El Niño Southern Oscillation

The El Niño–Southern Oscillation (ENSO) is a natural phenomenon that arises from variations in winds and sea-surface temperatures over the tropical Pacific Ocean. Through changes in large-scale atmospheric circulation, it affects seasonal winds, rainfall and temperature patterns worldwide. ENSO recurs every 2 to 7 years, oscillating irregularly between three phases: El Niño, La Niña and neutral. The positive phase of the oscillation, El Niño events, usually lasts 9 to 12 months, reaching a peak intensity between November and January. The negative phase of the oscillation, La Niña events, also typically peaks during the northern hemisphere winter. Both El Niño and La Niña events can extend well beyond one year. 

ENSO is the primary mode of climate variability on seasonal timescales and a strong driver of predictability in seasonal forecasts. To aid in interpreting seasonal forecasts from C3S, this page describes some aspects of the effect of ENSO globally, and specifically on Europe in winter, which manifest via atmospheric circulation teleconnections.  

ENSO index and events

Traditionally, the monitoring of ENSO evolution and intensity has used several indices based on sea-surface temperature (SST) or atmospheric variables. Most common among these are Niño indices – anomalies in SST averaged over four regions of the Equatorial Pacific: Niño 1+2 (0° - 10°S, 80°W - 90° W), Niño 3 (5°N - 5°S, 90°W - 150°W), Niño 3.4 (5°N - 5°S, 120°W - 170°W) and Niño 4 (5°N - 5°S, 160°E - 150°W).  

Following other operational centres around the world, in June 2026 ECMWF and C3S introduced in their seasonal forecast charts a relative version of the traditional Niño indices; one of these – relative Niño3.4 - is primarily used in this analysis. ERA5 SST data is used to define the index and explore the impacts of it, as this is the reference dataset used in the C3S seasonal forecast products and verification. More details on the calculation of the relative Niño3.4 index, and how it differs from the traditional index, are provided in the expandable box below.

The traditional and relative Niño3.4 indices are calculated from ERA5 SST data to characterise ENSO events and their impacts. The relative version of the index is discussed in an  ECMWF blog post, and it is now also available in the C3S seasonal forecast charts, as described in the product descriptions page (where the equation and scaling factors required for the calculation are provided).

Compared to the traditional Niño3.4 index, the relative version is less sensitive to climate change trends in SST. This means that ENSO events earlier in the period generally have more positive anomalies, and events later in the period generally have more lower, or more negative, anomalies. Here we use the same scaling factors for each month as used in the ECMWF and C3S seasonal forecasts, and also use the 1991-2020 period as the reference for the anomalies. 

To define positive and negative events the upper and lower quartiles are computed over the period 1970-2025 for a three-month rolling mean. El Niño and La Niña events are defined as those years falling above and below the upper and lower quartiles, respectively. The corresponding list of events and anomalies can be seen by expanding the box in the next section.

ENSO teleconnections 

ENSO events are characterised by large-scale anomalies in sea temperatures and atmospheric winds taking hold in the equatorial Pacific. These are associated with shifts in large-scale patterns of atmospheric circulation which result in local effects as well as effects a very long way from the centres of action (for example, changes to the Walker circulation 'propagate' the effects in a zonal direction, across the tropics). However, away from the tropics other factors, independent of ENSO or, more generally, of perturbations in the tropics, can affect the climate at certain times of year. They can come into competition with the ENSO teleconnections, modulating the latter in a variety of ways. The 'average' behaviour - often determined by calculating statistics conditioned on a phase of the oscillation - is by no means a guarantee of outcomes in a single instance (year). This is particularly the case for the North Atlantic/European sector, where influences from polar regions or the stratosphere can play a significant role in both winter and summer. 

Global effects - temperature and precipitation monthly composites

Using the ENSO events selection approach outlined above, typical effects on temperature and precipitation are illustrated, by displaying the number of ENSO years falling into the upper or lower tercile category of the distribution of the respective variable over the selected period (as represented in ERA5). Colours are only shown when the number of years is statistically significant. This concept and methodology are similar to those used in Davey et al. 2014.

These charts can be used to identify regions where, according to this analysis method, there is a statistically significant ENSO teleconnection for temperature or precipitation for each calendar month. With the dropdown selectors the calendar month as well as the index (Niño3.4 or relative Niño3.4) can be chosen.

