The Heat Is Online

From NATURE 398, pp: 694-697 (April 22, 1999)

Increased El Nino frequency in a climate model forced by future greenhouse warming


Max-Planck-Institut fur Meteorologie and * Deutsches Klimarechenzentrum, Bundesstrasse 55, D-20146 Hamburg, Germany

The El Nino/Southern Oscillation (ENSO) phenomenon is the strongest natural interannual climate fluctuation1. ENSO originates in the tropical Pacific Ocean and has large effects on the ecology of the region, but it also influences the entire global climate system and affects the societies and economies of many countries2. ENSO can be understood as an irregular low-frequency oscillation between a warm (El Nino) and a cold (La Nina) state. The strong El Ninos of 1982/1983 and 1997/1998, along with the more frequent occurrences of El Ninos during the past few decades, raise the question of whether human-induced 'greenhouse' warming affects, or will affect, ENSO3. Several global climate models have been applied to transient greenhouse-gas-induced warming simulations to address this question4-6, but the results have been debated owing to the inability of the models to fully simulate ENSO (because of their coarse equatorial resolution)7. Here we present results from a global climate model with sufficient resolution in the tropics to adequately represent the narrow equatorial upwelling and low-frequency waves. When the model is forced by a realistic future scenario of increasing greenhouse-gas concentrations, more frequent El-Nino-like conditions and stronger cold events in the tropical Pacific Ocean result.


Our global climate model8,9 uses a meridional resolution of 0.5° in the tropics. The model, which is 'flux corrected', simulates an irregular ENSO cycle similar to the observed one, and the amplitude and dynamics of the simulated ENSO cycle are consistent with those derived from observations8,9. The simulated ENSO period is too short, however, and amounts to about two years, whereas observations indicate a main period of about four years. Our model successfully predicted the onset and decline of the 1997/1998 El Nino several months in advance10.

Here, two experiments were performed. The first experiment is a 300-year-long control integration with fixed present-day atmospheric concentrations of greenhouse gases. The second experiment is a transient greenhouse warming simulation in which the model was forced by increasing levels of greenhouse gases as observed (1860-1990) and according to IPCC scenario IS92a11 (1990-2100).

The changes in the mean state at the surface of the tropical Pacific Ocean, as derived from the transient greenhouse warming simulation, are reminiscent of the anomalous climate state observed during present-day El Nino conditions. The sea-surface-temperature (SST) trend pattern is characterized by strongest warming in the equatorial east Pacific, accompanied by westerly near-surface wind anomalies in the equatorial region to the west of the maximum warming and strong equatorward flow off the Equator (not shown). The associated trend in rainfall is rather similar to that simulated during present-day El Ninos (not shown).

There has been some discussion about the relative roles of different feedbacks involved in the time-mean response of the tropical Pacific climate system to greenhouse warming. On the one hand, it has been suggested that regional differences in the cloud-albedo feedback will lead to surface warming that is strongest in the equatorial east Pacific12. The argument is that the equatorial west Pacific is so warm that even modest additional warming would lead to a cloud shielding effect, with high cirrus clouds, reducing incoming solar radiation at the surface and inhibiting further warming13. This 'thermostat' would be less efficient in the eastern equatorial Pacific, so it would warm more than the western Pacific. This would lead to a slackening of the winds along the Equator and result in overall conditions very similar to those observed during El Ninos.

On the other hand, it has been argued that the strong equatorial upwelling in the eastern equatorial Pacific will weaken the warming in this region, so the strongest warming will occur in the western equatorial Pacific7. This would lead to stronger winds along the Equator, more equatorial upwelling and a net cooling in the eastern Pacific. This 'dynamical thermostat' will lead to overall conditions resembling those observed during La Ninas, eventually retarding global warming. The global climate models applied so far to greenhouse warming simulations could not adequately address this problem7 as they were poor at resolving the equatorial upwelling. Our climate model includes adequate representations of both types of feedback, and the balance of all physical processes leads to a simulated warming pattern that is strongest in the east, consistent with ref. 12.

