastern-Pacific and Central-Pacific types of ENSO. J. Clim. 22, 615–632 (2009).

    ADS  Google Scholar 

  32. 32.

    Cai, W. et al. More frequent extreme La Niña events under greenhouse warming. Nat. Clim. Change 5, 132–137 (2015).

    ADS  Google Scholar 

  33. 33.

    Lorenz, E. N. Empirical Orthogonal Functions and Statistical Weather Prediction Statistical Forecast Project Report 1 (MIT Department of Meteorology, 1956).

  34. 34.

    Lorenz, E. N. Deterministic nonperiodic flow. J. Atmos. Sci. 20, 130–141 (1963).

    ADS  MathSciNet  MATH  Google Scholar 

  35. 35.

    Kay, J. E. et al. The community earth system model (CESM) large ensemble project: a community resource for studying climate change in the presence of internal climate variability. Bull. Am. Meteorol. Soc. 96, 1333–1349 (2015).

    ADS  Google Scholar 

  36. 36.

    Bayr, T., Dommenget, D. & Latif, M. Walker circulation controls ENSO atmospheric feedbacks in uncoupled and coupled climate model simulations. Clim. Dyn. 54, 2831–2846 (2020).

    Google Scholar 

  37. 37.

    Guilyardi, E. et al. Understanding El Niño in ocean–atmosphere general circulation models: progress and challenges. Bull. Am. Meteorol. Soc. 90, 325–340 (2009).

    ADS  Google Scholar 

  38. 38.

    Bellenger, H., Guilyardi, É., Leloup, J., Lengaigne, M. & Vialard, J. ENSO representation in climate models: from CMIP3 to CMIP5. Clim. Dyn. 42, 1999–2018 (2014).

    Google Scholar 

  39. 39.

    Bayr, T. et al. Mean-state dependence of ENSO atmospheric feedbacks in climate models. Clim. Dyn. 50, 3171–3194 (2018).

    Google Scholar 

  40. 40.

    Bayr, T. et al. Error compensation of ENSO atmospheric feedbacks in climate models and its influence on simulated ENSO dynamics. Clim. Dyn. 53, 155–172 (2019).

    Google Scholar 

  41. 41.

    Choi, J., An, S. I., Kug, J. S. & Yeh, S. W. The role of mean state on changes in El Niño’s flavor. Clim. Dyn. 37, 1205–1215 (2011).

    Google Scholar 

  42. 42.

    Vijayeta, A. & Dommenget, D. An evaluation of ENSO dynamics in CMIP simulations in the framework of the recharge oscillator model. Clim. Dyn. 51, 1753–1771 (2018).

    Google Scholar 

  43. 43.

    Jin, F. F. An equatorial ocean recharge paradigm for ENSO. Part I: conceptual model. J. Atmos. Sci. 54, 811–829 (1997).

    ADS  Google Scholar 

  44. 44.

    Johnson, N. C. & Xie, S.-P. Changes in the sea surface temperature threshold for tropical convection. Nat. Geosci. 3, 842–845 (2010).

    ADS  CAS  Google Scholar 

  45. 45.

    Sun, L., Alexander, M. & Deser, C. Evolution of the global coupled climate response to Arctic sea ice loss during 1990–2090 and its contribution to climate change. J. Clim. 31, 7823–7843 (2018).

    ADS  Google Scholar 

  46. 46.

    Hu, Z. Z. et al. Weakened interannual variability in the tropical Pacific Ocean since 2000. J. Clim. 26, 2601–2613 (2013).

    ADS  Google Scholar 

  47. 47.

    Guan, C. & McPhaden, M. J. Ocean processes affecting the twenty-first-century shift in ENSO SST variability. J. Clim. 29, 6861–6879 (2016).

    ADS  Google Scholar 

  48. 48.

    Hu, Z. Z., Kumar, A., Huang, B., Zhu, J. & Ren, H. L. Interdecadal variations of ENSO around 1999/2000. J. Meteorol. Res. 31, 73–81 (2017).

    Google Scholar 

  49. 49.

    Xu, K., Wang, W., Liu, B. & Zhu, C. Weakening of the El Niño amplitude since the late 1990s and its link to decadal change in the North Pacific climate. Int. J. Climatol. 39, 4125–4138 (2019).

