Last modified on Oct 13, 2026 15:28

Author: Franco Molteni

Contributors: Anca Brookshaw, Christopher Goddard, Eduardo Penabad

1. Defining El Niño

El Niño is a natural phenomenon whose main feature is the rise of sea surface temperature (SST) in the eastern equatorial Pacific above the climatological mean. This SST anomaly (i.e., departure from climate) is associated with significant changes over the whole tropical Indo-Pacific region, both in the strength and direction of near-surface winds and in the position and intensity of convective rainfall. Typically, an El Niño episode starts in late spring or early summer, reaches its peak around December and declines towards the end of boreal winter in the following year. This is usually followed by a reversal of the SST anomaly in the eastern Pacific, as well as in wind and rainfall anomalies, in what is referred to as a La Niña episode. The oscillation between these two phases occurs with irregular periods, with a full oscillation (i.e., the distance between episodes of the same sign) varying between 2 and 7 years. This cycle is referred to as the El Niño - Southern Oscillation (ENSO) phenomenon, since the variations of Pacific SST anomalies are closely related to changes in the sea-level pressure difference across the southern tropical Pacific (first documented in 1924 by Sir Gilbert Walker). 

To define objectively the occurrence and strength of ENSO episodes, scientists usually measure the SST anomaly in several pre-defined regions covering the eastern and central equatorial Pacific. These regions are shown in Fig. 1. The region closest to the South American coast (named Niño1+2 since it was defined as the combination of two smaller regions) is the one where the rise of SST anomalies is usually first detected at the beginning of El Niño episodes. Niño3 and Niño4 cover two adjacent regions in the eastern and central equatorial Pacific, respectively. While the Niño3 region was originally preferred to monitor the evolution of basin-wide SST anomalies, the anomaly in the Niño4 has a strong impact on the strength of the tropical rainfall response. The Niño3.4 region, covering part of the Niño3 and Niño4 regions, was defined with the goal of capturing SST variations close to those found on Niño3 but more representative of atmospheric impacts. Today, the SST anomaly in the Niño3.4 region is the primary index used to define the occurrence and strength of ENSO episodes: El Niño events are typically defined as periods when the 3-month running mean of the Niño3.4 SST anomaly (the so-called Oceanic Niño Index, ONI) exceed 0.5^oC for five consecutive months (other definitions are in use, which differ slightly in terms of amplitude and duration). 

The four panels of Fig. 2 show time series of 3-month mean SST anomalies in the four regions described above, computed from 1970 to mid-2026 as differences from the 1991-2020 climatology. The Niño3.4 anomaly is shown in the top-left panel.  It shows a moderate degree of asymmetry, with positive anomalies reaching higher peaks and negative anomalies usually persisting for longer periods. Before 2026, three major El Niño events had anomalies exceeding 2.5^oC, namely in winters 1982-83, 1997-98 and 2015-16, while for La Niña events only the 1988-89 episode had a comparable amplitude. Looking at the anomalies in the other three (non-overlapping) regions, Niño1+2 (top right) shows a strong asymmetry in the amplitude of positive and negative values. Here, the anomaly in the two strongest El Niño events exceeded 4^oC, while none of the La Niña events exceeded the -2^oC threshold. The time series of the Niño3 anomaly (bottom left) resembles closely the Niño3.4 anomaly, but with stronger positive extremes: the 1997-98 event exceeds 3.5^oC in the Niño3 index. Finally, the Niño4 anomaly (bottom right) has a smaller range than those in the other regions, with larger extreme values for La Niña events. However, the climatological SST in the Niño4 region has a larger value than in the other areas, which causes a stronger rainfall response to the same SST anomaly compared to regions further to the east. Indeed, the largest rainfall anomaly during ENSO episodes is usually found in this region, making it a major source of atmospheric teleconnections. Another notable fact is that, among the four Niño regions, this is the only one where a positive trend is evident in the anomaly time series for the last 56 years. 

