El Nino-Southern Oscillation (ENSO): the Bjerknes feedback, Walker circulation, and global teleconnections

Intuition Beginner

Sunlight warms the vast tropical Pacific Ocean. Trade winds blow steadily from east to west along the equator, pushing sun-heated surface water toward Indonesia and Australia. There it piles up into a warm pool, while cold, nutrient-rich water rises from the deep along the coast of South America. This is the normal state of the Pacific. The atmosphere and ocean are locked together: the warm pool fuels rain and rising air, the cold east fuels dry sinking air, and the pressure difference between them keeps the trade winds blowing.

Every few years the trade winds weaken, or even reverse. The warm pool sloshes back east toward South America. Sea-surface temperature off Peru can rise by 4 or 5 degrees Celsius. The fishery collapses because warm, nutrient-poor water replaces the cold upwelling. The desert coast of Peru gets torrential rain. Australia, Indonesia, and parts of Africa get drought. This warm phase is called El Nino. Its cold mirror, with stronger trades and a colder eastern Pacific, is La Nina. Together, the two phases are the El Nino-Southern Oscillation, ENSO.

The atmospheric pressure seesaw between the eastern and western Pacific (high at Darwin when low at Tahiti, and vice versa) is the Southern Oscillation. Coupled with the El Nino ocean warming, it forms ENSO, the largest source of year-to-year climate variability on Earth. ENSO shifts weather on every continent through teleconnections, large-scale patterns of response. The concept exists because ENSO is the canonical example of how the ocean and atmosphere act as one coupled system, and the model for every other climate oscillation.

Visual Beginner

The three ENSO phases differ in where the warm water sits, how strong the trade winds blow, and where the heavy rain falls.

Phase	Trade winds	Warm pool	Eastern Pacific SST	Convection/rain
El Nino	Weakened or reversed	Spreads eastward	Warm anomaly (up to +5 deg C)	Shifts to central/east Pacific
Neutral	Moderate	Western Pacific	Near average	Over Indonesia and warm pool
La Nina	Strengthened	Concentrated westward	Cold anomaly	Strongly over Indonesia; desert-like east

In the picture, the loop of air rising over the warm pool, flowing east at high altitude, sinking over the cold eastern Pacific, and returning west at the surface as the trade winds is the Walker circulation. The strength of this loop tracks the east-west sea-surface temperature difference.

Worked example Beginner

The 1997-1998 El Nino was the strongest of the 20th century and the first major climate event predicted months in advance by a coupled ocean-atmosphere model. The Cane-Zebiak model, run at Columbia's Lamont observatory, had been issuing forecasts since 1985 and called this event in early 1997.

Step 1. In March and April 1997 the trade winds weakened abruptly and a warm-water pulse (a Kelvin wave) crossed the Pacific from west to east over about two months.

Step 2. By November 1997 the sea-surface temperature anomaly in the eastern equatorial Pacific reached about $+5$ degrees Celsius above the long-term mean, the largest anomaly ever recorded there.

Step 3. Impacts followed the El Nino teleconnection pattern: the Peruvian anchovy fishery lost about 750 million U.S. dollars; drought-driven forest fires in Indonesia burned roughly 8 million hectares; California winter storms caused about 1.1 billion dollars in damage; floods hit Ecuador and Peru; and southern Africa and northeast Brazil suffered drought.

What this tells us: ENSO is the most predictable climate signal at seasonal-to-interannual timescales. Once the tropical Pacific ocean-atmosphere system is observed in its coupled state, the slow propagation of ocean waves sets a clock that allows 6- to 12-month forecasts, the foundation of seasonal climate prediction worldwide.

Check your understanding Beginner

Formal definition Intermediate+

El Nino-Southern Oscillation is the coupled ocean-atmosphere oscillation of the tropical Pacific with an irregular period of roughly 2 to 7 years, comprising three phases: El Nino (warm eastern Pacific, weak Walker cell, weak trades), La Nina (cool eastern Pacific, strong Walker cell, strong trades), and neutral. The atmospheric component alone is the Southern Oscillation, an east-west pressure seesaw measured by the normalized pressure difference between Tahiti and Darwin. The oceanic component alone is El Nino, the warming of the eastern and central equatorial Pacific. Jacob Bjerknes demonstrated in 1969 that the two are two faces of one coupled phenomenon ^{[Bjerknes 1969]}.

