8.2 El Niño & La Niña

The El Niño-Southern Oscillation (ENSO) is the dominant mode of interannual climate variability on Earth. This coupled ocean-atmosphere phenomenon involves the interaction of equatorial Kelvin and Rossby waves, the Bjerknes positive feedback, and teleconnections that affect weather patterns worldwide. El Niño ("the boy child," named by Peruvian fishermen for its tendency to appear around Christmas) and its cold counterpart La Niña ("the girl child") together constitute the oscillatory extremes of a system that reshapes global weather patterns, marine ecosystems, agricultural productivity, and public health on 2–7 year timescales.

Normal (Neutral) Conditions in the Tropical Pacific

Understanding ENSO requires first grasping the "baseline" state of the tropical Pacific. Under neutral conditions, the coupled ocean-atmosphere system maintains a characteristic east-west asymmetry driven by persistent trade winds:

Trade Winds and the Warm Pool

The tropical Pacific is dominated by easterly trade winds driven by the subtropical high-pressure systems in both hemispheres. These winds blow persistently from east to west along the equator at 5–8 m/s, creating fundamental asymmetries in the ocean:

Sea surface height: Wind stress pushes water westward, creating a ~40–60 cm sea level difference across the basin — the western Pacific stands physically higher than the eastern Pacific
Warm pool accumulation: Warm surface water (SST > 28°C) piles up in the western Pacific, forming the Indo-Pacific Warm Pool, the largest body of warm water on Earth (extending ~15,000 km²)
Thermocline tilt: The thermocline (the sharp temperature gradient separating warm surface waters from cold deep waters) slopes from ~200 m depth in the west to ~50 m in the east
Cold tongue: In the eastern equatorial Pacific, the shallow thermocline and trade-wind-driven upwelling bring cold, nutrient-rich deep water to the surface, creating the equatorial cold tongue (SST ~20–24°C off Peru and Ecuador)

The Walker Circulation

The east-west SST gradient drives a large-scale atmospheric circulation cell first described by Sir Gilbert Walker in the 1920s:

Rising branch: Over the western Pacific warm pool, warm moist air rises vigorously, producing deep convection and heavy rainfall over Indonesia, Papua New Guinea, and northern Australia (~2,000–3,000 mm/year)
Upper-level westerlies: At the tropopause (~15 km altitude), air flows eastward across the Pacific
Sinking branch: Over the eastern Pacific cold tongue, air subsides, creating dry conditions along coastal Peru and Ecuador (<50 mm/year precipitation in the Atacama Desert)
Surface easterlies: Air returns westward at the surface as the trade winds, completing the cell

The Walker circulation is self-reinforcing: strong trades maintain the cold tongue, which in turn maintains the pressure gradient that drives the trades. This coupling between ocean and atmosphere is the foundation of ENSO dynamics.

Ekman Transport and Equatorial Upwelling

The trade winds drive Ekman transport in the surface layer. Because the Coriolis parameter changes sign at the equator, Ekman transport is poleward on both sides of the equator, creating equatorial divergence and upwelling:

$$w_E = \frac{1}{\rho_0}\nabla \times \left(\frac{\boldsymbol{\tau}}{f}\right) \quad \text{(Ekman pumping velocity)}$$

Near the equator, the upwelling velocity reaches 1–2 m/day, bringing ~50 Sv of cold, nutrient-laden water to the surface. This supports one of the world's most productive marine ecosystems: the Humboldt Current system off Peru produces ~10% of the global fish catch from <0.1% of the ocean's area.

The Bjerknes Feedback: Engine of ENSO

Jacob Bjerknes (1969) identified the crucial positive feedback loop that couples the ocean and atmosphere in the tropical Pacific. Any initial perturbation to the system — whether a weakening of the trades, a warm SST anomaly, or a thermocline depth change — is amplified through this chain:

$$\text{Warm SST east} \rightarrow \text{weak trades} \rightarrow \text{deeper thermocline east} \rightarrow \text{warmer SST east} \rightarrow \ldots$$

This positive feedback loop amplifies initial perturbations, driving El Niño growth

The Bjerknes feedback operates through three coupled mechanisms:

1. Thermocline Feedback

Weakened trade winds reduce the east-west thermocline tilt. A deeper thermocline in the east cuts off the supply of cold water to the surface through upwelling. Even though upwelling may continue mechanically, the water brought up is now warm, allowing SST to rise.

$$T'_{\text{sfc}} \propto -\bar{w}\frac{\partial T'}{\partial z} \approx -\bar{w}\frac{T_{\text{sfc}} - T_{\text{sub}}}{H_{\text{mix}}}$$

2. Zonal Advective Feedback

Weakened trade winds reduce the westward surface current, allowing the warm pool edge to migrate eastward. Anomalous eastward currents advect warm water from the western Pacific into the central Pacific:

$$\frac{\partial T'}{\partial t} \sim -u'\frac{\partial \bar{T}}{\partial x}$$

This feedback is particularly important for Central Pacific (Modoki) El Niño events.

3. Ekman Feedback

Weakened trades reduce equatorial upwelling and Ekman divergence. Less cold water reaches the surface, allowing SST to warm. Additionally, reduced wind speed decreases evaporative cooling (latent heat flux scales as $Q_L \propto |\mathbf{u}|(q_s - q_a)$), further warming SST.

The Bjerknes Index (BI) quantifies the total positive feedback strength:

$$\text{BI} = \mu_a \beta_u \beta_h \frac{\partial \bar{T}}{\partial z}\bigg|_{\text{sub}} + \mu_a \beta_u \frac{\partial \bar{T}}{\partial x} - \alpha_s$$

where $\mu_a$ is the atmospheric coupling, $\beta_u$ is the ocean dynamical response, $\beta_h$ is the thermocline response, and $\alpha_s$ is the thermal damping. BI > 0 indicates growth.