2. Ocean-atmosphere interactions 

Although ENSO events are defined based on SST in a relatively narrow region, they affect atmospheric and oceanic variability on a much larger scale. To illustrate this, Fig. 3 shows the anomalies of SST, precipitation and lower tropospheric winds (at 850 hPa) in the December-to-February (DJF) season during two recent events, the El Niño of 2015-16 and the La Niña of 2010-11, covering the entire tropical and subtropical regions. Starting from El Niño (left-hand panels), the equatorial maximum of positive SST anomalies in the eastern equatorial Pacific is surrounded by weaker sub-tropical anomalies of the same sign on the eastern side of the ocean, while negative SST anomalies form a C-shaped (or 'horseshoe') pattern in the western Pacific and the subtropical regions around (or east of) the 180-degree meridian (the date line). Warm anomalies also cover the tropical Indian Ocean, while in the tropical Atlantic positive SST anomalies around 20°S and 30°S are separated by an area of negative anomalies north of the Equator. Precipitation anomalies over the Indo-Pacific Ocean are mostly in phase with the SST anomalies, but their maximum is shifted westward towards the date line, and negative rainfall anomalies appear around the Maritime Continent even in places where SST anomalies are weakly positive. Anomalies of low-level zonal wind are strong and positive in the central equatorial Pacific (indicating a weakening of the climatological trade winds), negative over Indonesia. This wind pattern allows warm surface waters to spread eastwards from the 'warm pool' region in the western tropical Pacific and creates anomalous low-level divergence and convergence which sustain the tripolar pattern of rainfall anomalies in the Indian Ocean, Maritime Continent and central Pacific. 

Looking at the La Niña event of 2010-11 (right-hand panels), all the main features of the anomalies described above can be found but with an opposite sign, indicating a fairly symmetric behaviour between warm and cold ENSO episodes. Some deviations from symmetry are worth noting: primarily, the westward shift in the central Pacific SST anomalies, with a much weaker signal along the South American coast (see comments on the Niño1+2 SST time series above). Other deviations from symmetry can be found in the intensity of anomalies in the Indian Ocean and northern tropical Atlantic. However, Fig. 3 shows two specific episodes, and the intensity and exact location of anomalies can vary substantially between events. In the case of El Niño events, one important distinction is between episodes showing the largest SST anomaly in the eastern Pacific or closer to the date line; the latter events are referred to as ‘Central Pacific’ or ‘Modoki’ events and are characterised by large anomalies in the Niño4 index. The features of specific events can be seen in the anomaly maps for individual El Niño and La Niña episodes in the expandable boxes below.

To describe the interactions between oceanic and atmospheric anomalies during ENSO events, it is useful to start describing how the climatological mean state over the tropical Pacific is maintained. Four main elements should be considered: the near-surface winds, the SST pattern, the tropical thermocline (i.e., the thin layer of water separating warm surface waters from cold deep waters, where temperature changes rapidly with depth), and the main areas of convective activity - all in the left panel of Fig. 4. The easterly near-surface winds (commonly called trade winds) which blow over the tropical Pacific have a double effect:  

• they induce a northward/southward flow in the subsurface waters respectively north and south of the Equator, creating horizontal divergence and upwelling of cold waters at the Equator, and leading to the creation of the 'equatorial cold tongue' in SST; 

• the surface wind stress forces warm surface waters to pile up on the western side of the tropical Pacific, so that the thermocline is deeper in the west and closer to the surface in the east Pacific; this thermocline tilt is reflected in the formation of the so-called 'SST warm pool'  at the western edge of the tropical Pacific and around the Maritime Continent, and the presence of relatively cold surface waters near the South American coast (also supported by further upwelling off the coast of Peru). 