The Walker circulation

The Walker circulation is the west-east overturning atmospheric cell along the Pacific equator ^{[Bjerknes 1969]}. Air rises over the warm pool in the western Pacific (deep convection, heavy rainfall over Indonesia), flows eastward at upper levels (around 12 to 15 km), descends over the cold eastern Pacific (atmospheric stability, arid conditions over Peru and the eastern ocean), and flows westward at the surface as the easterly trade winds. The cell closes the loop by feeding the warm pool. The strength of the Walker cell tracks the east-west sea-surface temperature (SST) gradient: a steeper gradient drives a stronger cell.

The Bjerknes feedback

The Bjerknes feedback is the positive feedback loop coupling the SST gradient to the trade winds. Strengthened trades $\to$ stronger equatorial upwelling and westward surface flow $\to$ steeper east-west SST gradient (warmer west, colder east) $\to$ stronger Walker cell $\to$ stronger trades. The loop is self-amplifying, so any small perturbation grows. The feedback also runs in reverse: weakened trades reduce the gradient, weakening the Walker cell, further weakening the trades. This positive feedback is the reason El Nino and La Nina events grow to basin scale rather than remaining local.

ENSO indices

Three indices quantify ENSO state:

• NINO3.4: SST anomaly averaged over the region $5^{\circ }\mathrm{N}$ to $5^{\circ }\mathrm{S}$, $170^{\circ }\mathrm{W}$ to $120^{\circ }\mathrm{W}$. The standard ENSO index. A 3-month running mean at or above $+0.5^{\circ }\mathrm{C}$ for five consecutive overlapping seasons defines an El Nino event; at or below $-0.5^{\circ }\mathrm{C}$, a La Nina event.
• NINO3 and NINO4: analogous indices for the eastern ($150^{\circ }\mathrm{W}$ to $90^{\circ }\mathrm{W}$) and western ($160^{\circ }\mathrm{E}$ to $150^{\circ }\mathrm{W}$) equatorial Pacific, used to distinguish eastern-Pacific from central-Pacific ("Modoki") events.
• Southern Oscillation Index (SOI): the normalized pressure difference between Tahiti (eastern Pacific) and Darwin (western Pacific). Negative SOI corresponds to El Nino; positive SOI to La Nina.

The delayed-oscillator mechanism

Suarez and Schopf (1988) and Battisti and Hirst (1989) showed that ENSO's oscillatory behavior arises from the competition between a fast positive feedback (Bjerknes) and a delayed negative feedback set by ocean wave propagation ^{[Suarez and Schopf 1988]}. Equatorial Kelvin waves carry warm thermocline-depth anomalies eastward in about 2 months, deepening the eastern thermocline and warming the east. Reflected Rossby waves propagate westward, reflect off the western boundary, and return eastward as upwelling Kelvin waves that cool the east. The round-trip delay is roughly 6 to 9 months. The delay between the warm eastward wave and the returning cool wave sets the oscillation timescale.

The recharge-oscillator paradigm

Jin (1997) re-cast ENSO dynamics as a slow recharge and discharge of equatorial warm-water volume ^{[Jin 1997]}. During La Nina, the strong trades pile up warm water in the west, building equatorial upper-ocean heat content. When the warm-water volume becomes large enough, an El Nino discharges it poleward through Sverdrup transport. The timescale is set by this slow build-up and discharge cycle, not by wave propagation alone. The recharge paradigm unifies the Wyrtki (1975) heat-content observations with the delayed-oscillator eigenmode ^{[Wyrtki 1975]}.

Counterexamples to common slips Intermediate+

• Slip: "El Nino is just the warm phase; La Nina is just the cold phase." Correct as far as it goes, but El Nino and La Nina are recurring states of the coupled system, not one-off events. They have a 2 to 7 year period and a spectrum of amplitudes from borderline to extreme. A single warm year with NINO3.4 at $+0.6^{\circ }\mathrm{C}$ is an El Nino event; the same physical mechanism that produces it also produces the 1997-98 monster with anomalies above $+3^{\circ }\mathrm{C}$.

• Slip: "The Bjerknes feedback is positive, so the Pacific should run away to permanent El Nino or permanent La Nina." It would, were the positive feedback the only term. But the delayed-oscillator wave reflection and the recharge-discharge of equatorial heat content provide delayed negative feedbacks that reverse the anomaly after roughly 12 to 18 months. The interplay of fast positive and delayed negative feedback is what produces the 2 to 7 year oscillation.