El Niño: The Warm Phase

An El Niño event represents a dramatic reorganization of the tropical Pacific ocean-atmosphere system. Events typically develop through distinct phases over 12–18 months:

Precursor Phase (Boreal Spring)

Before El Niño onset, equatorial heat content (warm water volume above the 20°C isotherm) builds up during a "recharge" phase. Positive WWV anomalies of 1–2 × 10¹&sup4; m³ typically precede events by 6–9 months. Subsurface temperature anomalies along the equator — visible as downwelling Kelvin wave pulses in TAO/TRITON mooring data — signal the developing event. Westerly wind bursts (WWBs) in the western Pacific, often triggered by the Madden-Julian Oscillation (MJO), provide the stochastic trigger.

Growth Phase (Boreal Summer–Fall)

The Bjerknes feedback amplifies initial perturbations. Key oceanic changes include:

Trade winds weaken by 2–5 m/s across the central Pacific
Thermocline deepens by 20–50 m in the eastern Pacific and shoals by 10–30 m in the west
Equatorial undercurrent (Cromwell Current) weakens, reducing the supply of cold water to the east
Warm pool edge migrates eastward by 2,000–4,000 km to the central Pacific
Sea level rises by 20–30 cm along South American coast and drops in the western Pacific

Mature Phase (Boreal Winter, DJF)

El Niño events typically peak in December–February. At peak:

SST anomalies in the Niño 3.4 region reach +1 to +3°C (strong events exceed +2.5°C)
Atmospheric convection center shifts from the Maritime Continent to the central Pacific (date line)
Southern Oscillation Index (SOI) becomes strongly negative (< −1.5)
Upwelling off Peru and Ecuador virtually ceases — anchovy and sardine fisheries collapse
Global mean surface temperature rises ~0.1–0.2°C with a 3–6 month lag

Decay Phase (Following Boreal Spring–Summer)

Reflected Rossby waves from the western boundary propagate eastward as upwelling Kelvin waves, shoaling the thermocline and terminating the warm event. Simultaneously, heat discharge from the equatorial band (poleward Sverdrup transport) depletes the warm water volume, preconditioning the system for La Niña. Most El Niño events decay rapidly through the following spring, though some strong events persist into a second year.

Historical El Niño Events

1982–83 (Very Strong)

Niño 3.4 peak: +2.2°C. Caught forecasters by surprise — no ENSO observing system existed. Devastating floods in Peru killed 600+. Australian drought/bushfires. Coral bleaching across Pacific. Estimated $8 billion in global damages. Led to creation of the TAO array.

1997–98 (Very Strong)

Niño 3.4 peak: +2.4°C. Strongest on record at the time. Indonesian fires and haze crisis. 23,000 deaths and $33+ billion damages globally. Massive coral bleaching (16% of world's reefs died). Successfully predicted months in advance. Followed by strong La Niña.

2015–16 (Very Strong)

Niño 3.4 peak: +2.6°C. Contributed to 2016 being the warmest year on record. Third mass global coral bleaching. Horn of Africa drought. Well-predicted. Combined with long-term warming trend, some impacts exceeded 1997–98 despite similar ENSO amplitude.

La Niña: The Cold Phase

La Niña is not simply the mirror image of El Niño. While involving opposite-signed SST anomalies, La Niña events have distinct dynamics, duration, and impacts. La Niña represents an intensification of the normal Pacific state rather than a reversal.

Oceanic Characteristics

Enhanced trade winds: Easterlies strengthen by 1–3 m/s, increasing the east-west sea level gradient and thermocline tilt
Intensified upwelling: Stronger equatorial upwelling and coastal upwelling off Peru bring anomalously cold water to the surface (SST anomalies of −1 to −2°C in Niño 3.4)
Expanded cold tongue: The equatorial cold tongue extends further westward than normal, pushing the warm pool edge to ~150°E
Shoaling thermocline: The eastern Pacific thermocline rises to ~30 m depth, enhancing the efficiency of upwelling in cooling the surface
Enhanced biological productivity: Stronger upwelling brings nutrients to the photic zone, supporting high primary productivity and anchovy fisheries

Key Asymmetries with El Niño

La Niña exhibits important asymmetries compared to El Niño:

Duration: La Niña events tend to persist longer (often 2–3 years, e.g., 2020–23 "triple dip"), while El Niño events rarely last beyond 12–15 months. This is because the mean state upwelling acts as a stronger negative SST feedback during warm events.
Amplitude: La Niña SST anomalies are typically weaker (−0.5 to −2.0°C) compared to strong El Niño events (+1.5 to +2.6°C). The SST response is bounded below by the temperature of upwelled water (~15°C).
Transition: El Niño is often followed by La Niña (via the recharge-discharge mechanism), but La Niña transitions back to El Niño less reliably. Multi-year La Niña events may require a new WWB trigger to terminate.
Nonlinearity: The SST-wind stress relationship is nonlinear — atmospheric convection responds more strongly to warm anomalies than cold anomalies (since convection requires SST > ~27.5°C).