    Google Scholar 

  50. 50.

    Philander, S. et al. Why the ITCZ is mostly north of the Equator. J. Clim. 9, 2958–2972 (1996).

    ADS  Google Scholar 

  51. 51.

    Xie, S.-P. in The Hadley Circulation: Present, Past and Future (eds Diaz, H. F. & Bradley, R. S.) 121–152 (Advances in Global Change Research Vol. 21, Kluwer Academic Publishers, 2005).

  52. 52.

    Zuo, H., Balmaseda, M. A. & Mogensen, K. The new eddy-permitting ORAP5 ocean reanalysis: description, evaluation and uncertainties in climate signals. Clim. Dyn. 49, 791–811 (2017).

    Google Scholar 

  53. 53.

    Rayner, N. A. et al. Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. 108, 4407 (2003).

    Google Scholar 

  54. 54.

    Huang, B. et al. Extended reconstructed sea surface temperature version 5 (ERSSTv5), upgrades, validations, and intercomparisons. J. Clim. 30, 8179–8205 (2017).

    ADS  Google Scholar 

  55. 55.

    Kalnay, E. et al. The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 77, 437–472 (1996).

    ADS  Google Scholar 

  56. 56.

    Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. https://doi.org/10.1002/qj.3803 (2020).

Download references

Acknowledgements

This work is supported by the Strategic Priority Research Program of Chinese Academy of Sciences, grant number XDB40000000. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modelling groups for producing and making available their model output. For CMIP, the US Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. We are grateful to various reanalysis groups for make the datasets available to us. PMEL contribution number 5077. W.C., B.N. and A.S. are supported by CSHOR and the Earth System and Climate Change Hub of the Australian Government’s National Environment Science Program. CSHOR is a joint research Centre for Southern Hemisphere Oceans Research between QNLM and CSIRO.

Author information

Affiliations

Authors

Contributions

W.C. conceived the study and wrote the initial manuscript. B.N. performed analysis of butterfly experiments and T.G. carried out analysis of CMIP5 and CMIP6. All authors contributed to interpreting results, discussion of the associated dynamics and improvement of this paper.

Corresponding authors

Correspondence to Wenju Cai or Lixin Wu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Tobias Bayr and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Ensemble averaged warming and ENSO change in the butterfly effect experiments.

The results from 40 butterfly effect experiments of the CESM-LE. a, Time series of global mean SST. The red curve represents the ensemble mean. b, Multi-experiment ensemble mean SST and wind stress difference between the last 50-year period (2050–2099) and the initial 50-year period (1920–1969). c, Same as in b but for ocean temperature along the Equator averaged between 5° S–5° N, showing an intensification of stratification along the equatorial upper ocean as in ref. 17, enhancing the ocean–atmosphere coupling. d, E-index variability in the initial 50-year period (blue bars) and the last 50-year period (red bars) for each experiment and the multi-experiment ensemble mean. The error bars represent 1 s.d. value of interexperiment E-index variability for the two periods, respectively.

Extended Data Fig. 2 ENSO properties in initial and future climate in the butterfly effect experiments.

The results from 40 butterfly effect experiments of the CESM-LE. a, Interexperiment relationship between the E-index and the C-index s.d. for the initial 50-year period (1920–1969). b, As in a, interexperiment relationship between the E-index variability and variability of the EP (5° S–5° N, 150° W–90° W) net heat flux (s.d.). c, d, The same as a, b, respectively, but for the future 50-year period (2050–2099). The blue stars and orange diamonds represent the ten experiments with the weakest and strongest initial E-index variability, respectively. Solid black circles represent other experiments. Experiments with a greater E-index variability systematically produce greater heat flux variability, and greater C-index variability as strong El Niño events lead to strong La Niña events. These properties are seen in both initial and future climate. Statistics (that is, correlation (R) and P value) of a linear fit (red solid line) are shown. The relationship is statistically significant above the 99% confidence level.