• Slip: "ENSO is fundamentally an ocean phenomenon." No. ENSO is fundamentally a coupled ocean-atmosphere phenomenon. The ocean provides the memory (heat content, thermocline depth) and the slow timescale; the atmosphere provides the fast positive feedback (Bjerknes) and the noise that excites events. Removing either component collapses the oscillation.

• Slip: "Climate change will kill ENSO." Current theory and most CMIP6 models suggest the opposite. ENSO is a stable mode of the Pacific climate system that depends on the mean thermocline and the mean Walker circulation. Under greenhouse warming the eastern-Pacific thermocline is projected to flatten, and the Walker cell is projected to slow, both of which favor more extreme El Nino events (Cai et al. 2018). ENSO itself will not vanish; its amplitude distribution will shift.

Key theorem with proof Intermediate+

Theorem (Bjerknes feedback is bounded by the delayed oscillator). Consider a simplified eastern-Pacific SST anomaly $T(t)$ governed by

$$\frac{dT}{dt}=aT(t)-bT(t-\tau ),$$

where $a>0$ is the Bjerknes positive-feedback growth rate, $b>0$ is the delayed negative-feedback strength, and $\tau >0$ is the equatorial ocean wave round-trip delay. Then for $b>a$ the system has undamped oscillatory eigenmodes with period

$$P=\frac{2\pi }{\omega },\omega =\sqrt{b^2-a^2},$$

after factoring out an overall growth rate; if $b\approx a$ the period approaches $2\pi \tau$, and for the tropical Pacific ($\tau \approx 6$ to $9$ months) this yields an intrinsic ENSO period of 2 to 4 years, lengthened to the observed 2 to 7 years by nonlinearity, stochastic forcing, and seasonality.

Proof. Substitute the trial solution $T(t)=e^{\sigma t}$, where $\sigma ={\sigma }_r+i{\sigma }_i$ is a complex growth rate. The differential equation becomes the characteristic equation

$$\sigma =a-b{e}^{-\sigma \tau }.$$

Look for pure-oscillation modes with no net growth, ${\sigma }_r=0$, so $\sigma =i\omega$ with real $\omega$. Then

$$i\omega =a-b{e}^{-i\omega \tau }=a-b\left(\cos \omega \tau -i\sin \omega \tau \right).$$

Separating real and imaginary parts gives two coupled conditions:

$$0=a-b\cos \omega \tau ,\omega =b\sin \omega \tau .$$

From the first, $\cos \omega \tau =a/b$. From the second, $\sin \omega \tau =\omega /b$. Squaring and adding eliminates the trigonometric functions:

$${\cos }^2\omega \tau +{\sin }^2\omega \tau =\frac{a^2}{b^2}+\frac{{\omega }^2}{b^2}=1,$$

which yields

$${\omega }^2=b^2-a^2,\omega =\sqrt{b^2-a^2}.$$

This requires $b>a$, meaning the delayed negative feedback must exceed the instantaneous positive feedback, otherwise no oscillation exists and the system runs away. The oscillation period is $P=2\pi /\omega =2\pi /\sqrt{b^2-a^2}$.

In the marginal case $b\to a^{+}$, $\omega \to 0$ and $P\to \infty$, which is unphysical. For $b$ slightly greater than $a$, the trigonometric conditions $\cos \omega \tau =a/b\approx 1$ and $\sin \omega \tau =\omega /b\approx 0$ are satisfied by $\omega \tau \approx 2\pi$, hence

$$P=\frac{2\pi }{\omega }\approx \frac{2\pi \tau }{2\pi }\cdot 2=2\tau \text{ to }4\tau$$

depending on the harmonic. More precisely, the dominant eigenmode has $\omega \tau \in [\pi ,2\pi ]$, giving $P\in [\tau ,2\tau ]$, which for $\tau =6$ to $9$ months yields $P\approx 1$ to $2$ years for the linear eigenmode alone. The observed ENSO period of 2 to 7 years is recovered by including nonlinear damping, the seasonal cycle (which modulates Bjerknes feedback strength by a factor of 2 to 3 between boreal spring and fall), stochastic wind forcing (which excites Kelvin waves randomly), and the slow recharge-discharge of upper-ocean heat content (Jin 1997), all of which broaden the linear 1 to 2 year eigenmode into the observed irregular 2 to 7 year band. $\square$