La Niña Impacts

Pacific and Americas

Enhanced Atlantic hurricane season (reduced wind shear)
Drought in southwestern US and southern South America
Cooler, wetter winters in Pacific Northwest and western Canada
Enhanced anchovy fisheries off Peru
Global mean temperature suppressed by ~0.1°C

Asia-Pacific and Africa

Enhanced Australian monsoon and flooding (e.g., 2010–11 Queensland floods)
Increased rainfall over Maritime Continent and Southeast Asia
Stronger Indian summer monsoon with more rainfall
Flooding in parts of East Africa, drought in East Africa's Horn
Enhanced Pacific typhoon activity

Notable La Niña Events

1988–89 (Strong)

Niño 3.4: −1.8°C. Followed the 1987 El Niño. Severe drought in central US (agricultural losses >$40 billion). Record hurricane season. Flooding in Bangladesh.

1998–2001 (Prolonged)

Niño 3.4: −1.7°C (peak). Three-year event following the massive 1997–98 El Niño. US drought. Record Atlantic hurricanes. Cool global temperatures masked greenhouse warming trend.

2020–23 ("Triple Dip")

Unprecedented three consecutive La Niña years. Queensland flooding. Record US tornado outbreaks. Horn of Africa drought. Contributed to the temporary slowdown in global temperature rise despite record greenhouse gas concentrations.

Equatorial Kelvin and Rossby Waves in ENSO

Equatorial waves are the ocean's adjustment mechanism and provide the delayed negative feedback that terminates ENSO events. The equator acts as a waveguide because the Coriolis parameter vanishes there, trapping wave energy within a few degrees of latitude:

$$\text{Kelvin wave speed: } c_K = \sqrt{g'H_1} \approx 2.8 \text{ m/s (first baroclinic mode)}$$

$$\text{Rossby wave speed: } c_R = -\frac{c_K}{2n+1}, \quad n = 1, 2, 3, \ldots$$

$$\text{Equatorial Rossby radius: } R_{eq} = \sqrt{\frac{c_K}{\beta}} \approx 300 \text{ km}$$

$g' = g\Delta\rho/\rho$ is the reduced gravity, $H_1$ is the upper layer thickness, $\beta = 2\Omega\cos\phi/R_E \approx 2.3 \times 10^{-11}$ m$^{-1}$s$^{-1}$

Kelvin Waves

Equatorial Kelvin waves propagate eastward only, trapped within $\pm R_{eq}$ of the equator. Downwelling Kelvin waves (deepening thermocline) are generated by westerly wind bursts and cross the 15,000 km Pacific basin in ~2 months. They deepen the thermocline in the eastern Pacific, suppressing upwelling and warming SST. Upon reaching the eastern boundary, Kelvin waves partially reflect as westward-propagating Rossby waves and partially transmit poleward along the coast as coastal Kelvin waves (detectable as sea level anomalies in tide gauges along the Americas).

Rossby Waves

Westward-propagating Rossby waves travel at $c_K/(2n+1)$, so mode-1 Rossby waves are 3× slower than Kelvin waves and take ~6 months to cross the Pacific. Off-equatorial upwelling Rossby waves, generated by wind curl anomalies during El Niño, shoal the thermocline as they propagate westward. Upon reaching the western boundary, they partially reflect as upwelling Kelvin waves that propagate back eastward, shoaling the thermocline in the eastern Pacific and terminating the warm event. This round-trip wave propagation provides the "memory" of the system and sets the ENSO period.

Westerly Wind Bursts (WWBs)

Stochastic westerly wind bursts in the western Pacific, often associated with the Madden-Julian Oscillation (MJO), are the primary trigger for equatorial Kelvin wave generation. WWBs lasting 5–20 days with wind anomalies of 5–10 m/s can initiate El Niño development. Their stochastic nature contributes to the irregularity of ENSO timing. State-dependent noise: WWBs are more likely during warm pool expansion (when the warm pool edge extends east of ~155°E), creating a multiplicative noise effect that makes ENSO an example of a noise-driven nonlinear oscillator.

Derivation: Shallow Water Equations on the Equatorial Beta-Plane

The equatorial wave dynamics are derived from the linearized shallow water equations on an equatorial beta-plane:

Step 1: Governing Equations

Start with the reduced-gravity shallow water equations linearized about a state of rest:

$$\frac{\partial u}{\partial t} - \beta y v = -g'\frac{\partial h}{\partial x} + \frac{\tau^x}{\rho_0 H_1}$$

$$\frac{\partial v}{\partial t} + \beta y u = -g'\frac{\partial h}{\partial y}$$

$$\frac{\partial h}{\partial t} + H_1\left(\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y}\right) = 0$$

where $f = \beta y$ is the equatorial beta-plane approximation.

Step 2: Non-Dimensionalization

Scale using the equatorial deformation radius $R_{eq} = (c_K/\beta)^{1/2}$ and wave speed $c_K = (g'H_1)^{1/2}$:

$$x \to R_{eq}\hat{x}, \quad y \to R_{eq}\hat{y}, \quad t \to (R_{eq}/c_K)\hat{t}$$

$$u \to c_K\hat{u}, \quad v \to c_K\hat{v}, \quad h \to H_1\hat{h}$$

Step 3: Kelvin Wave Solution

Set $v = 0$ (no meridional velocity). The equations reduce to:

$$\frac{\partial u}{\partial t} = -\frac{\partial h}{\partial x}, \quad yu = -\frac{\partial h}{\partial y}, \quad \frac{\partial h}{\partial t} + \frac{\partial u}{\partial x} = 0$$

The first and third equations give an eastward-propagating wave. The second equation constrains the meridional structure. Solution:

$$u = h = q(x - t)\exp\left(-\frac{y^2}{2}\right)$$

The Gaussian $e^{-y^2/2}$ trapping means Kelvin waves are confined within ~300 km of the equator.

