ENSO and seasonal rainfall, season by season

What the map shows

For each land cell (0.25°, about 28 km) and each three-month season, the correlation r between that season’s rainfall and the Niño3.4 index, 1981–2025 (44 or 45 years, depending on the season and the lead time). Positive r (blue) means the season has tended to be wetter in El Niño years and drier in La Niña years; negative r (brown) the reverse.

The darker shades (|r| ≥ 0.5) mark places where ENSO accounts for at least a quarter of the year-to-year variation in that season’s rain; the lighter ones (0.3 to 0.5) for roughly a tenth to a quarter. Grey: no clear relationship (|r| below 0.3, or not significant at p < 0.05). Hatched: these three months bring less than a quarter of the year’s rain in that cell (or under 0.25 mm a day), so the season is not assessed.

The box for a cell also shows its 12 seasons side by side, so the seasons when ENSO matters there are visible at a glance.

How it is computed

• Rain: ERA5 monthly precipitation, averaged over the three months of the season, one value per year and cell.
• ENSO: NOAA’s monthly Niño3.4 anomaly (ERSST v5, 1991–2020 base), averaged over three months, either the same three months or a window ending one, two or three months earlier. The map uses whichever of these four gives the largest |r|; the box says which.
• Which seasons count: a season is assessed in a cell only if it brings at least 25% of the year’s rain there (at least its share if rain were spread evenly over the year) and averages at least 0.25 mm a day. Where rain falls all year, as in much of Indonesia, this skips some seasons that are still wet. Drier months: Show colours those seasons too, hatched, wherever they average at least 0.25 mm a day; the box always gives their r.
• Colour: Pearson r, two-tailed p < 0.05 and |r| ≥ 0.3. “All values” shows every r in assessed seasons, significant or not.

This is the method of the global teleconnection survey. The survey’s pixel ENSO map shows each cell’s strongest season; this map shows every season. On the cells both maps cover, taking the strongest season here reproduces the survey’s map exactly.

Reading it with care

• Each cell is computed on its own, without smoothing. Even with no ENSO effect, about one cell in twenty would pass p < 0.05 by chance (somewhat more, because the best of four lead times is kept), so trust patches of the same colour, not isolated cells.
• Choosing the best of four lead times makes r slightly larger than a single fixed lead would.
• The correlation describes 1981–2025. It is not a forecast, and it says nothing about how large a particular El Niño’s effect will be.
• ERA5 rainfall comes from a weather model constrained by observations. Where gauges are few it can differ from satellite and station records; check CHIRPS or station data before relying on one area.
• Coverage: all land between 60°S and 84°N. The survey’s maps stop at 100°W, 36°S and 56°N. Coastlines are Natural Earth 1:10m with minor islands, so small islands have cells; coastal cells mix land and sea.
• Projection: with the whole world in view the map is in Equal Earth, which shows every region at its true relative size (the projection the UN General Assembly encouraged in September 2026). Dragging turns the globe, so any place, including those across 180° such as Fiji, can be brought to the middle. Zooming in changes the projection gradually to a Lambert azimuthal equal-area projection centred on the middle of the view, which keeps the area you look at true to shape, with north up. Both are equal-area. In between, the map is a blend of the two: areas are close to true in the middle of the view but can be off by up to about a third towards the corners, most at high latitudes.