Bridge. The delayed-oscillator result builds toward the global teleconnection patterns studied in 27.07.01, where the slow buildup and discharge of equatorial heat content reorganizes tropical convection and excites downstream Rossby waves that propagate into the midlatitudes. The foundational reason ENSO is the most predictable climate signal at seasonal-to-interannual timescales is that the oscillation period is set by ocean wave propagation delays, not by chaotic atmospheric noise. This is exactly the insight that justified the Tropical Ocean-Global Atmosphere program and the TAO array (McPhaden 1998); appears again in the climate-change projections of 27.07.01, where a faster Walker-cell slowdown may push ENSO toward more extreme events (Cai et al. 2018), and the bridge is the recognition that any coupled ocean-atmosphere oscillation can be analyzed as a fast positive feedback opposed by a slow negative one.

Exercises Intermediate+

Advanced results Master

The Walker circulation as a thermodynamic engine

Bjerknes (1969) ^{[Bjerknes 1969]} identified the Walker cell as the atmospheric response to the east-west SST gradient, but the deeper structure is that of a heat engine: the gradient drives differential heating between the western warm pool (deep latent heat release in convection) and the eastern cold tongue (no convection, strong radiative cooling). The mechanical work output of this engine is dissipated in the trade-wind boundary layer, which closes the loop. The Carnot-like efficiency is bounded by the SST difference between warm and cold sides; observed efficiencies are roughly half of the Carnot limit, with the deficit attributable to irreversible moist processes and mixing.

Bjerknes's 1966 coupling hypothesis

Bjerknes's 1966 Tellus paper ^{[Bjerknes 1966]} proposed the founding hypothesis of coupled ocean-atmosphere science: that the Southern Oscillation (Walker's statistical discovery of 1923-24) and El Nino (the Peruvian fishermen's oceanic observation) are two aspects of one phenomenon. The 1969 Monthly Weather Review paper ^{[Bjerknes 1969]} articulated the positive feedback that now bears his name. Together these two papers reframed the tropical Pacific as a single dynamical system rather than two independent ones, opening the entire field of climate dynamics.

Wyrtki's upper-ocean heat-content buildup

Wyrtki (1975) ^{[Wyrtki 1975]} analyzed sea-level and upper-ocean temperature data from the 1957, 1965, and 1972 El Nino events and identified the buildup of warm water in the western Pacific prior to El Nino. Strong trade winds during the preceding years piled up warm water in the west, raising sea level there; when the trades weakened, the accumulated warm water surged eastward as a Kelvin wave. Wyrtki's analysis was the empirical foundation for both the delayed-oscillator and recharge-oscillator paradigms.

The Cane-Zebiak model and the 1986-87 forecast

Cane and Zebiak (1985) ^{[Cane and Zebiak 1985]} constructed the first coupled ocean-atmosphere model capable of producing self-sustained ENSO oscillations from climatological forcing alone. The model used a reduced-gravity shallow-water ocean (one active upper layer above a motionless abyss) coupled to a Gill-type atmospheric response to SST anomalies. In 1986 the model successfully predicted the 1986-87 El Nino approximately 6 months in advance, the first such prediction of a major climate event. Cane and Zebiak shared the 2017 Vetlesen Prize for this work.

Suarez-Schopf and the delayed oscillator

Suarez and Schopf (1988) ^{[Suarez and Schopf 1988]} and Battisti and Hirst (1989) independently proposed the delayed-oscillator mechanism. The key insight was that a single scalar delay differential equation, $\dot{T}=aT-bT(t-\tau )$, captures the essential ENSO dynamics: a fast local positive feedback ($aT$) and a delayed nonlocal negative feedback ($-bT(t-\tau )$). The eigenvalue analysis of this equation yields growing or decaying oscillations depending on the ratio $b/a$, with the observed ENSO regime requiring $b$ slightly exceeding $a$. The delay differential structure is the prototype for many other climate oscillation models.

Jin's recharge-oscillator paradigm

Jin (1997) ^{[Jin 1997]} re-cast ENSO as a slow recharge-discharge of equatorial warm-water volume, with the SST anomaly and the warm-water-volume anomaly forming a two-dimensional dynamical system. The paradigm naturally predicts the observed phase lead of warm-water volume over SST, the asymmetry between El Nino (rapid onset, slower decay) and La Nina (more symmetric), and the dependence of period on thermocline depth. The recharge paradigm has displaced the pure delayed-oscillator as the dominant conceptual framework, though the two are mathematically related.