Step 4: Rossby Wave Dispersion

For the general case with $v \neq 0$, seek solutions $\propto e^{i(kx - \omega t)}$. Eliminating $u$ and $h$ yields a second-order equation for $v(y)$ that is the quantum harmonic oscillator equation. Solutions exist only when:

$$\omega^2 - k^2 - \frac{k}{\omega} = 2n + 1, \quad n = 0, 1, 2, \ldots$$

For low-frequency, long wavelength ($\omega \ll k$), this gives Rossby waves:

$$\omega = \frac{-k}{k^2 + 2n + 1} \quad \Rightarrow \quad c_R = \frac{-c_K}{2n+1} \text{ (long-wave limit)}$$

Theoretical Models of ENSO

Several conceptual models capture the essential physics of ENSO oscillation. Each emphasizes different aspects of the delayed negative feedback that creates the oscillatory behavior:

Delayed Oscillator (Suarez & Schopf, 1988)

The SST anomaly T in the eastern Pacific evolves as:

$$\frac{dT}{dt} = aT(t) - bT(t - \delta) - cT^3$$

The first term ($a \approx 2$ year$^{-1}$) is the Bjerknes positive feedback. The second term ($b \approx 3.5$ year$^{-1}$) is the delayed negative feedback from reflected Rossby waves arriving after delay $\delta \approx$ 6–12 months. The cubic term provides amplitude saturation. The oscillation period is approximately $P \approx 2\delta \cdot (1 + a/b) \approx$ 3–5 years.

Physical interpretation: when T > 0 (El Niño), the Bjerknes feedback amplifies warming. But the warm anomaly also generates off-equatorial Rossby waves that eventually reflect and return as upwelling Kelvin waves after delay δ, cooling the eastern Pacific and initiating La Niña.

Recharge-Discharge Model (Jin, 1997)

ENSO oscillation explained through two coupled variables — eastern Pacific SST anomaly (T) and equatorial heat content anomaly (h):

$$\frac{dT}{dt} = aT + bh$$

$$\frac{dh}{dt} = -cT - rh$$

The four-phase ENSO cycle in this framework:

El Niño (T > 0, h decreasing): Warm SST drives anomalous Sverdrup transport that discharges heat poleward
Transition to La Niña (h < 0, T decreasing): Depleted heat content shoals thermocline, reducing SST
La Niña (T < 0, h increasing): Cold SST reverses Sverdrup transport, recharging equatorial heat
Transition to El Niño (h > 0, T increasing): Recharged heat deepens thermocline, increasing SST

The heat content leads SST by ~90° in phase, providing predictability: positive WWV anomalies predict El Niño 6–9 months ahead.

Western Pacific Oscillator (Weisberg & Wang, 1997)

Emphasizes the role of off-equatorial wind anomalies in the western Pacific as the source of negative feedback. During El Niño, the anomalous Walker circulation generates an off-equatorial anticyclonic wind stress curl in the western Pacific, which forces upwelling Rossby waves. These propagate westward, reflect as Kelvin waves, and terminate the warm event. This model highlights the importance of western Pacific boundary reflection coefficients.

Unified Oscillator (Wang, 2001)

Combines elements of all three models into a comprehensive framework with SST, thermocline depth, wind stress, and off-equatorial Rossby wave variables. Shows that the delayed oscillator, recharge-discharge, and western Pacific oscillator are all limiting cases of the same underlying coupled ocean-atmosphere dynamics, differing mainly in which feedback pathway dominates.

Derivation: Recharge-Discharge Oscillator

Step 1: Start from Reduced Gravity Shallow Water

Begin with the zonally averaged equatorial shallow water equations. The thermocline depth anomaly $h$ in the eastern Pacific depends on the basin-wide average thermocline depth $\langle h \rangle$ plus a tilt contribution from wind stress:

$$h_E = \langle h \rangle + \alpha \tau^x$$

where $\alpha$ parameterizes the Sverdrup balance response and $\tau^x$ is the zonal wind stress anomaly.

Step 2: SST Equation

The SST anomaly in the eastern Pacific is controlled by upwelling of subsurface water whose temperature depends on thermocline depth, plus thermal damping:

$$\frac{dT}{dt} = \gamma h_E - \epsilon_T T = \gamma(\langle h \rangle + \alpha\tau^x) - \epsilon_T T$$

Since wind stress responds to SST ($\tau^x = \mu T$, the atmospheric Gill response), this becomes:

$$\frac{dT}{dt} = (\gamma\alpha\mu - \epsilon_T)T + \gamma\langle h \rangle = aT + b\langle h \rangle$$

Step 3: Heat Content Equation

The basin-averaged heat content changes due to meridional heat transport (Sverdrup transport driven by wind stress curl):

$$\frac{d\langle h \rangle}{dt} = -r\langle h \rangle - \delta_h \tau^x = -r\langle h \rangle - cT$$

During El Niño, anomalous westerlies drive poleward Sverdrup transport, discharging heat from the equatorial band ($\frac{d\langle h\rangle}{dt} < 0$ when $T > 0$).

Step 4: Eigenvalue Analysis

The linearized system has the Jacobian matrix and eigenvalues:

$$J = \begin{pmatrix} a & b \\ -c & -r \end{pmatrix}, \quad \lambda = \frac{(a-r) \pm \sqrt{(a-r)^2 - 4(bc - ar)}}{2}$$

When $(a-r)^2 < 4(bc-ar)$, the eigenvalues are complex conjugates with oscillation period:

$$P = \frac{2\pi}{\text{Im}(\lambda)} = \frac{2\pi}{\sqrt{bc - ar - \left(\frac{a-r}{2}\right)^2}} \approx 3\text{--}5 \text{ years}$$

ENSO Indices and Monitoring

ENSO is monitored through a network of indices that capture both oceanic and atmospheric components. The TAO/TRITON mooring array (70 moorings spanning the tropical Pacific) provides real-time subsurface data, complemented by satellite altimetry, SST measurements, and atmospheric reanalysis.