Extended Data Fig. 3 Nonlinear thermal damping aggregated over observational datasets and over 27 selected CMIP5 and CMIP6 models.

a, The observed monthly E-index versus normalised monthly surface net heat flux anomalies over the EP (5° S–5° N, 150° W–90° W) for the period of 1979–2017. b, As in a, but for the observed monthly C-index versus normalized monthly surface net heat flux anomalies over the CP (5° S–5° N, 160° E–150° W). Also shown are the quadratic fit (red solid line), for example, with E-index, in terms of NHF(t) = αT(E-index(t))2 + βTE-index(t) + γT, and corresponding thermodynamic nonlinear coefficient αT associated with EP ENSO and CP ENSO. Three SST reanalysis products and two atmospheric reanalyses52,53,54,55,56 are used here (see ‘Atmosphere thermal feedback and its nonlinearity’ in Methods). c, d, As in a, b, respectively, but for 27 selected CMIP5 and CMIP6 models (Extended Data Fig. 7).

Extended Data Fig. 4 ENSO thermal damping and cumulative ocean heat flux in the butterfly effect experiments.

The results from 40 butterfly effect experiments of the CESM-LE. a, b, Time series E-index (black) and net heat flux (red) in the EP (at 0°, 105° W) in experiments with strongest (run 14; a) and weakest (run 24; b) E-index variability in the initial (1920–1969) 50 years. c, EP cumulative net heat flux for the two experiments. Raw monthly net heat flux fields referenced to the 70-year (1850–1919) common monthly climatology before the butterfly effect is constructed first before accumulation. Greater cumulative heat loss by 1969 (end of the initial 50 years, indicated by the vertical black line) is generated owing to greater initial ENSO variability, reducing upper ocean warming due to the greenhouse effect.

Extended Data Fig. 5 ENSO thermal damping in the initial 100 years after the butterfly effect.

The results from 40 butterfly effect experiments of the CESM-LE. a, b, Relationship between the monthly E-index and the monthly net heat flux over the EP (5° S–5° N, 150° W–90° W; a), and between the monthly C-index and the monthly CP (5° S–5° N, 160° E–150° W; b), quadratically detrended net heat flux into ocean (W m−2) for the initial 100 years (1920–2019). Thermal damping takes heat out of the ocean during El Niño and puts heat into the ocean during La Niña, but because El Niño is greater in amplitude, after several ENSO events, net heat is taken out of the ocean. c, Interexperiment relationship showing that greater initial ENSO variability, hence a greater amount of cumulative ocean heat loss (at 0°, 105° W, indicated by the black cross in e, positive out of the ocean), is generated. Raw monthly net heat flux fields referenced to the 70-year (1850–1919) common monthly climatology before the butterfly effect is constructed first before accumulation. The cumulative oceanic heat loss can be surrogated by heat flux variability, as seen in d, showing that a greater cumulative heat loss is associated with greater heat flux variability. The blue stars and orange diamonds in c and d represent the ten experiments with the weakest and strongest initial E-index variability, respectively. Solid black circles represent other experiments. Correlation (R) and P value of a linear fit (red solid line) are shown. e, Interexperiment regression of 40 cumulative heat flux fields onto 40 values of E-index variability, both over the initial 100 years (1920–2019), showing an ENSO pattern of cumulative heat flux. In experiments in which the butterfly effect leads to greater initial ENSO variability, a greater cumulative ocean heat loss is generated along the Equator. Statistical significance above the 90% and 95% confidence levels based on a two-tailed Student’s t-test is indicated as black stippling and the green solid contour, respectively.

Extended Data Fig. 6 Difference between two groups of experiments with strong and weak initial E-index variability.

The results from 40 butterfly effect experiments of the CESM-LE. The difference indicates the impact due to different ENSO variability between the two groups. a, SST (°C) and wind stress (N m−2) difference between the ten experiments with the strongest E-index variability in the initial 50-year period (1920–1969) and the ten experiments with the weakest E-index variability for the same period (see Fig. 1b, orange diamonds and blue stars, respectively). b, Same as in a but for the upper 150-m ocean temperature (°C). Stippling indicates where the difference between the two ensembles is significant above the 90% confidence level, based on a two-tailed Student’s t-test, and the green solid contour represents the 95% confidence level.