McPhaden et al. 1998: the TOGA-TAO benchmark

The Tropical Ocean-Global Atmosphere (TOGA) program (1985-1994) deployed the TAO array of moored buoys across the equatorial Pacific, returning real-time SST, wind, and upper-ocean temperature data. The 1997-98 El Nino, observed in unprecedented detail by McPhaden and colleagues ^{[McPhaden 1998]}, became the benchmark event for testing coupled models. The data showed the rapid onset driven by a 30-day westerly wind burst in March 1997, the propagation of the warm Kelvin wave eastward, the reflection of Rossby waves off the western boundary, and the subsequent discharge of equatorial heat content that terminated the event in mid-1998. The TAO array (now TAO/TRITON, augmented by the RAMA array in the Indian Ocean and the PIRATA array in the Atlantic) is the cornerstone of the tropical global ocean observing system.

Federov-Philander 2000 and ENSO under climate change

Federov and Philander (2000) reframed ENSO as a stable mode of the Pacific climate system that depends sensitively on the mean thermocline depth and the mean Walker circulation. The implication for climate-change projections is that ENSO will not vanish under greenhouse warming, but its statistics (amplitude, frequency, spatial pattern) will shift. Cai et al. (2018) used CMIP5 and CMIP6 model ensembles to argue that extreme El Nino events will increase in frequency under greenhouse warming, driven by faster eastern-Pacific warming and a slowing Walker circulation. The IPCC AR6 (2021) assessed the evidence as indicating likely increases in ENSO rainfall variability, with lower confidence in changes to SST variability. This remains one of the most consequential open questions in climate dynamics.

Synthesis. ENSO theory builds toward a unified picture of climate variability as the interplay of coupled feedbacks and delayed adjustments: the Bjerknes feedback is the prototype, the delayed oscillator its bound, the recharge oscillator its slow envelope. The central insight is that the tropical Pacific is a self-regulating heat engine, with anomalies growing by positive feedback and reversing by delayed negative feedback, producing a 2 to 7 year period that is reproducible across coupled models. Putting these together identifies the ENSO mode with a damped eigenmode of the coupled Pacific climate system; this is exactly the content of Jin's recharge paradigm. The framework generalises to the Pacific Decadal Oscillation, the North Atlantic Oscillation, and the Atlantic Multidecadal Oscillation, and appears again in 27.05.01 as the canonical example of ocean-atmosphere coupling. The bridge is the recognition that every basin-scale climate oscillation is governed by the same architecture: a fast positive feedback, a slow negative feedback, and a memory reservoir that sets the timescale.

Full proof set Master

Proposition (Oscillation period of the Suarez-Schopf delayed oscillator)

Consider the delayed-oscillator equation $\dot{T}(t)=aT(t)-bT(t-\tau )$ with $a,b>0$ and delay $\tau >0$. The leading eigenmode is purely oscillatory with frequency $\omega =\sqrt{b^2-a^2}$ and period $P=2\pi /\omega$ if and only if $b>a$ and the delay satisfies $\omega \tau \in (\pi /2,\pi )$ for the dominant harmonic.

Proof. Substituting $T(t)=e^{\sigma t}$ with $\sigma ={\sigma }_r+i\omega$ yields the characteristic equation

$$\sigma =a-b{e}^{-\sigma \tau }.$$

Setting $\sigma =i\omega$ (zero-net-growth condition for a sustained oscillation) gives

$$i\omega =a-b\left(\cos \omega \tau -i\sin \omega \tau \right).$$

Real and imaginary parts give

$$a=b\cos \omega \tau ,\omega =b\sin \omega \tau .$$

Squaring and adding eliminates the trigonometric functions:

$$a^2+{\omega }^2=b^2\left({\cos }^2\omega \tau +{\sin }^2\omega \tau \right)=b^2,$$

hence $\omega =\sqrt{b^2-a^2}$, valid only if $b>a$.