Niño Region Indices

SST anomaly indices defined over specific equatorial Pacific regions:

Niño 1+2: 0–10°S, 90–80°W (coastal Peru; earliest signal)
Niño 3: 5°N–5°S, 150–90°W (eastern equatorial Pacific)
Niño 3.4: 5°N–5°S, 170–120°W (primary ENSO index; best coupled signal)
Niño 4: 5°N–5°S, 160°E–150°W (central Pacific; Modoki events)

Southern Oscillation Index (SOI)

Normalized sea level pressure difference between Tahiti and Darwin:

$\text{SOI} = 10 \times \frac{P_{\text{Tahiti}} - P_{\text{Darwin}} - \overline{(P_T - P_D)}}{\text{std}(P_T - P_D)}$

Negative SOI corresponds to El Niño (reduced pressure gradient weakens trades). SOI below −8 sustained for several months indicates a significant El Niño.

ENSO period is irregular, ranging from 2–7 years. Strong events (1982–83, 1997–98, 2015–16) produce Niño 3.4 anomalies exceeding +2°C.

Oceanic Niño Index (ONI)

NOAA's primary operational index. 3-month running mean of Niño 3.4 SST anomalies relative to a 30-year centered base period (updated every 5 years). El Niño: ONI ≥ +0.5°C for 5 overlapping seasons. La Niña: ONI ≤ −0.5°C. The centered base period removes the effect of long-term warming trends on ENSO classification.

Warm Water Volume (WWV)

Volume of water above the 20°C isotherm between 5°S–5°N across the Pacific. Serves as a predictor: positive WWV anomalies precede El Niño by 6–9 months (recharge phase). Used in the recharge-discharge framework as the h variable. WWV is the single best predictor of ENSO state 2–3 seasons ahead, explaining ~50% of the variance in subsequent Niño 3.4 values.

Multivariate ENSO Index (MEI)

Combines six variables: sea level pressure, zonal and meridional surface wind, SST, surface air temperature, and total cloudiness over the tropical Pacific. Uses empirical orthogonal function (EOF) analysis to extract the leading coupled mode. Provides the most comprehensive single measure of the ENSO state.

Equatorial SOI

Standardized anomaly of the difference between area-averaged sea level pressure in the eastern Pacific (80–130°W) and Indonesia (90–140°E). More directly measures the equatorial Walker circulation strength than the traditional Tahiti–Darwin SOI, and provides a cleaner ENSO signal.

Global Teleconnections

ENSO influences weather and climate far beyond the tropical Pacific through atmospheric teleconnections — large-scale Rossby wave trains excited by the anomalous tropical heating. The displaced convection center acts as a Rossby wave source, generating wave trains that propagate poleward and eastward along great circle paths.

Pacific-North American (PNA) Pattern

El Niño shifts the subtropical jet stream southward across the southern US. The PNA teleconnection pattern features alternating high and low pressure anomalies arcing from the tropical Pacific through western North America. Result: warmer, drier winters in the Pacific Northwest and western Canada; wetter conditions across the southern US from California to Florida. La Niña reverses this pattern, with enhanced storminess in the Pacific Northwest.

Pacific-South American (PSA) Pattern

The Southern Hemisphere counterpart of the PNA. During El Niño, a Rossby wave train extends from the tropical Pacific across South America. Enhanced rainfall in southeastern South America (Argentina, Uruguay, southern Brazil), drought in northeast Brazil (the Nordeste), and anomalous warming on the Antarctic Peninsula. These patterns reverse during La Niña.

PDO / IPO

Pacific Decadal Oscillation: ENSO-like SST pattern on 20–30 year timescale. When PDO and ENSO are in phase (both positive or both negative), teleconnection impacts are amplified. When out of phase, impacts are muted. The Interdecadal Pacific Oscillation (IPO) is the basin-wide (hemisphere-symmetric) expression. PDO positive phases (1977–98, 2014–) may favor stronger El Niño events.

Indian Ocean Dipole (IOD)

SST gradient between western and eastern Indian Ocean. Positive IOD (warm west, cool east) often co-occurs with El Niño through the Walker circulation bridge — subsidence over the Maritime Continent suppresses eastern Indian Ocean convection. IOD can independently modulate Australian and East African rainfall. In some years, IOD effects can amplify or counteract ENSO teleconnections over Australia.

Detailed Regional Impacts

Tropical Cyclones

El Niño suppresses Atlantic hurricane activity through increased vertical wind shear (200–850 hPa zonal wind difference increases by 5–10 m/s) and enhanced upper-level westerlies. Conversely, eastern and central Pacific tropical cyclone activity increases (the 2015 season had a record 16 central Pacific tropical cyclones). La Niña enhances Atlantic hurricane activity — 8 of the 10 most active Atlantic hurricane seasons since 1950 occurred during La Niña or neutral conditions.

Agriculture and Food Security

ENSO affects global crop yields through altered temperature and precipitation patterns. El Niño reduces maize and rice yields in Southeast Asia, Australia, and southern Africa by 5–15%. Wheat yields in Australia can drop by 25–50% during strong events. Conversely, South American soybean production may benefit from increased rainfall. The global economic impact of ENSO on agriculture is estimated at $4–8 billion per event, disproportionately affecting developing nations. ENSO forecasts are now used operationally by agricultural ministries in many countries for crop planning.