Extended Data Fig. 7 Selection of CMIP5 and CMIP6 models.

The results are for 27 models, that is, 18 out of 34 CMIP5 models and 9 out of 15 CMIP6 models, that produce both a dynamic nonlinear coefficient αD less than −0.155, that is, greater than 50% of the observed amplitude17 and a thermodynamic nonlinear coefficient αT less than 0 with all but one simulating 50% of the observed value. In general, a greater αT is associated with a greater αD, with a correlation coefficient of 0.47 using all models. Selected models are marked by symbols filled in different colours, while non-selected models are indicated in black and grey without filling. Each ensemble member and the multimember ensemble mean for CESM-LE are shown in filled blue and red circles, respectively.

Extended Data Fig. 8 ENSO properties in CMIP5 and CMIP6 models.

a, b, Intermodel relationship of E-index variability with strong ENSO frequency and with EP (5° S–5° N, 150° W–90° W) heat flux variability, respectively, for the initial 50-year period (1900–1949). c, d, The same as a, b, respectively, but for the last 50-year period (2050–2099). Models with a higher E-index variability systematically generate a higher frequency of strong ENSO events and a stronger heat flux variability. The results are for 27 models, that is, 18 out of 34 CMIP5 models and 9 out of 15 CMIP6 models, that produce both a dynamic nonlinear coefficient αD of less than −0.155, that is, greater than 50% of the observed amplitude17 and a thermodynamic nonlinear coefficient αT less than 0. Strong ENSO frequency in a, c is defined as the total number per 50 years of strong El Niño events (E-index more than 1.5 s.d.) plus the total number of strong La Niña events (C-index less than −1.5 s.d.) in the ENSO peak season of December–February. Correlation (R) and P value of a linear fit (red solid line) are shown. In all scatter plots, the relationship is statistically significant above the 99% confidence level.

Extended Data Fig. 9 Evolution of ENSO variability in CMIP5 and CMIP6 models.

a, Intermodel relationship between the last (2050–2099) and initial (1900–1949) 50-year period in E-index variability, showing an inverse relationship that is statistically significant above the 99% confidence level, that is, models that generate a greater variability in the initial period systematically produce a smaller future variability. The results are for 27 models, that is, 18 out of 34 CMIP5 models and 9 out of 15 CMIP6 models, that produce both a dynamic nonlinear coefficient αD less than −0.155 (greater than 50% of the observed amplitude17) and a thermodynamic nonlinear coefficient αT less than 0. b, Evolution of E-index variability, measured by a 50-year running window, moving forward every year from 1900 and recorded at the initial year, for ten models with strong initial E-index variability (red box in a) and ten models with weak initial E-index variability (blue box in a). Solid red (blue) lines and red (blue) shadings indicate multimodel average and intermodel spread (1 s.d. value), respectively, of the ten models with strong (weak) initial E-index variability. ENSO variability in models with weaker initial variability shows a faster increase in response to greenhouse warming during the ensuing periods, with a final amplitude that exceeds that in models with stronger initial ENSO variability. Different running window lengths (for example, 40 years or 60 years) and different sample sizes of model groups for averaging (for instance, 7 or 13 models with largest initial E-index variability versus 7 or 13 models with smallest initial E-index variability, respectively) produce qualitatively similar behaviour.

Extended Data Fig. 10 Initial ENSO variability and its future change in all available runs of CMIP5 and CMIP6 models.

The results are for the 27 selected models, that is, 18 out of 34 CMIP5 models and 9 out of 15 CMIP6 models, that produce both a dynamic nonlinear coefficient αD less than −0.155 (greater than 50% of the observed amplitude17) and a thermodynamic nonlinear coefficient αT less than 0. a, Intermodel relationship between initial 50-year (1900–1949) E-index variability and its future change (2050–2099 minus 1900–1949) scaled by the corresponding global-mean SST warming in each model. b, As in a, but for C-index. Run numbers are indicated next to the model names (for example, r1, r2 and so on).

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cai, W., Ng, B., Geng, T. et al. Butterfly effect and a self-modulating El Niño response to global warming. Nature 585, 68–73 (2020). https://doi.org/10.1038/s41586-020-2641-x

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links