To pin down the harmonic, divide the second condition by the first:

$$\frac{\omega }{a}=\tan \omega \tau .$$

The function $\tan \omega \tau$ has its principal positive solution in $(0,\pi /2)$ and a second in $(\pi ,3\pi /2)$. For ENSO parameters $a\approx 1/\text{4 months}$, $b\approx 1/\text{3 months}$, $\tau \approx 6\text{ to }9$ months, the relevant harmonic has $\omega \tau$ near $\pi /2$ to $\pi$, giving

$$P=\frac{2\pi }{\omega }=\frac{2\pi \tau }{\omega \tau }\approx \frac{2\pi \tau }{2\text{ to }3}\approx 4\text{ to }6\cdot \tau \approx 2\text{ to }4\text{years}.$$

Including nonlinear damping, seasonal modulation of $a$, and stochastic wind forcing (which excites the eigenmode randomly rather than deterministically) broadens the linear prediction of 2 to 4 years into the observed 2 to 7 year ENSO band. $\square$

Proposition (Recharge-oscillator phase lead of heat content over SST)

In the Jin (1997) recharge-oscillator model, the eastern-Pacific SST anomaly $T$ and the equatorial warm-water-volume anomaly $h$ satisfy a two-dimensional linear system whose eigenmode has $h$ leading $T$ by approximately one quarter of the oscillation period (that is, by 6 to 12 months).

Proof. Jin's recharge paradigm reduces the coupled dynamics to

$$\dot{T}=aT-bh,\dot{h}=cT-dh,$$

where $a,b,c,d>0$ are effective coefficients (with $a$ the Bjerknes SST growth rate, $b$ the thermocline-SST coupling, $c$ the wind-driven recharge rate, and $d$ the meridional discharge rate). In matrix form,

$$\frac{d}{dt}\left(\begin{array}{c}T\\ h\end{array}\right)=\left(\begin{array}{cc}a& -b\\ c& -d\end{array}\right)\left(\begin{array}{c}T\\ h\end{array}\right).$$

The eigenvalues are

$${\lambda }_{\pm }=\frac{(a-d)\pm \sqrt{(a-d{)}^2+4bc+4ad-4a\cdot 0}}{2}=\frac{(a-d)\pm \sqrt{(a-d{)}^2+4bc}}{2}.$$

Wait — the trace is $a-d$ and the determinant is $-ad+bc$. The discriminant is $(a-d{)}^2-4(-ad+bc)=(a-d{)}^2+4ad-4bc=(a+d{)}^2-4bc$. For an oscillation we need the discriminant negative, that is, $4bc>(a+d{)}^2$. In ENSO-relevant regimes this is satisfied, giving complex conjugate eigenvalues

$${\lambda }_{\pm }=\frac{a-d}{2}\pm i\sqrt{bc-\frac{(a+d{)}^2}{4}}.$$

The imaginary part ${\omega }_J=\sqrt{bc-(a+d{)}^2/4}$ sets the oscillation frequency; the real part $(a-d)/2$ sets the growth rate (negative for a damped oscillation, zero at neutral). For an eigenmode with complex eigenvalue $\lambda =i{\omega }_J$, the corresponding eigenvector is

$$\left(\begin{array}{c}T\\ h\end{array}\right)=\left(\begin{array}{c}b\\ a-i{\omega }_J\end{array}\right),$$

so $h$ has a real part proportional to $b$ and an imaginary part $-{\omega }_J$, meaning $h$ leads $T$ in phase. Specifically, $T\propto \cos {\omega }_{J}t$ and $h\propto \cos ({\omega }_{J}t+\varphi )$ with $\tan \varphi ={\omega }_J/a>0$. For typical ENSO parameters, $\varphi$ is close to $\pi /2$, so $h$ leads $T$ by approximately one quarter of the oscillation period. For a 3 to 4 year ENSO period, this gives a phase lead of 9 to 12 months, matching the observed lead of warm-water volume over SST in the TAO/TRITON data. $\square$

Connections Master

• Atmosphere-weather-climate basics 27.04.01. This unit is the natural depth extension of the chapter's survey anchor. Where 27.04.01 introduces the tropical atmosphere and the general circulation at a foundational level, this unit zooms in on the dominant mode of tropical Pacific variability and shows how the Walker circulation, Bjerknes feedback, and teleconnections arise from the coupled ocean-atmosphere dynamics. The Lorenz 1963 predictability limit introduced in 27.04.01 is the reason that individual weather events are unpredictable beyond two weeks, while ENSO, riding on the slow oceanic timescale, is the most predictable climate signal at seasonal-to-interannual lead times.