Marine Ecosystems and Fisheries

During El Niño, suppressed upwelling off Peru reduces primary productivity by 50–80%, causing collapse of anchovy fisheries (catch drops from ~8 million tons to <1 million tons). The warm water drives tuna stocks westward. Coral bleaching occurs when SST exceeds the local maximum monthly mean by >1°C for 4+ weeks — the 2015–16 El Niño triggered the third global coral bleaching event, affecting 70% of the world's reefs. Galapagos marine iguanas, sea lions, and seabirds experience mass mortality during strong events.

Public Health

ENSO modulates disease vectors through temperature and precipitation changes. Malaria outbreaks increase in the East African highlands during El Niño (warmer, wetter conditions expand mosquito habitat). Dengue fever outbreaks in Southeast Asia correlate with La Niña. Cholera in Bangladesh shows ENSO-related periodicity. Hantavirus pulmonary syndrome in the southwestern US increases after El Niño (rodent populations boom from enhanced vegetation). These linkages enable climate-informed early warning systems for epidemic preparedness.

ENSO and Climate Change

How ENSO will respond to anthropogenic climate change is one of the most important open questions in climate science. The interaction between the mean state changes and ENSO variability involves competing mechanisms:

Arguments for Stronger ENSO

Faster warming in the east: The equatorial Pacific cold tongue warms faster than the western Pacific in many climate models, reducing the mean east-west SST gradient and making the Walker circulation more susceptible to disruption
Enhanced thermocline feedback: Increased upper-ocean stratification (due to surface warming) sharpens the thermocline, making SST more sensitive to thermocline depth changes
Increased moisture: Clausius-Clapeyron scaling (~7%/°C) means warmer air holds more moisture, amplifying precipitation anomalies during ENSO events even without changes in circulation

Arguments for Weaker ENSO

Increased thermal damping: Higher SST increases outgoing longwave radiation ($\propto T^4$) and evaporative cooling, providing stronger thermal damping of SST anomalies
Weakened Walker circulation: The tropical overturning circulation weakens under greenhouse forcing because precipitation increases slower than moisture (~2%/°C vs 7%/°C), potentially reducing the Bjerknes feedback
Deeper thermocline: Ocean warming may deepen the mean thermocline, reducing the effectiveness of the thermocline feedback

Current Assessment

CMIP6 model projections show no robust consensus on ENSO amplitude changes, but do agree on several points: (1) ENSO-driven precipitation variability will increase even if SST variability remains constant, due to the moisture increase; (2) extreme El Niño events (like 1997–98 and 2015–16) are projected to become more frequent, roughly doubling under RCP8.5 by 2100; (3) the proportion of Central Pacific events may increase; (4) ENSO teleconnections will shift as the mean jet stream positions change. Paleoclimate records suggest ENSO has varied substantially over geological time — ENSO was suppressed during the early Holocene (~6,000–8,000 years ago) when orbital forcing changed the seasonal cycle.

Python: ENSO Index, Delayed Oscillator & Wavelet Analysis

Python

!/usr/bin/env python3

script.py93 lines

#!/usr/bin/env python3
"""el_nino.py - ENSO delayed oscillator, recharge-discharge, wavelet"""
import numpy as np
from scipy.integrate import odeint
import matplotlib.pyplot as plt

# --- 1. Delayed oscillator model ---
def delayed_oscillator(T_hist, dt, a, b, delay_steps):
    """Delayed oscillator: dT/dt = aT(t) - bT(t-delta)."""
    nsteps = len(T_hist)
    T = T_hist.copy()
    for i in range(delay_steps, nsteps-1):
        dTdt = a * T[i] - b * T[i - delay_steps]
        T[i+1] = T[i] + dTdt * dt
        # Add stochastic forcing (westerly wind bursts)
        T[i+1] += 0.05 * np.random.randn()
    return T

dt = 0.01  # years
nsteps = 10000
t = np.arange(nsteps) * dt
delay = 0.5  # years
delay_steps = int(delay / dt)
a, b = 2.0, 3.5

T_init = np.zeros(nsteps)
T_init[0] = 0.5  # initial perturbation
T_do = delayed_oscillator(T_init, dt, a, b, delay_steps)

fig, axes = plt.subplots(2, 2, figsize=(14, 10))

ax1 = axes[0, 0]
ax1.plot(t, T_do, 'r-', lw=0.8)
ax1.axhline(0.5, color='orange', ls='--', alpha=0.5, label='El Nino')
ax1.axhline(-0.5, color='blue', ls='--', alpha=0.5, label='La Nina')
ax1.set_xlabel('Time (years)'); ax1.set_ylabel('SST anomaly (K)')
ax1.set_title('Delayed Oscillator Model')
ax1.legend(); ax1.grid(True, alpha=0.3); ax1.set_xlim(0, 50)

# --- 2. Recharge-discharge model ---
def recharge_discharge(y, t, a=2.0, b=1.0, c=1.0, r=0.5):
    """Jin (1997) recharge-discharge ENSO model."""
    T, h = y
    dTdt = a * T + b * h
    dhdt = -c * T - r * h
    return [dTdt, dhdt]

t_rd = np.linspace(0, 30, 3000)
y0 = [1.0, 0.0]
sol = odeint(recharge_discharge, y0, t_rd,
             args=(1.5, 1.0, 1.0, 0.8))

ax2 = axes[0, 1]
ax2.plot(t_rd, sol[:, 0], 'r-', lw=1.5, label='SST anomaly T')
ax2.plot(t_rd, sol[:, 1], 'b-', lw=1.5, label='Heat content h')
ax2.set_xlabel('Time (years)'); ax2.set_ylabel('Anomaly')
ax2.set_title('Recharge-Discharge ENSO Model')
ax2.legend(); ax2.grid(True, alpha=0.3)

