Changes in the tropical Pacific induced by strong El Niño and climate change
Figure 1a shows the MEM El Niño SST anomalies under present day. The anomalies closely resemble the pacemaker target SST pattern, with the canonical eastern Pacific structure of strong El Niño. In the tropical Pacific, strong El Niño events drive an eastward shift of the precipitation maximum, leading to precipitation anomalies of above 10 mm day−1 over the central and eastern equatorial Pacific (Fig. 1b), similar to the observed precipitation pattern during El Niño (Fig. S1). Those positive anomalies contrast with drier conditions off the equator around 10°N and in the South Pacific Convergence Zone.
There are very small differences in the El Niño SST anomalies in the future (Fig. 1c), indicating the pacemaker method works as intended (see “Methods” and Fig. S2). There are, however, significant differences in El Niño-driven precipitation in the future (Fig. 1d), with higher precipitation rates in the eastern (more wetting) and western (less drying) equatorial Pacific of up to 2 mm day−1 and negative anomalies (less wetting) of at least 2 mm day−1 over the central equatorial Pacific. We note that these differences do not fully match the projected change in El Niño-driven precipitation anomalies in the tropical Pacific suggested by the literature, consisting of an eastward shift of precipitation during El Niño under greenhouse38 forcing42. This might indicate an important role for El Niño-SST changes in the future El Niño-precipitation relationship seen in coupled climate models, or be specific to the effect of climate change on the precipitation response to strong El Niño events, which are generally not isolated in other studies. It is also possible the pacemaker set–up influences the associated precipitation response, for example because it constrains surface turbulent fluxes, but similar differences were found in an experiment with weaker flux restoring (not shown).
In terms of mean state changes, the tropical Pacific warms by around 2.5 K in the future (Fig. 1e), with the Niño4 (5°N-5°S, 160°E-150W) and Niño3 (5°N-5°S, 150°W-90°W) regions being 4.7% and 6.6% warmer, respectively, compared with the present day. Consequently, the zonal east-west equatorial Pacific SST gradient decreases by 19.5%, similar to the response in other CMIP6 models43. Coincident with the mean SST changes, there is a weakening of the Walker circulation, evidenced by a decrease in precipitation over the warm pool and an increase in precipitation over the Central and Eastern Pacific (Fig. 1f).
To understand how strong El Niño conditions would appear in the future compared to present day, we next consider the combined mean state plus El Niño anomalies (Fig. 1g and 1h). Since the El Niño SSTs are largely unchanged in the future, the total near-surface temperature signal mainly reflects the change in tropical Pacific mean state, whereas for the total precipitation anomaly, the substantial negative anomalies between 170°E-150°W seen as a difference in the response to El Niño in the future, expand the location of dry conditions further towards the central Pacific. Such a change might be expected to alter the teleconnections from El Niño, which is explored in the next section.
Teleconnections from strong El Niño events are modulated by climate change
In the present day, the modelled strong El Niño drives significant surface temperature impacts across the globe in DJF (Fig. 2a). Northern South America experiences warm anomalies reaching up to 3 K over the Amazon, as well as large parts of the African continent, with the exception of near Lake Victoria. Western Australia experiences hotter summers during strong El Niño, and south Asia also sees warm anomalies. There are some regions, though, where El Niño leads to cold anomalies in DJF, such as the southern and western US and south-east South America (SESA). Note that the EC-Earth3-CC model captures the amplitude and spatial location of the winter surface response to El Niño over northern South America, North America and Southern Africa similar to observations (Fig. S1).
Surface temperature (K, left) and precipitation (mm day−1, right) anomalies during El Niño in the present (a, b), differences in future and present El Niño anomalies (c, d), mean state changes (e, f) and changes in absolute values between future and present (g, h). Hatched areas are not statistically significant at the 95% confidence level.