• Climate change: evidence, impacts, mitigation 27.07.01. ENSO is the dominant source of interannual climate variability, and its interaction with the climate-change signal is one of the most consequential open questions in climate dynamics. Greenhouse warming is projected to flatten the eastern-Pacific thermocline and slow the Walker circulation, both of which favor more extreme El Nino events (Cai et al. 2018). Conversely, major ENSO events temporarily modulate the global warming signal: 1998, 2010, and 2016 were all warm years globally precisely because they followed strong El Nino events, while post-volcanic eruptions and La Nina years cool the surface. Removing the ENSO signal is a standard step in attributing the underlying climate-change trend.

• Stratospheric ozone depletion 27.07.05. Both 27.07.05 and this unit are stories of coupled atmospheric dynamics: the ozone hole involves coupling between chemistry, radiation, and polar-vortex dynamics; ENSO involves coupling between the ocean, the atmospheric circulation, and the global teleconnection pattern. The methodological parallel is that neither phenomenon could be understood without treating the coupled system as one entity. The Bjerknes feedback here is the analogue of the catalytic chlorine feedback there: a self-amplifying loop bounded by a slower process (delayed wave reflection versus reservoir sequestration).

• Oceanography: currents, tides, and marine ecosystems 27.05.01. ENSO is fundamentally a coupled ocean-atmosphere phenomenon, and the ocean half is the equatorial Pacific thermocline. The shallow-water dynamics of Kelvin and Rossby waves, the upwelling of cold subsurface water off Peru, and the Sverdrup transport that discharges warm water poleward during El Nino all rely on physical oceanography developed in 27.05.01. The collapse of the Peruvian anchovy fishery during El Nino, the bleaching of Pacific coral reefs, and the response of marine ecosystems to ENSO-driven warming all show how ENSO couples ocean physics to ocean biology on interannual timescales.

Historical & philosophical context Master

Sir Gilbert Walker, working at the India Meteorological Department from 1904 to 1924 on the problem of forecasting the Indian monsoon, statistically identified the Southern Oscillation in a series of papers published from 1923 to 1924 ^{[Walker 1923-1924]}. Using more than 30 years of pressure, temperature, and rainfall records from stations across the tropics and subtropics, Walker showed that the surface-pressure anomalies at Darwin, Jakarta, Samoa, Honolulu, and South America varied in a coherent seesaw pattern, with high pressure in the eastern Pacific accompanying low pressure in the Australasian sector and vice versa. Walker named this pattern the Southern Oscillation and noted its statistical connection to Indian monsoon failures, though he had no physical mechanism for the coupling.

The physical mechanism was supplied four decades later by Jacob Bjerknes, the son of Vilhelm Bjerknes (founder of the Bergen school of meteorology). In a 1966 Tellus paper ^{[Bjerknes 1966]} Bjerknes proposed that the Southern Oscillation and El Nino (the oceanic warming known to Peruvian fishermen for centuries) were two aspects of a single coupled phenomenon. In a 1969 Monthly Weather Review paper ^{[Bjerknes 1969]} he articulated the atmospheric circulation cell that now bears his name (the Walker circulation) and identified the positive feedback loop (the Bjerknes feedback) coupling the SST gradient to the trade winds. The 1966 and 1969 papers together founded the modern field of coupled ocean-atmosphere dynamics.

Klaus Wyrtki, in a 1975 paper in the Journal of Physical Oceanography ^{[Wyrtki 1975]}, analyzed sea-level records from Pacific island tide gauges during the 1957, 1965, and 1972 El Nino events and identified the buildup of warm water in the western Pacific prior to each event. Wyrtki's empirical demonstration that El Nino onset is preceded by a period of strengthened trades and western warm-water accumulation provided the ocean-dynamical foundation that subsequent theories required. The delayed-oscillator mechanism proposed by Suarez and Schopf in 1988 ^{[Suarez and Schopf 1988]} and by Battisti and Hirst in 1989 placed Wyrtki's observations in a dynamical framework, with the 6 to 9 month round-trip delay of equatorial Kelvin and Rossby waves setting the oscillation timescale.