# --- 3. Phase space diagram ---
ax3 = axes[1, 0]
ax3.plot(sol[:, 0], sol[:, 1], 'purple', lw=1)
ax3.plot(sol[0, 0], sol[0, 1], 'ro', ms=8, label='Start')
ax3.set_xlabel('SST anomaly T'); ax3.set_ylabel('Heat content h')
ax3.set_title('ENSO Phase Space')
ax3.legend(); ax3.grid(True, alpha=0.3)
ax3.set_aspect('equal')

# --- 4. Synthetic Nino 3.4 index with spectral analysis ---
ax4 = axes[1, 1]
# Create synthetic Nino 3.4 with irregular period
np.random.seed(42)
t_monthly = np.arange(0, 50*12) / 12  # 50 years monthly
nino34 = np.zeros(len(t_monthly))
for period in [2.5, 3.5, 5.0, 7.0]:
    phase = np.random.uniform(0, 2*np.pi)
    amp = np.random.uniform(0.5, 1.5)
    nino34 += amp * np.sin(2*np.pi*t_monthly/period + phase)
nino34 += 0.3 * np.random.randn(len(t_monthly))

ax4.fill_between(t_monthly, nino34, 0,
                 where=nino34>0.5, color='red', alpha=0.5, label='El Nino')
ax4.fill_between(t_monthly, nino34, 0,
                 where=nino34<-0.5, color='blue', alpha=0.5, label='La Nina')
ax4.plot(t_monthly, nino34, 'k-', lw=0.5)
ax4.axhline(0, color='gray', ls='-', alpha=0.3)
ax4.set_xlabel('Time (years)'); ax4.set_ylabel('Nino 3.4 Index (K)')
ax4.set_title('Synthetic Nino 3.4 Index')
ax4.legend(); ax4.grid(True, alpha=0.3)

plt.tight_layout()
plt.savefig('el_nino.png', dpi=150)
plt.show()

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Fortran: Equatorial Ocean Model (Kelvin & Rossby Waves)

Fortran

Simple reduced-gravity equatorial ocean model

program.f90116 lines

program equatorial_waves
  ! Reduced-gravity equatorial ocean model
  ! Propagates Kelvin and Rossby waves along the equator
  ! Uses forward Euler with Laplacian diffusion for stability
  implicit none
  integer, parameter :: dp = selected_real_kind(15)
  integer, parameter :: nx = 200, ny = 40
  real(dp) :: Lx, Ly, dx, dy, dt, c1, beta_eq
  real(dp) :: u(nx,ny), v(nx,ny), h(nx,ny)
  real(dp) :: u_new(nx,ny), v_new(nx,ny), h_new(nx,ny)
  real(dp) :: g_prime, H0, f_local, y_val, rho0
  real(dp) :: wind_x, x_pos, t_current, nu, lap
  real(dp) :: day_val, h_eq_east, h_eq_west, h_eq_mid
  integer :: i, j, step, nsteps, iout

! Parameters for first baroclinic mode
  g_prime = 0.03_dp       ! reduced gravity (m/s^2)
  H0 = 150.0_dp           ! mean upper layer depth (m)
  c1 = sqrt(g_prime * H0) ! Kelvin wave speed (~2.1 m/s)
  beta_eq = 2.3e-11_dp    ! equatorial beta (1/m/s)
  rho0 = 1025.0_dp        ! reference density

Lx = 15000.0e3_dp       ! 15,000 km (Pacific width)
  Ly = 2000.0e3_dp        ! 2,000 km (10 deg each side)
  dx = Lx / real(nx-1, dp)
  dy = Ly / real(ny-1, dp)
  dt = 0.3_dp * min(dx, dy) / c1  ! CFL < 1

! Diffusion coefficient for numerical stability
  nu = 0.1_dp * dx * dx / dt

nsteps = nint(180.0_dp * 86400.0_dp / dt)  ! 180 days

! Initialize
  u = 0.0_dp; v = 0.0_dp; h = 0.0_dp

print *, '=== Equatorial Kelvin & Rossby Wave Model ==='
  print '(A,F6.2,A)', ' Kelvin wave speed c1 = ', c1, ' m/s'
  print '(A,F8.1,A)', ' dt = ', dt, ' s'
  print '(A,I6)', ' Total steps: ', nsteps
  print *
  print *, 'Thermocline depth anomaly along equator (m):'
  print *, 'Day   West_h(m)   Mid_h(m)   East_h(m)   max|h|(m)'

iout = max(nsteps / 30, 1)  ! output ~30 times

do step = 1, nsteps
    t_current = real(step, dp) * dt

do j = 2, ny-1
      do i = 2, nx-1
        y_val = (real(j-1, dp) - real(ny/2, dp)) * dy
        f_local = beta_eq * y_val

! Wind forcing: westerly wind burst, western Pacific, first 30 days
        x_pos = real(i-1, dp) * dx
        wind_x = 0.0_dp
        if (t_current < 30.0_dp*86400.0_dp .and. &
            x_pos < 4000.0e3_dp .and. &
            abs(y_val) < 300.0e3_dp) then
          wind_x = 0.02_dp * exp(-(y_val/200.0e3_dp)**2)
        end if

! Laplacian diffusion for u
        lap = (u(i+1,j) - 2.0_dp*u(i,j) + u(i-1,j)) / (dx*dx) + &
              (u(i,j+1) - 2.0_dp*u(i,j) + u(i,j-1)) / (dy*dy)