Alongside surface temperature changes, strong El Niño events drive dry anomalies of up to 2 mm day−1 over the Amazon basin, although based on the model evaluation these anomalies may be overestimated (Fig. S1), and around 1 mm day−1 over Southern Africa. There is a dipole precipitation anomaly over the North Atlantic, likely driven by a southward shift of the North Atlantic jet stream and storm track (Fig. 2b), characteristic of the link between El Niño and the negative phase of the winter North Atlantic Oscillation (NAO)44. Over the Congo basin and the Horn of Africa, strong El Niño leads to weak but significant positive precipitation anomalies. Precipitation also increases by 2 mm day−1 over SESA during strong El Niño, leading to wetter than usual summers in the north of Argentina, Uruguay and Paraguay. EC-Earth3-CC effectively captures the sign and spatial location of positive precipitation anomalies during El Niño, but slightly underestimates its amplitude compared to observations (Fig. S1). In terms of the ability of the model to represent the complexity of the amazon climate, it has shown to correctly simulate the annual cycle of temperature and precipitation, and outperforms most CMIP6 models in the representation of the south American monsoon dynamics45.
In the future, the DJF surface temperature response to strong El Niño increases in several regions (Fig. 2c). Strong El Niño drives even hotter conditions in most of Africa (except in its southern part), and hotter summers in the south-west of the Amazon and Congo basins and eastern Australia, being the future response to El Niño in the latter stronger than in the present. Over north-east North America, the cold anomalies driven by strong El Niño are also enhanced in the future. On the other hand, mean DJF temperature responses driven by strong El Niño in the future weaken in other regions potentially due to shifts in atmospheric circulation. For example, cold anomalies over north-east Argentina, Paraguay and south Brazil extend more towards the north, following the northward shift of precipitation shown in Fig. 2d. Over southern Africa, the temperature and precipitation responses to strong El Niño weaken in our simulation, leading to milder summers.
In Australia, the influence of future climate change on the response to strong El Niño shows dipole anomalies consisting of higher precipitation rates over western Australia and drier conditions over the east in DJF, which is opposite to the projected precipitation trend in austral summer in the model (Fig. 2f). We note that other climate models tend to show an increase in Australian summer rainfall in future projections33, which would constructively interfere with the altered strong El Niño signal in western Australia. In a limited number of regions, the precipitation response to strong El Niño events weakens in the future compared to present day. For instance, in the present day strong El Niño drives dry anomalies over the Amazon basin (Fig. 2b), but this changes in the future, where the meteorological drought induced by strong El Niño becomes significantly weaker (Fig. 2d).
Figure 2e, f show the mean state surface temperature and precipitation changes between the future and present day. By the late 21st century, EC-Earth3-CC simulates an overall statistically significant increase in temperature across the globe, especially in the Northern Hemisphere (in agreement with Lee et al.33), where land areas experience a warming of above 4 K in DJF. In the Southern Hemisphere, regions like western Australia, southern Africa and the Amazon basin are warming faster than others. In terms of precipitation, besides the projected eastward shift in tropical Pacific precipitation resulting from a future weakening of the Walker circulation, there’s an overall increase in precipitation across the tropics, in agreement with CMIP6 projections33. Regions like Australia, North America and East Asia are expected to experience a slight increase precipitation in DJF, in agreement with multimodel projections, although the change in precipitation is not as evident as in temperature due to the higher variability of the former and the larger model uncertainty regarding dynamical changes as a response to climate change. The total anomalies during strong El Niño in the future compared to present day (Fig. 2g, h) show constructive interference and exacerbated El Niño-driven warm anomalies over eastern Australia, Northern South America and Central Africa during the austral summer months, which would lead to increased heat-related risks46,47. The mean warming under climate change masks any enhancement of the cooling effect of strong El Niño, such as over North America or East Asia. Nevertheless, over South America, Fig. 2h highlights the contribution of changes in strong El Niño impacts to the absolute response over the Amazon basin, which drives an increase in precipitation in DJF, opposing the mean drying under global warming.
