Sources & methods

NOAA publishes the official index weekly. To get a value every day, this site downloads daily SST from OISST v2.1 (NOAA CoastWatch ERDDAP server) on a 1° grid over the equatorial Pacific and computes the area-weighted mean in each Niño box.

OISST’s ready-made anomaly uses the 1971–2000 climatology, while the official indices use 1991–2020 — in Niño 3.4 the difference exceeds 0.2 °C. So the climatology is recomputed here: the 1991–2020 mean fitted with three annual harmonics (a smooth annual cycle, free of day-to-day noise). The same method, applied cell by cell (sampled every 5 days, 1991–2020), produces the maps.

The result is continuously validated against the CPC weekly index on the Ocean page. The most recent days come from the preliminary product (1–2 days of lag) and are replaced by the final one (~2 weeks) automatically.

IRI/Columbia publishes the plume as a matplotlib SVG figure, not as a table. The collector reads the SVG: the grid lines give the pixel → °C transform, the comments carry the labels, and the legend (color + marker) ties each curve to a model.

At every update, the observed value reconstructed from the figure is checked against CPC’s sstoi.indices; if they differ by more than 0.15 °C, the update logs a warning. The probabilities (El Niño/neutral/La Niña) come from the bar heights in the probability figure.

By Niño 3.4 (or ONI): weak ≥ 0.5 °C · moderate ≥ 1.0 · strong ≥ 1.5 · very strong ≥ 2.0. The ≥ 2.5 °C range is highlighted because CPC uses it as the threshold for a "historic" event (above any El Niño since 1950 on the RONI). An event only becomes official after 5 consecutive seasons with ONI ≥ 0.5.