Over the tropical oceans, the presence of convective activity is strongly conditioned by the SST below; as a result, during boreal winter the strongest convective rainfall is found over the warm pool, with two further bands of rainfall north and south of the equatorial cold waters: the Pacific Intertropical Convergence Zone (ITCZ) crossing the whole Pacific basin north of the Equator, and the South Pacific Convergence Zone (SPCZ) tilting in a south-easterly direction from the southern edge of warm pool. In turn, the low-level convergence associated with warm-pool convection reinforces the easterly trade winds over the western and central Pacific, while causing westerly near-surface winds over the equatorial Indian Ocean. Therefore, a positive feedback exists between the strength of the easterly winds over the Pacific, the east-to-west gradient of SST, and the near-surface convergence associated with the areas of convective activity; this feedback is referred to as the Bjerknes feedback. 

During an El Niño event (see central panel in Fig.4), the trade winds in the central tropical Pacific weaken, allowing warm water to propagate from west to east and reducing the longitudinal tilt of the thermocline. The positive SST anomalies which develop in the eastern and central Pacific cause an eastward shift of the convective activity, with positive rainfall anomalies occurring around the 180-degree meridian and drier-than-average conditions prevailing over Indonesia and northern Australia. Consistently with the Bjerknes feedback, the shift of the convective activity modifies the near-surface winds in a way to sustain the weakening of trade winds on the western side of the basin.  

In La Niña events (right panel of Fig.4), the opposite events occur, reinforcing the SST and rainfall differences between the eastern and western parts of the tropical Pacific. The trade winds get stronger, increasing the equatorial upwelling and pushing warm waters (and the associated convection) further to the west. The thermocline gets shallower in the central Pacific, resulting in negative SST anomalies which usually reach their maximum amplitude closer to the date line than the maximum positive anomalies in a typical El Niño event. 

 Changes in the tropical atmospheric circulation during ENSO events, however, are not limited to the Pacific sector. The shift in the location of the convective activity in the western and central Pacific modifies the location of areas of ascending and descending motion across the whole tropical belt, altering the so-called Walker circulation which connects the main convection centres along the equatorial belt. Changes in the Walker cells propagate the impact of El Niño and La Niña events outside the Pacific Ocean and are responsible for creating significant rainfall anomalies across the tropical continents. We’ll examine these in more detail in the following section. 

3. ENSO impacts in the tropics and extratropics

The changes in the Walker circulation induced by ENSO events alter the location and intensity of rainfall across most of the tropical and subtropical belt. Fig. 5 provides a measure of such impacts by showing regions where El Niño and La Niña events increase the frequency of rainfall anomalies being either above the upper tercile or below the lower tercile of the climatological distribution (computed from 56 years, 1970 to 2025).  

Since the intensity of ENSO anomalies grows from boreal summer to winter, let’s first look at the frequency maps for the June-to-August season (JJA, top row in Fig. 5). Apart from the increase in rainfall over the tropical Pacific, the strongest impact in the JJA rainfall is seen over Indonesia and eastern Australia, where dry conditions prevail during El Niño events and increased rainfall occurs during La Niña events. Regions where this kind of response also occurs, although with some asymmetry between the phases of the ENSO cycles, are the north-eastern part of South America, the Sahel region and India, although some of these are less evident in the maps shown.  

Moving to the September-to-November season (SON, central row), a dipolar structure of rainfall anomalies emerges over the tropical Indian Ocean and the adjacent continents. During El Niño events, dry conditions over Indonesia and Australia intensify and spread to the eastern side of the Indian Ocean, while increased rainfall occurs over the western Indian Ocean and tropical east Africa. Opposite conditions occur during La Niña seasons. To a large extent, these anomalies are amplified by the response of the Indian Ocean to the surface wind changes induced by ENSO, which often takes the form of an SST dipole between the western and eastern side of the Indian Ocean (see a further discussion on the role of the Indian Ocean Dipole, or IOD, below).  

In the December-to-February season (DJF, bottom row), the dry conditions over the northern part of South America intensify (and remain strong over north-east Brazil also in the following months, not shown). A significant rainfall reduction also occurs over southern Africa; in both these regions, the opposite response is found in La Niña events. Conversely, wet conditions during El Niño and dry conditions during La Niña are found over south-eastern United States. 