Tropospheric circulation response to strong El Niño and its changes under future conditions
To understand future changes in teleconnections of strong El Niño, we explore tropospheric geopotential height at 500 hPa (z500) and zonal wind at 200 hPa (u200) anomalies in DJF. The extratropical teleconnections of El Niño depend on the background flow, specifically on the upper tropospheric zonal wind48, which is stronger in the Northern hemisphere in DJF due to a stronger subtropical jet. In the present day, strong El Niño generates positive z500 anomalies of up to 60 m off the coast of Japan (Fig. 3a), a deepening of the Aleutian low and barotropic Rossby wave trains that project onto the Pacific South American (PSA1) and Pacific North American (PNA) patterns. As opposed to moderate and weak El Niño events, strong episodes shift the PNA further east24, making these events unique and incomparable with El Niño events of lesser amplitude.
Eddy geopotential height at 500 hPa, (deviations from the zonal mean, m, filled contours) and zonal wind at 200 hPa anomalies (m s−1, black contours, where solid lines denote positive anomalies and dashed lines show negative anomalies) during El Niño in the present (a), differences in future and present El Niño anomalies (b), mean state changes (c) and changes in absolute values between future and present (d). Hatched areas are not statistically significant at the 95% confidence level.
Outside of the Pacific sector, strong El Niño drives negative geopotential anomalies over the south-east USA, which extend eastward over the subtropical Atlantic Ocean. Those negative anomalies contrast with positive anomalies spanning from the north-east of Canada to the Norwegian Sea. This corresponds to a negative phase of the NAO and a southward shift of the North Atlantic westerly jet stream, in agreement with the southward shift of precipitation anomalies shown in Fig. 2b.
Under future conditions, strong El Niño drives a weaker PNA response, with the simulations showing a relatively stronger Aleutian low in boreal winter with positive z500 anomalies of above 40 m (Fig. 3b). The eastward shift of z500 anomalies in the North Pacific and North America, in agreement with Johnson et al.49 and Geng et al.50 coincide with below normal temperatures shown in Fig. 2c. Similar conditions are shown over central and northern Europe, where strong El Niño events drive relatively low pressure anomalies associated with colder winters. There is a high pressure anomaly in the Caribbean region during strong El Niño events in the future, which translates into the drying trend shown in Fig. 2c. In the Southern hemisphere, strong El Niño triggers an anomalous wavetrain spanning the extratropics and subpolar regions and driving a dipole of positive and negative z500 anomalies in the west and eastern coasts of South America, respectively. The projected changes in El Niño-driven z500 over South America, do not resemble those shown in Johnson et al.49 possibly due to the weaker climate change signal in our simulations (compared to SSP5-8.5 used in their study) and the fact that changes in El Niño anomalies modulated by climate change are only expected to emerge at the end of the century. Also, as it is the case of the teleconnection towards north America, the fact that amplitude of the regional response to El Niño is not linearly related to the strength of the event makes it hard to compare with studies based on El Niño composites of all strengths as in Johnson et al.49 and McGregor et al.38.
In terms of mean state changes, Fig. 3c shows a deepening of the Aleutian low by 20 m in the future climatology, compared to present-day conditions. In the South Pacific Ocean, there is a strengthening of the zonal low pressure anomalies across SESA, and positive z500 anomalies in the Southern Ocean off the coast of South America. It is important to note that the amplitude of future changes in El Niño-driven z500 anomalies (Fig. 3b) is twice as large as the mean state changes (Fig. 3c), highlighting the significance of changes in ENSO teleconnections in the future relative to changes in the background flow. The absolute changes in z500 and zonal wind during strong El Niño events in the future are dominated by changes in El Niño teleconnections rather than by mean state changes, as suggested by Beverley et al.51.
Regional extreme temperature responses to strong future El Niño events
Building on the evidence in the previous sections showing regional signatures of the modulation of strong El Niño teleconnections under future climate change, we now investigate regional changes in meteorological extremes under strong El Niño, in North America, southern South America and southern Africa. Special attention is paid to shifts in the distribution of daily maximum temperature from the present to the future considering both cool and warm days (see Methods).