In the tables below ENSO events are indicated by the shading of the cells corresponding to the rolling three-month means printed in the cells: red indicates El Niño and blue La Niña (following the definition given above). The table for the traditional Niño3.4 index is shown on the left, and the table for the relative Niño3.4 is shown on the right.


   


ENSO impacts on European winters

ENSO events can have an effect on European winters, as discussed in the scientific literature. They tend to peak at the end of the calendar year; accordingly, impacts over Europe tend to be stronger in winter than at other times of year. Below, these effects, as present in ERA5, are introduced and discussed.  

The following plots are ERA5 mean sea level pressure (MSLP) composites for the northern hemisphere, constructed for El Niño and La Niña events (defined as upper and lower quartiles of DJF relative Niño3.4, as shown in the tables above). They show the average circulation patterns which have occurred during these events during winter (Dec-Feb), separately for the early (Nov-Dec) and the late part (Jan-Feb) of the season. The composites are the mean of the anomalies during the corresponding ‘events’, using the 1991-2020 period as the reference. 

Average effects

The composites above reveal differences in the average response in North Atlantic-European (NAE) atmospheric circulation between early winter and late winter. For early winter (November-December), the response shows a strong 'symmetry' (high spatial correlation) between El Niño and La Niña composites, but the amplitude of the signal is larger during the El Niño phase, when a strong negative anomaly is located over the North Atlantic and positive anomalies are found to the south and east. The late-winter (January-February) teleconnection is visibly different from the early-winter equivalent; the spatial correlation between El Niño and La Niña composites decreases. 

These circulation anomalies would be associated with anomalies in temperature and precipitation patterns (not shown). 

Diversity of outcomes

The average behaviour in the global composites shown above is not guaranteed to occur during each ENSO event.  

This can be illustrated by looking at past European winters during ENSO events, specifically at MSLP, near-surface air temperature (t2m), precipitation and 10m windspeed. The 'postage stamp' charts below show early- and late-winter (November-December and January-February) anomalies for El Niño/La Niña years. Here, El Niño/La Niña years are selected based on the December-February (DJF) relative Niño3.4 index, calculated using ERA5 data from 1970 to 2025 (as described above: using the upper/lower quartiles as thresholds, so these are the same years which contributed to the MSLP composites above). The anomalies are calculated using the 1991-2020 period as the reference.  

These 'postage stamp' charts highlight that, despite the ‘typical’ behaviour described above, there is a large diversity of outcomes in the North-Atlantic and Europe, in both atmospheric circulation and surface conditions. The plots are labelled by the year in which the aggregation falls, and with the Niño3.4 index for the period used to make the selection, Dec-Feb. Each set of charts can be expanded by clicking the links below. 

These MSLP anomalies, relative to the reference-period mean, are in units of hPa.

The relative Niño3.4 anomalies printed above each map are DJF means, which were used for the selection. The year denotes the year in which the aggregation falls, for example for the 1982/83 event ND is labelled as 1982 and JF as 1983.


El Niño years, ND anomaly, MSLP


La Niña years, ND anomaly, MSLP


El Niño years, JF anomaly, MSLP

La Niña years, JF anomaly, MSLP

These temperature anomalies, relative to the reference-period mean, are in degrees Celsius.

The relative Niño3.4 anomalies printed above each map are DJF means, which were used for the selection. The year denotes the year in which the aggregation falls, for example for the 1982/83 event ND is labelled as 1982 and JF as 1983.

El Niño years, ND anomaly, t2m

La Niña years, ND anomaly, t2m

El Niño years, JF anomaly, t2m

La Niña years, JF anomaly, t2m

These precipitation anomalies are percentages of the reference-period mean.

The relative Niño3.4 anomalies printed above each map are DJF means, which were used for the selection. The year denotes the year in which the aggregation falls, for example for the 1982/83 event ND is labelled as 1982 and JF as 1983.

El Niño years, ND anomaly, precip

La Niña years, ND anomaly, precip

El Niño years, JF anomaly, precip

La Niña years, JF anomaly, precip

These 10m windspeed anomalies are percentages of the reference-period mean.

The relative Niño3.4 anomalies printed above each map are DJF means, which were used for the selection. The year denotes the year in which the aggregation falls, for example for the 1982/83 event ND is labelled as 1982 and JF as 1983.

El Niño years, ND anomaly, windspeed

La Niña years, ND anomaly, windspeed

El Niño years, JF anomaly, windspeed


La Niña years, JF anomaly, windspeed

References