A full sequence of 'impact maps' for each month of the year, and for both precipitation and near-surface air temperature, can be found on the page: ENSO impacts (based on ERA5 reanalysis).

The response to ENSO events is not limited to the tropics but may also extend to the extratropics through propagation of large-scale Rossby waves; these connections between climate anomalies in regions distant from one another are called teleconnections in the scientific literature. The changes in the location and intensity of tropical convective activity modify the location and amount of diabatic heating released into the atmosphere, and the associated wind divergence/convergence in the upper troposphere. The interaction of the anomalous divergent wind with the climatological flow modifies the sources of Rossby waves which propagate from the tropics to the extratropics. In turn, these anomalous waves interact with the planetary waves originated in the extratropics and, in the northern hemisphere, can also modify the amount of heat fluxes into the polar stratosphere.

The result is that, although ENSO episodes are able to affect the extratropical circulation, the exact pattern of the anomalous response is highly variable from year to year, and therefore more difficult to detect through a statistical analysis. While the extratropical response over the north Pacific in boreal winter is fairly robust and reproducible, over the North Atlantic and European (NAE) sector the cold-season response changes significantly between late autumn/early winter and the later part of winter. This is illustrated in Fig. 6, which shows composites of mean-sea-level pressure (MSLP) anomalies in El Niño and La Niña events during the 1970-2025 period, for November-December (ND) and January-February (JF). Over the north Pacific, we find negative MSLP anomalies in both sub-seasons during El Niño events, which intensify and shift to the east in late winter; this corresponds to an intensification of the Aleutian low-pressure centre. The opposite impact is found during La Niña event, although with a stronger amplification of the anomalies between ND and JF.

On the NAE sector much stronger differences are found between the early- and late-winter response. In ND, the North Atlantic response to positive and negative ENSO events is fairly strong and anti-symmetric, with a negative pressure anomaly west of the British Isles during El Niño and a positive anomaly during La Niña events. As a results, the westerly flow over western Europe is intensified in the El Niño response, weakened (or even reversed) during La Niña events. Instead, during JF, the El Niño response on the western side of the north Atlantic looks like the negative phase of the North Atlantic Oscillation (NAO), a condition that shifts the Atlantic storm track south of its climatological position. Negative pressure anomalies cover north-eastern Europe. Over the Atlantic, the MSLP late-winter response to La Niña events is not simply the opposite of the El Niño response but still produces a northward shift of the Atlantic storm track, leading to dry conditions over the Mediterranean region. 

The impacts on near-surface air temperature over Europe are mainly the result of changes in the westerly flow: in the presence of a strong surface temperature difference between the warm eastern Atlantic and the cold eastern Europe, increased westerly winds bring relatively warmer air into the European continent, and vice versa. During El Niño events, warm anomalies cover most of western and central Europe in ND, while in JF they are confined to the eastern part of the continent. Cold anomalies prevail during La Niña events, again being stronger and more widespread in early winter. 

4. Combined teleconnections from multiple tropical sources 

Although the shift in tropical convection from to the 'warm pool' to the central Pacific is the largest source of diabatic heating anomalies during ENSO events, and therefore a major driver of the ENSO teleconnections, the full extratropical response is also determined by Rossby waves originated by heating anomalies in other parts of the tropics, which in turn are caused by the changes in the Walker circulation mentioned above. Among these secondary sources of teleconnection, a particularly important one for the autumn-to-winter response over the north Atlantic and Europe is the teleconnection from rainfall anomalies in the Indian Ocean.

The climatological distribution of SST in the Indian Ocean also shows a gradient from east to west, but with reversed sign with respect to the Pacific Ocean: namely, waters are warmer on the eastern side of the basin, being connected to the west Pacific warm pool by the Indonesian Throughflow. During the boreal cold season, the intense convection over the Maritime Continent drives near-surface westerly winds across the equatorial Indian Ocean, pushing warm surface waters eastwards. When, during the development of an El Niño event, convection decreases over the Indonesian region, these westerly winds decrease or even reverse locally, allowing the upwelling of cold water west of Sumatra and the rise of positive SST anomaly on the western side of the Indian Ocean, near the African coast. The opposite phenomenon occurs when the convection over the Maritime Continent intensifies, as it usually happens during a La Niña event.  