We first consider cool days in North America using a cool day index (see Methods) to understand changes to the distribution of cold anomalies driven by El Niño in the future. Note this index is percentile-based and therefore effectively removes the climatological shift in cool days driven by anthropogenic forcing.
In the present day, strong El Niño drives an increase of cool days in North America of up to 7% per winter, especially along the Rocky Mountains and central-eastern US (Fig. 4a), and a 6% reduction in cool days per winter over eastern Canada. There is a relative increase in cool days per winter during strong El Niño events over most of the USA in the future compared to present day (Fig. 4b), meaning an intensification of cold anomalies in the east and the appearance of cool anomalies over the northern US and Alaska. The mean climate change signal shifts the local climatological 10th percentile of maximum daily temperature towards higher values in the whole North America (See Fig. S3c), especially over the Northeast of the United States, Canada and Alaska, where the threshold for cool days increases by up to 8 °C. The changes in relative occurrence of cold days during strong El Niño could have societal implications, since a focus on adaptation to a warmer climate may reduce the focus on preparedness for cold days. At the same time, percentile-based indicators might not be as good a proxy for climate impacts on health in the future as they are in the present, i.e. cold days in the future will be less harmful for human health. This difference overrides the intensification of the cold El Niño anomalies expected at the end of the century, since the overall temperature will be higher due to the dominance of the global warming signal for the total temperature anomaly.
Anomalous cool days per winter (DJF) during strong El Niño in North America in (a) present conditions and (b) future – present conditions. Only grid cells corresponding to land surface areas are used in this plot. Areas where strong El Niño events do not drive statistically significant anomalies in the present-day simulations appear as hatched in subpanel (a). Hatching in subpanel (b) show non significant changes in the impacts of strong El Niño events between the future and present-day simulations.
Northern South America is one of the regions where strong El Niño events have the largest impacts on daily temperature extremes. From December to February, strong El Niño leads to an increase of warm days of up to 40%, especially around the Amazon basin in Brazil, Venezuela and the Pacific coast of Colombia and Ecuador. In the future, the striking impact of strong El Niño events in the region gets muted in the Amazon basin, where the percentage of warm days decreases up to 15%. In southwest Brazil and Bolivia, however, there is a significant strengthening of the El Niño signal, with up to 15% more warm days from December to February. Changes in extreme temperature impacts driven by El Niño in the future will play a key role in the frequency and intensity of forest fires in the Amazon region, fueled by a negative precipitation and soil moisture trend52.
According to Fig. 6, during strong El Niño, areas of Africa south of the equator experience a strong increase in warm days in summer, with the largest increases over Equatorial Guinea, Republic of the Congo and Angola, with up to 25% additional warm days per summer. In the future experiment, there is a relatively smaller increase in warm days per summer due to strong El Niño (approximately 7% less in some areas). In addition, there are enhanced warm anomalies towards the northern regions of lakes Tanganyika and Malawi, Tanzania and the Democratic Republic of the Congo, where the number of anomalous warm days per summer during strong El Niño is found to increase by up to 10%.
Figure S3g-i shows the shift of the 90th Tmax percentile towards higher values in southern Africa, especially over the west and southwest of the region, where the temperature threshold for a day to be considered warm will rise from 30 °C to 35 °C. The smaller increase in warm days in the future relative to present-day conditions is caused by a raise of the Tmax threshold for warm days. Nevertheless, in the future strong El Niño still leads to an increase in warm days in southern Africa, further exacerbating the underlying signal of climate change consisting of a shift towards higher temperatures in the region in austral summer. In the coastal areas of Mozambique and Tanzania there is a significant increase of up to 10% in the number of El Niño-driven warm days in summer, which constructively interfere with climate change leading to an exacerbation of warm summer days in the region. As shown in Figure S3i, the largest shift in the climatological warm day threshold is found in Namibia, Botswana and Mozambique.