As can be seen from the time evolution of the eastern equatorial SST in the transient greenhouse warming simulation, there is considerable interannual variability superimposed on the warming trend in the tropical Pacific (Fig. 2). Our model simulates an irregular ENSO cycle under enhanced greenhouse conditions, with a main period close to that derived from the control integration. However, the ENSO cycle evolves under different mean conditions to those currently found so we may expect the statistics of the ENSO cycle to change under enhanced greenhouse conditions. It can be seen from the SST time series that the level of the interannual variability increases strongly towards the end of the greenhouse warming simulation. This is also seen in the time series of the interannual SST standard deviations, which shows a strong increase towards the end of the transient greenhouse warming integration. The observations also show that the internannual variability has intensified during the past several decades, but this is well within the range of our control simulation.

The statistical significance of the increase in ENSO variability is an important issue to be addressed. In addition to our 300-year control integration, we can use the first 100 years of the scenario integration in the estimation of the natural variability as the external forcing is small up to 1960. We also performed two scenario integrations that included anthropogenic sulphate aerosols14, which started in 1860 and were integrated up to the year 2050 (not shown). Although these two runs are too short to study changes in the statistics of the interannual variability, their first 100 years can be used in the estimation of the noise, so we have 600 years in total. We did not find any period within the whole 600 years that exceeds a running standard deviation of 1.15 * C. The enhanced ENSO variability seen towards the end of our greenhouse warming integration is therefore highly significant.

To investigate further the changes in the ENSO statistics, we computed the frequency distributions of monthly SST anomalies. The distribution obtained from the first half of the transient integration is narrower than that obtained from the second half, so the year-to-year variability becomes more extreme under enhanced greenhouse conditions. Furthermore, although the distributions of the SST anomalies in the control integration (not shown) and during the first half of the transient integration are almost symmetric, the distribution for the second half of the transient integration is skewed: strong cold extremes become more frequent, while the statistics of the strong warm extremes do not change .

Which changes in the mean state lead to changes in the statistics of the interannual variability? First, the changes in the mean state near the surface could favour a reduction in ENSO-type variability as the zonal asymmetries across the equatorial Pacific are reduced15. Second, interactions between the air and the sea may be more energetic in a warmer climate, increasing the interannual variability. Third, changes in the vertical density structure of the ocean may alter the level of the interannual variability. The equatorial thermocline becomes stronger in response to greenhouse warming temperatures near the surface rise, but those at deeper ocean levels fall. This cooling at subsurface levels can be attributed to a greater inflow of cold waters in response to the intensification of the atmospheric Hadley Circulation, especially in the Southern Hemisphere. In order to gain further insight into the dynamics of the changes in the ENSO statistics, we computed atmospheric and oceanic sensitivities as functions of time (Fig. 4). The most important atmospheric forcing for equatorial oceans is the zonal wind stress component. We computed first the sensitivity of central equatorial zonal wind stress anomalies to the SST anomalies in the eastern equatorial Pacific averaged over the Nino-3 region (150* W-90* W, 5* N-5* S). No significant change was found. We then computed the sensitivity of Nino-3 SST anomalies to changes in the central equatorial zonal wind stress. This exhibited a significant increase towards the end of the transient greenhouse warming simulation, indicating that changes in the ocean dynamics arising from the strengthening thermocline are responsible for the enhanced interannual variability. This result is consistent with findings obtained from simpler coupled models16-18.

The simple models also predict a skew in the frequency distribution of the thermocline depth anomalies in the eastern equatorial Pacific (a quantity closely related to eastern equatorial SST anomalies) for a sharpening thermocline, with strong cold events becoming more frequent18. This is consistent with our greenhouse warming simulation. We therefore conclude that the most important change in the mean state of the tropical Pacific Ocean-atmosphere system affecting the ENSO statistics appears to be the strengthening of the equatorial thermocline.

The tropical Pacific climate system is thus predicted to undergo strong changes if emissions of greenhouse gases continue to increase. The climatic effects will be threefold. First, the mean climate in the tropical Pacific region will change towards a state corresponding to present-day El Nino conditions. It is therefore likely that events typical of El Nino will also become more frequent. Second, a stronger interannual variability will be superimposed on the changes in the mean state, so year-to-year variations may become more extreme under enhanced greenhouse conditions. Third, the interannual variability will be more strongly skewed, with strong cold events (relative to the warmer mean state) becoming more frequent. Although the model was successful in simulating and predicting ENSO, the ENSO response to greenhouse warming may depend on processes that are not well understood, such as cloud feedback, so we cannot exclude the possibility that the results are sensitive to the model formulation.

Received 21 December 1998;accepted 22 February 1999.



