Presets
Playback
Experiment
ENSO model (same as the 2D page)
Playback
Camera
Layers
Quality
Model check
The science
What is modelled
The El Niño–Southern Oscillation (ENSO) is the largest year-to-year climate swing on Earth. This page animates the classic recharge oscillator (Jin 1997; Burgers et al. 2005) on a 3D section of the equatorial Pacific: the warm pool and Indonesia on the left, the cold tongue and the Andes on the right, the 20 °C isotherm (the thermocline) as a glowing sheet, trade winds, convective clouds, TAO-type buoys and an anchovy school. The time series is the same model, same constants and same random seed as the 2D page, so numbers match it exactly.
Equations
dT/dt = R T + F h − ε T³ + σ ξ(t)dh/dt = −r h − α T(time in months). T is the east-Pacific SST anomaly (K), h the equatorial heat content / thermocline depth anomaly.- Net growth
R = c (μ − 1): the Bjerknes feedback (warm SST → weaker trades → more warming) at coupling μ minus damping. Constants c = 0.03, r = 0.02, α = 0.17, F = 0.1 per month, ε = 0.004. - Linear eigenvalues
λ = (R − r)/2 ± i√(αF − ((R + r)/2)²): period2π/√(αF − ((R + r)/2)²)(about 4 years) and growth rate(R − r)/2. The cycle becomes self-sustained aboveμc = 1 + r/c = 1.667. - Integration: RK4 with a 0.1 month step; wind bursts are white noise added as
σ √Δt N(0,1)after each step (Euler–Maruyama). The record is computed once for 40 years and then played back and scrubbed.
How the two variables become a picture (illustrative mapping)
- SST along the equator = a climatological gradient (29 °C west, 24 °C east) plus
Ttimes an east-weighted pattern (and a small opposite western pattern). The “Nino-3 SST” readout is this field at 5.5 on the −9.5 … 9.5 axis. - Thermocline depth = climatology (150 m west, 50 m east)
+ 10 m·h + 9 m/K·T·(east weighting): heat content moves the whole sheet, a warm east deepens its eastern end so the tilt flattens. - Trade winds =
1 − 0.3 Tof normal (clamped); the convection centre moves east by 2.6 units per kelvin; the anchovy school scales with1 − 0.45 T(weaker upwelling). These are qualitative rules of thumb, not fitted relations. - Temperature in the section follows
T(z) = Td + (Ts − Td)/(1 + e(z − D)/22 m)with Td = 40 − Ts, so the 20 °C isotherm sits at the thermocline depth D.
What to try
- “Quiet ocean” (σ = 0, μ = 1.9): a pure self-sustained cycle; watch the phase portrait turn into a loop and the thermocline see-saw.
- “Damped, noise-driven”: without wind bursts the signal dies out; with them events appear irregularly.
- “Linear, no saturation”: the amplitude grows without bound because the cubic term is off.
- Run the coupling-threshold experiment and compare the measured μc with 1 + r/c.
Honest simplifications
- Two boxes, constant coefficients, no seasonal cycle, no spatial structure (no central-Pacific events, no Kelvin or Rossby wave propagation), white-noise wind bursts. Parameters are tuned for a plausible period and variance, not fitted to observed Nino-3.4 data; the output is not a forecast.
- The scene is a 300 m deep section with a vertical exaggeration of roughly 16,000 times (1 unit is about 790 km sideways but 50 m deep); the real basin is about 15,000 km wide and 4 km deep. Buoys, clouds, rain, boat and fish are decoration driven by the model state; fish numbers are illustrative only.
- “El Nino / La Nina” labels use a +/-0.5 K threshold on T, while the event counters (as in the 2D page) count crossings of +/-1 K.
- Typical literature values (for you to spot-check): ENSO period 2–7 years, thermocline about 150 m in the west and 50 m in the east, west-east SST difference about 5 K.
Checks run in the page
- Measured period and growth rate of the linear limit against the analytic formulas.
- RK4 step halving.
- RK4 trajectory against the exact eigen-solution of the linear system.
- Experiment: bisection for the coupling threshold μc against 1 + r/c.
Loading model...
Loading Blender model...