This pattern of SST anomalies is referred to as the Indian Ocean Dipole (IOD); it usually develops in late summer, reaches its peak during boreal autumn and decreases after December, when the maximum of SST and convective activity moves from the Equator towards southern latitudes, and radiative effects tend to invert the sign of SST anomalies in the eastern Indian Ocean. The connection between ENSO and IOD variability is responsible for the dipolar pattern of rainfall anomalies seen in the rainfall response to ENSO in SON (central panels in Fig. 5). The dipolar pattern of rainfall anomalies usually persists longer than the SST dipole, and is itself an important source of teleconnections into the extratropics.  

The IOD amplitude is quantified by the difference of SST anomalies in two regions of the Indian Ocean: a western region close to Africa (50°E-70°E, 10°S-10°N), and an eastern region west of Sumatra (90°E-110°E, 10°S-0°N). Fig. 7 compares 3-month-mean time series of the El Niño3.4 index with those of the IOD index (top panel) and of the SST anomalies in the two individual poles of the dipole. The variation of the (contemporaneous) correlation between these indices across the seasonal cycle is also shown in a colour-coded table.  

The time series in the top panel show a tendency for positive IOD and El Niño events to occur at the same time (see e.g. the strong IOD events in autumn 1997 and 2023). However, some El Niño episodes were only accompanied by weak or moderate IOD anomalies, while a strong, positive IOD event occurred in autumn 2019 in near-neutral ENSO conditions.  The correlation table indicates that, while a positive correlation between the Niño3.4 and western Indian Ocean SST exists throughout the year, the anti-correlation of Niño3.4 with the eastern Indian Ocean SST only occurs in boreal summer and autumn, when conditions allow the emergence of opposite SST anomalies on the two sides of the Indian Ocean.  

Once a dipolar rainfall anomaly pattern is established over the Indian Ocean as a response to the IOD SST anomalies, this acts as a further source of teleconnection to the northern extratropics through the interaction of anomalous upper-level winds with the South Asian jet. During early winter (typically in November and December) the Rossby wave train emerging from this region propagates through the Pacific and reaches the north Atlantic in phase with the waves originated in the central Pacific. A schematic of this early-winter interaction, as presented in a recent study [ref. 4.1] is shown in Fig. 8. The result of the wave superposition is to reinforce the low-pressure signal over the north Atlantic, so that the total response to Pacific and Indian Ocean anomalies resembles the positive phase of the NAO (although with a southward shift, which actually makes it a combination of NAO and eastern Atlantic patterns). 

5. ENSO and global-mean temperature

So far, we have discussed the ENSO variability and its impacts with respect to climatological mean conditions, using the most recent decades as a reference period to define the climatology. However, we know that the world climate is changing, mainly as a result of the increasing concentration of gases (such as CO2 and methane) which contribute to the greenhouse effect. Therefore, in recent times a growing attention has been paid to the relationship between ENSO variability and global warming. Here, we start from two straightforward, interrelated questions: does ENSO variability affect global-mean temperature, and does the increase in global-mean temperature affect the definition of ENSO events? 

 To answer these questions, let’s look at the time series of global-mean SST variability since 1950, as shown by anomalies with respect to the 1991-2020 climate mean. This is shown in the top panel of Fig. 9, together with the global-mean anomalies of surface air temperature (SAT) over sea. From 1980 onwards, these two timeseries are almost identical, while global-mean SST appears to be slightly warmer than global-mean SAT up to 1975. As expected, the variability in the two timeseries is strongly correlated, in terms of both the overall warming trend (hardly visible before 1975 and then growing after 1980) and the interannual fluctuations due to natural phenomena. The increase of global-mean SST in the past 50 years is approximately of 0.8^oC, while the difference between consecutive minima and maxima on the interannual scale ranges from 0.15^oC to 0.3^oC. Therefore, on average in the last 50 years, interannual phenomena account for global SST variations which are about 20 to 40% as large as the long-term warming. 