  1. Philander, S. G. H. El Nino, La Nina, and the Southern Oscillation (Academic, San Diego, 1990). Links Save Citation
  2. Glantz, M. H., Katz, R. W.& Nicholls, N. Teleconnections Linking Worldwide Climate Anomalies (Cambridge Univ. Press, 1991). Links Save Citation
  3. Trenberth, K. & Hoar, T. The 1990-1995 El Nino-Southern Oscillation event: Longest on record. Geophys. Res. Lett. 23, 57-60 (1996). Links Save Citation
  4. Knutson, T. R., Manabe, S.& Gu, D. Simulated ENSO in a global coupled ocean-atmosphere model: multidecadal amplitude modulation and CO2 sensitivity. J. Clim. 10, 138-161 (1997). Links Save Citation
  5. Tett, S. Simulation of El Nino-Southern Oscillation-like variability in a global AOGCM and its response to CO2 increase. J. Clim. 8, 1473-1502 (1995). Links Save Citation
  6. Meehl, G. A., Brantstator, G. W. & Washington, W. M. Tropical Pacific interannual variability and CO2 climate change. J. Clim. 6, 42-63 (1993). Links Save Citation
  7. Cane, M. A. et al. Twentieth-century sea surface temperature trends. Science 275, 957-960 (1997). Links Save Citation
  8. Roeckner, E., Oberhuber, J. M., Bacher, A., Christoph, M. & Kirchner, I. ENSO variability and atmospheric response in a global atmosphere-ocean GCM. Clim. Dynam. 12, 737-754 (1996). Links Save Citation
  9. Bacher, A., Oberhuber, J. M. & Roeckner, E. ENSO dynamics and seasonal cycle in the tropical Pacific as simulated by the ECHAM4/OPYC3 coupled general circulation model. Clim. Dynam. 14, 431-450 (1997). Links Save Citation
  10. Oberhuber, J. M., Roeckner, E., Christoph, M., Esch, M.& Latif, M. Predicting the '97 El Nino event with a global climate model. Geophys. Res. Lett. 25, 2273-2276 (1998). Links Save Citation
  11. Houghton, J. T., Callander, B. A. & Varney, S. K. V. (eds) IPCC Climate Change 1992. The Supplementary Report to the IPCC Scientific Assessment (Cambridge Univ. Press, 1992). Links Save Citation
  12. Meehl, G. A. & Washington, W. M. El Nino-like climate change in a model with increased atmospheric CO2 concentrations. Nature 382, 56-60 (1996). Links Save Citation
  13. Ramanathan, V. & Collins, W. Thermodynamic regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Nino. Nature 351, 27-32 (1991). Links Save Citation
  14. Roeckner, E., Bengtsson, L., Feichter, J., Lelieveld, J. & Rodhe, H. Transient climate change simulations with a coupled atmosphere-ocean GCM including the sulfur cycle. J. Clim. (in the press). Links Save Citation
  15. Battisti, D. S. & Hirst, A. C. Interannual variability in the tropical atmosphere/ocean system: Influence of the basic state, ocean geometry and non-linearity. J. Atmos. Sci. 46, 1687-1712 (1989). Links Save Citation
  16. Munnich, M., Cane, M. A. & Zebia, S. E. A study of self-excited oscillations of the tropical ocean-atmosphere system, Part II: Nonlinear cases. J. Atmos. Sci. 48, 1238-1248 (1991). Links Save Citation
  17. Tziperman, E., Stone, L., Cane, M. A. & Jarosh, H. El Nino chaos: overlapping of resonances between the seasonal cycle and the Pacific Ocean-Atmosphere Oscillator. Science 264, 72-74 (1994). Links Save Citation
  18. Stone, L., Saparin, P. I., Huppert, A. & Price, C. El Nino chaos: The potential role of noise and stochastic resonance on the ENSO cycle. Geophys. Res. Lett. 25, 175-178 (1998). Links Save Citation

Acknowledgements. We thank T. P. Barnett for discussion and M. Munnich for help with data processing. This work was sponsored by the German government under its programme 'Klimavariabilitat und Signalanalyse' and by the European Union through its 'SINTEX' programme. The climate model integrations were performed at the Deutsches Klimarechenzentrum.

Correspondence and requests for materials should be addressed to M.L. (e-mail:

Nature (c) Macmillan Publishers Ltd 1999 Registered No. 785998 England.