! Momentum: du/dt = fv - g' dh/dx + wind/rho0*H0 + nu*lap_u
        u_new(i,j) = u(i,j) + dt * ( &
          f_local * v(i,j) - g_prime * &
          (h(i+1,j) - h(i-1,j)) / (2.0_dp*dx) + &
          wind_x / (rho0*H0) + nu*lap )

! Laplacian diffusion for v
        lap = (v(i+1,j) - 2.0_dp*v(i,j) + v(i-1,j)) / (dx*dx) + &
              (v(i,j+1) - 2.0_dp*v(i,j) + v(i,j-1)) / (dy*dy)

! dv/dt = -fu - g' dh/dy + nu*lap_v
        v_new(i,j) = v(i,j) + dt * ( &
          -f_local * u(i,j) - g_prime * &
          (h(i,j+1) - h(i,j-1)) / (2.0_dp*dy) + nu*lap )

! Laplacian diffusion for h
        lap = (h(i+1,j) - 2.0_dp*h(i,j) + h(i-1,j)) / (dx*dx) + &
              (h(i,j+1) - 2.0_dp*h(i,j) + h(i,j-1)) / (dy*dy)

! Continuity: dh/dt = -H0*(du/dx + dv/dy) + nu*lap_h
        h_new(i,j) = h(i,j) - dt * H0 * ( &
          (u(i+1,j) - u(i-1,j)) / (2.0_dp*dx) + &
          (v(i,j+1) - v(i,j-1)) / (2.0_dp*dy) ) + dt*nu*lap
      end do
    end do

! Boundary conditions (radiation + no-slip)
    h_new(1,:) = h_new(2,:); h_new(nx,:) = h_new(nx-1,:)
    u_new(1,:) = u_new(2,:); u_new(nx,:) = 0.0_dp
    v_new(:,1) = 0.0_dp; v_new(:,ny) = 0.0_dp

u = u_new; v = v_new; h = h_new

if (mod(step, iout) == 0) then
      day_val = t_current / 86400.0_dp
      ! Sample equatorial thermocline at west, mid, east
      h_eq_west = h(nx/5, ny/2)
      h_eq_mid  = h(nx/2, ny/2)
      h_eq_east = h(4*nx/5, ny/2)
      print '(F6.1, 3F12.4, F12.4)', day_val, &
            h_eq_west, h_eq_mid, h_eq_east, maxval(abs(h))
    end if
  end do

print *
  print *, 'Simulation complete: 180 days of wave propagation'
  print *, 'Downwelling Kelvin wave propagates east at ~2.1 m/s'
  print *, 'Upwelling Rossby wave propagates west at ~0.7 m/s'
end program equatorial_waves

Click Run to execute the Fortran code

Code will be compiled with gfortran and executed on the server

ENSO Prediction and the Spring Predictability Barrier

ENSO prediction is one of the most important seasonal forecasting challenges. Dynamical models and statistical methods provide useful forecasts 6–12 months ahead, but face a "spring predictability barrier":

$$\text{Correlation skill} = \frac{\text{Cov}(\text{predicted}, \text{observed})}{\sigma_{\text{pred}} \cdot \sigma_{\text{obs}}}$$

Spring Barrier

Forecast skill drops sharply when predictions must cross boreal spring (March–May). ENSO events typically develop in summer/fall, so spring is the transition period with lowest signal-to-noise ratio.

Forecast Models

Dynamical: NCEP CFSv2, ECMWF SEAS5, GFDL SPEAR. Statistical: CPC/IRI consolidated. Multi-model ensembles generally outperform individual models. Useful skill out to ~9 months.

ENSO Diversity: Central Pacific vs Eastern Pacific Events

Not all El Niño events are alike. Research since the 2000s has revealed two distinct "flavors" of ENSO with different spatial patterns, dynamics, and teleconnections:

Eastern Pacific (EP) / Canonical El Niño

Maximum SST anomaly in Niño 3 region (eastern equatorial Pacific). Driven by thermocline dynamics and Bjerknes feedback. Strong events (1982–83, 1997–98). Large global temperature impact. Strong PNA teleconnection pattern. Preceded by warm water volume buildup.

Central Pacific (CP) / El Niño Modoki

Maximum SST anomaly in Niño 4 region (central equatorial Pacific). Driven by zonal advective feedback rather than thermocline dynamics. More frequent since 2000 (2002–03, 2004–05, 2009–10). Different teleconnections: enhanced Australian drought, less US winter impact.

The El Niño Modoki Index (EMI) is used to distinguish CP events:

$$\text{EMI} = \overline{SST_A} - 0.5\overline{SST_B} - 0.5\overline{SST_C}$$

Region A: 165°E–140°W, Region B: 110–70°W, Region C: 125–145°E (all 10°S–10°N)

Whether the apparent increase in CP events reflects anthropogenic climate change (weakened Walker circulation, warmer western Pacific) or natural multidecadal variability remains an active research question.

Key Concepts Summary

Bjerknes Feedback

Positive ocean-atmosphere coupling: warm SST → weak trades → deeper thermocline → warmer SST. Drives El Niño growth. Negative feedback from equatorial waves terminates events.

Wave Dynamics

Equatorial Kelvin waves propagate eastward (~2.8 m/s), Rossby waves westward (~0.9 m/s for mode-1). Their reflection and propagation provide the delayed oscillator mechanism.

Recharge-Discharge

Equatorial heat content (warm water volume) is discharged poleward during El Niño and recharged during La Niña. The heat content anomaly precedes SST anomaly.

Global Teleconnections

ENSO affects global weather through atmospheric Rossby wave trains. PDO modulates ENSO on decadal scales. IOD interacts with ENSO in the Indian Ocean.

← 8.1 Ocean-Atmosphere 8.3 Ocean Heat Content →