The bottom panel of Fig. 9 compares the time series of global-mean SAT over sea and land; here, the amplitude of the series has been multiplied by the fraction of sea and land respectively, so that the result gives the contribution of each term to the total global-mean SAT. It is worth noting that the two curves are strongly correlated, even on the interannual time scale, implying that heat transferred from ocean to the atmosphere is a major contributor to SAT anomalies over land. 

Comparing the time series in Fig. 9 with the Niño3.4 index (as shown in Fig. 2), most of the interannual variations appear related to ENSO events, with El Niño events leading to increasing global temperature, and the opposite occurring during La Niñas. As an example, the arrows in the top panel point to local maxima in SST and SAT anomalies corresponding to the El Niño events of 1972/73, 1982/83, 1991/92, 1997/98, 2009/10, 2015/16 and 2023/24. With positive SST anomalies prevailing in the tropical belt during El Niño events, surface fluxes transfer a larger-than-average amount of heat and moisture to the atmosphere, also increasing the greenhouse effect due to water vapour. 

However, interannual variations in global SAT are not only due to ENSO, even though variability in the central and eastern tropical Pacific (including decadal phenomena such as Interdecadal Pacific Oscillation) has been shown to be the major driver of fluctuations around the long-term trend [ref  5.1]. During boreal winter, a large amount of heat is transferred from the ocean to the atmosphere in the regions around the western boundary currents on the northwestern edge of the Atlantic and Pacific oceans. Here, strong surface heat fluxes are driven by the large temperature difference between the continental air advected by north-westerly winds and the underlying ocean temperature. Changes in the atmospheric circulation in these regions can significantly affect the amount of these fluxes, which are reflected in large SAT variations over the northern continents [ref 5.2, 5.3]. In addition, global temperatures are affected by major volcanic eruptions, which decrease the amount of solar radiation reaching the surface by injecting vast amounts of aerosols in the stratosphere (see for example the dip in global SAT following the eruption of Mount Pinatubo in June 1991). 

The superposition of a long-term warming trend due to changing concentration of greenhouse gases (GHG) and interannual SAT variations induced by natural phenomena poses a further question. As we have seen, the increase in global-mean SST in the last 50 years is of the order of 0.8^oC, while the standard deviation of the Niño3.4 index is close to 1^oC. Therefore, if we measure the amplitude of ENSO events over a multi-decadal period with respect to a fixed climatology, positive Niño3.4 anomalies corresponding to El Niño events tend to be stronger in recent years because of the global warming trend; the opposite occurs for La Niña events.  

To provide a fairer comparison between the amplitude of ENSO events across decades, relative ENSO indices have been proposed, in which the difference is taken between the SST anomaly in one of the four Niño regions shown in Fig. 1, and the SST anomaly averaged across the whole tropical belt from 20^oS to 20^oN (also shown in yellow in Fig. 1). The latter anomaly is taken an indicator of the global warming trend in the tropics, even though some natural variability is found even in such a large-scale index. A comparison between the uncorrected and the relative Niño3.4 index, averaged over three months, and their difference is shown in Fig. 10. Further details about the definition of the relative index used are available in an ECMWF blog post, and the C3S seasonal forecast product descriptions.    

Looking first at the difference between the original and the relative index (bottom panel in Fig. 10), once can see that the values of the correction term range from -0.6^oC to 0.6^oC in individual years, with decadal means varying similarly to the global-mean SST value. The impact of the correction means that the 1982/83 event stands out as the strongest El Niño events in the last 56 years; conversely, the intensity of the recent 2023/24 event is reduced, while the multi-year La Niña event of 2020-2023 looks stronger. Relative versions of other Niño indices are shown in the expandable box below.