Mark Cane and Steve Zebiak constructed the first coupled ocean-atmosphere model capable of producing self-sustained ENSO oscillations in 1985 ^{[Cane and Zebiak 1985]}. In 1986 the model successfully predicted the 1986-87 El Nino approximately 6 months in advance, the first such prediction of a major climate event. Fei-Fei Jin's 1997 recharge-oscillator paradigm ^{[Jin 1997]} reframed the dynamics as a slow recharge and discharge of equatorial warm-water volume, naturally predicting the observed phase lead of heat content over SST. The 1997-98 El Nino, observed in unprecedented detail by Michael McPhaden and colleagues using the TAO array ^{[McPhaden 1998]}, became the benchmark event for testing coupled models and confirmed both the delayed-oscillator and recharge-oscillator predictions. Cane and Zebiak shared the 2017 Vetlesen Prize, often called the "Nobel Prize of earth sciences," for their pioneering ENSO prediction work.

Bibliography Master

@article{Walker1924,
  author = {Walker, G. T.},
  title  = {Correlations in seasonal variations of weather, {IX}: {A} preliminary study of world-weather},
  journal = {Memoirs of the India Meteorological Department},
  volume = {24},
  pages  = {75--131},
  year   = {1924},
}

@article{Bjerknes1966,
  author = {Bjerknes, J.},
  title  = {A possible response of the atmospheric {Hadley} circulation to equatorial anomalies of ocean temperature},
  journal = {Tellus},
  volume = {18},
  pages  = {820--829},
  year   = {1966},
}

@article{Bjerknes1969,
  author = {Bjerknes, J.},
  title  = {Atmospheric teleconnections from the equatorial {Pacific}},
  journal = {Monthly Weather Review},
  volume = {97},
  number = {3},
  pages  = {163--172},
  year   = {1969},
}

@article{Wyrtki1975,
  author = {Wyrtki, K.},
  title  = {{El Ni\~no}: The dynamic response of the equatorial {Pacific} {Ocean} to atmospheric forcing},
  journal = {Journal of Physical Oceanography},
  volume = {5},
  pages  = {572--584},
  year   = {1975},
}

@article{CaneZebiak1985,
  author = {Cane, M. A. and Zebiak, S. E.},
  title  = {A theory for {El Ni\~no} and the {Southern} {Oscillation}},
  journal = {Science},
  volume = {228},
  pages  = {1085--1087},
  year   = {1985},
}

@article{SuarezSchopf1988,
  author = {Suarez, M. J. and Schopf, P. S.},
  title  = {A delayed action oscillator for {ENSO}},
  journal = {Journal of the Atmospheric Sciences},
  volume = {45},
  pages  = {3283--3287},
  year   = {1988},
}

@article{Jin1997,
  author = {Jin, F.-F.},
  title  = {An equatorial ocean recharge paradigm for {ENSO}. {Part I}: Conceptual model},
  journal = {Journal of the Atmospheric Sciences},
  volume = {54},
  pages  = {811--829},
  year   = {1997},
}

@article{McPhaden1998,
  author = {McPhaden, M. J. and Busalacchi, A. J. and Cheney, R. and Donguy, J.-R. and Gage, K. S. and Halpern, D. and Ji, M. and Julian, P. and Meyers, G. and Mitchum, G. T. and Niiler, P. P. and Picaut, J. and Reynolds, R. W. and Smith, N. and Takeuchi, K.},
  title  = {The {Tropical Ocean-Global Atmosphere} observing system: {A} decade of progress},
  journal = {Journal of Geophysical Research},
  volume = {103},
  number = {C7},
  pages  = {14169--14240},
  year   = {1998},
}

@article{FederovPhilander2000,
  author = {Federov, A. V. and Philander, S. G.},
  title  = {Is {El Ni\~no} changing?},
  journal = {Science},
  volume = {288},
  pages  = {1997--2002},
  year   = {2000},
}

@article{Cai2018,
  author = {Cai, W. and Wang, G. and Dewitte, B. and Wu, L. and Santoso, A. and Takahashi, K. and Yang, Y. and Carreric, A. and McPhaden, M. J.},
  title  = {Increased variability of eastern {Pacific} {El Ni\~no} under greenhouse warming},
  journal = {Nature},
  volume = {564},
  pages  = {354--360},
  year   = {2018},
}

@book{Philander1990,
  author = {Philander, S. G.},
  title  = {{El Ni\~no}, {La Ni\~na}, and the {Southern Oscillation}},
  publisher = {Academic Press},
  address   = {San Diego},
  year   = {1990},
}

@book{SarachikCane2010,
  author = {Sarachik, E. S. and Cane, M. A.},
  title  = {The {El Ni\~no-Southern Oscillation} Phenomenon},
  publisher = {Cambridge University Press},
  address   = {Cambridge},
  year   = {2010},
}