Click here to see the traditional and relative versions of the other Niño indices

6. Does climate change affect ENSO properties and impacts? 

The introduction of relative indices corrects the first-order effect of climate change on the estimation of the amplitude of ENSO SST anomalies, practically removing the warming trend from the time series of ENSO indices. However, there are other ways in which climate change can affect ENSO. Is ENSO variability, as measured by the standard deviation of (detrended) ENSO indices, increasing in a warming climate? Are regional ENSO impacts changing, as a result of a modified climatological state and/or of a change in teleconnection patterns? These questions can be posed both in relation to the past observed multi-decadal record and the future, by considering the differences between current climate and future climate scenarios. 

Starting from the question on the amplitude, the relative Niño3.4 time series shown in Fig. 10 does not suggest a significant increase in the ENSO amplitude since 1970, although it has been argued that such variability changes can only be significantly detected when comparing much longer periods, of the order of a century.  Although no evidence of a change in ENSO amplitude in future climate scenarios has been reported in the IPCC 6^th Assessment Report (AR6), some scientists who looked recently at historical and future scenario simulations run under the CMIP6 protocol have detected an increase in ENSO variability and frequency, at least for some emission scenarios [refs 6.1, 6.2].   

A greater consensus is found with regard to the strength of ENSO impacts, with the AR6 stating that 'it is very likely that ENSO rainfall variability, used for defining extreme El Niños and La Niñas, will increase significantly, regardless of amplitude changes in ENSO SST variability, by the second half of the 21st century' in most emission scenarios. This change is related to the overall increase in the strength of the hydrological cycle in a warmer climate. It has also been suggested that the impact of ENSO on global-mean SST will increase under greenhouse warming [ref 6.3].  

Unfortunately, assessing the significance of changes in ENSO teleconnections requires the analysis of very large records of events to obtain statistically significant results. Such long records can possibly be obtained from ensembles of multi-decadal model simulations, but comparing model results with observations during the historical period is hampered by the limits of the observational record (even in a 70-year record we only have less than 20 full ENSO cycles). It is interesting to note that one change in the observed ENSO impact which has been widely discussed in the literature is actually a decrease in the strength of the local rainfall response: namely, the decrease in the negative correlation between ENSO SST and the All India Rainfall (AIR) index, describing the variability of state-averaged rainfall during the South Asian monsoon season. 

Fig. 11 shows a scatter diagram of AIR data, as provided by the Indian Meteorological Department (source: https://www.data.gov.in/catalog/rainfall-india, accessed September 2026), against Niño3 SST anomalies from ERA5, with different colours for data before and after 1980 and linear regression fits to data in the two parts of the 76-year sample, as well as in the full sample. Indeed, the correlation changes from -0.60 in the 1940-1979 period to -0.48 from 1980 onwards. Although these statistics suggest a decrease in the strength of the relationship between ENSO and Indian rainfall (with El Niño events leading to below-average monsoon rain), a significance test on the difference between the two coefficients shows that it is likely that the difference originates from random sampling from the same data distribution. The same conclusion was reached in a study [ref 6.4], based on large ensembles of seasonal forecasting experiments for the years 1980-2011, which confirms that sampling uncertainty can be responsible for the differences in correlation over time. 

A similar situation is found with respect to inter-decadal changes of other ENSO teleconnections found in the observed record, although with different degrees of uncertainty. A recent study on multi-decadal variability of tropical rainfall [ref 6.5] noted that the correlation between January-February rainfall anomalies in the western Indian Ocean and in the Niño4 region has increased from 0.34 to 0.65 (in ERA5 data) when considering 30-year periods before and after 1980. Since rainfall anomalies in the western Indian Ocean have a positive correlation with the NAO index, while Niño4 rainfall has a negative correlation with NAO in this late-winter period, the increased connection between rainfall in the two tropical regions may explain why the late-winter ENSO teleconnection to the North Atlantic region shows a weaker negative NAO signature after 1980, because of the superposition of responses of opposite sign. When the significance of the correlation difference in rainfall data is tested, it turns out that we can attach a confidence level greater than 90%, but lower than 95%, to this result.  

A way to overcome the sampling problems arising from the observed record may come from using large ensembles of historical simulations or seasonal reforecasts, but this approach would require a skilful simulation of tropical and extratropical teleconnections and their drivers from the model (or models) used in such studies. As shown in multiple studies, while most climate models reproduce well the canonical, late-winter extratropical teleconnections of ENSO, the same cannot be said for the early-winter response or the impact of Indian Ocean anomalies [ref 6.6]. In this situation, and especially for the Atlantic-European region, assessing potential changes in teleconnections driving the extratropical response to ENSO in future climate scenarios remains a very arduous task. 

References

NB: The scientific literature on ENSO is huge, and most statements in this web page are supported by a large number of sources. Here, references are provided for some recent studies which cover relatively new (and therefore not so well established) aspects of research on ENSO variability in the current and future climates.

[4.1] Abid, M.A., F. Kucharski, F. Molteni et al., 2021: Separating the Indian and Pacific Ocean impacts on the Euro-Atlantic response to ENSO and its transition from early to late winter. J. Climate, 34, 1531-1548. doi.org/10.1175/JCLI-D-20-0075.1 

[5.1] Kosaka, Y., Xie, SP. (2016) The tropical Pacific as a key pacemaker of the variable rates of global warming. Nature Geosci  9, 669–673 (2016). https://doi.org/10.1038/ngeo2770 

[5.2] Molteni, F., R. Farneti, F. Kucharski, and T.N. Stockdale, 2017: Modulation of air-sea fluxes by extratropical planetary waves and its impact during the recent surface warming slowdown, Geophys. Res. Lett., 44, 1494–1502, doi:10.1002/2016GL072298 

[5.3] Yang, JC., Lin, X., Xie, SP. et al. Synchronized tropical Pacific and extratropical variability during the past three decades. Nat. Clim. Chang. 10, 422–427 (2020). https://doi.org/10.1038/s41558-020-0753-9

[6.1] Cai, W., Ng, B., Wang, G. et al. Increased ENSO sea surface temperature variability under four IPCC emission scenarios. Nat. Clim. Chang. 12, 228–231 (2022). https://doi.org/10.1038/s41558-022-01282-z 

[6.2] Coquereau, A., Sévellec, F., Huck, T., & Fedorov, A. V. (2025). Increase in ENSO frequency and intensity under 20th and 21st century warming: Insights from CMIP6 large ensembles. Geophysical Research Letters, 52, e2025GL116541. https://doi.org/10.1029/2025GL116541 

[6.3] Hong, SJ., Kim, GI., Shin, Y. et al. Stronger ENSO-induced global SST variability in a warming climate. Nat Commun 17, 4231 (2026). https://doi.org/10.1038/s41467-026-70140-9 

[6.4] Cash, B.A., R. Barimalala, R., J.L Kinter III et al., 2017: Sampling variability and the changing ENSO–monsoon relationship. Climate Dyn 48, 4071–4079, doi:10.1007/s00382-016-3320-3 

[6.5] Molteni, F. and A. Brookshaw, 2024: Multi-decadal variability of tropical rainfall: reconciling ECMWF re-analyses and GPCP data. ECMWF Technical Memorandum 922, 24 pp. doi.org/10.21957/5952e84bb7 

[6.6] Molteni F. and A. Brookshaw, 2023: Early- and late-winter ENSO teleconnections to the Euro-Atlantic region in state-of-the-art seasonal forecasting systems. Climate Dyn 61, 2673–2692. doi.org/10.1007/s00382-023-06698-7 

Related articles

• Page:
  C3S Seasonal Forecasts
• Page:
  Conversion table for accumulated variables (total precipitation/fluxes)
• Page:
  Climate oscillations and variability
• Page:
  C3S Seasonal Forecast known issues
• Page:
  ENSO impacts (based on ERA5 reanalysis)