Where every number comes from

• The intensity gauge is on the conventional scale, and only the conventional scale. Niño 3.4 anomaly below 0.5°C is neutral; 0.5–0.9 weak; 1.0–1.4 moderate; 1.5–1.9 strong; 2.0°C and above very strong. The current reading is +2.7°C. This is a deliberate choice, not an oversight: the gauge's bands are calibrated to NOAA's conventional index specifically, so RONI and BOM's relative reading are never fed into it — doing so would silently apply the wrong threshold to the wrong number. See the site's DECISIONS.md for the full reasoning.
• Three headline numbers, three agencies, three thresholds — never interchange them. The weekly pair leads because it's the cleanest like-with-like comparison: the Bureau of Meteorology's own weekly relative Niño 3.4 reading (+2.20°C for the week ending 2026-08-09, BOM's own El Niño threshold +0.8°C) against the conventional weekly Niño 3.4 anomaly above (NOAA's ±0.5°C threshold). NOAA's RONI — +1.0°C for MJJ 2026, also ±0.5°C — is the official classification metric: the ONI calculation minus the average sea-surface-temperature anomaly of the whole tropical belt, so it isolates the Pacific's excess warmth rather than a raw reading against a fixed climatology, published seasonally rather than weekly. CPC itself notes RONI values can revise for up to two months after first posting.
• The race chart aligns events by calendar month (January of the onset year through June of year two). Historical lines use the CPC monthly Niño 3.4 series; the 2026 line uses monthly means of the weekly readings, because the monthly file lags — months with fewer than three weekly values are omitted. Climatology baselines differ slightly between the two files (roughly a tenth of a degree), which we accept for a comparison chart and note here.
• Caching and fallbacks. Data is fetched server-side and cached (daily for weekly SSTs, weekly for the long series). If NOAA is unreachable, bundled snapshots render instead and an amber “cached data” badge appears — the site never shows a blank chart. Every API response carries fetchedAt and isFallback.
• Probabilities are transcribed by hand. The official outlook is published as prose and graphics, not a feed. We transcribe the headline numbers after each monthly outlook; per-season splits between the headline anchors are editorial interpolations until the next official table, and the file records the outlook date.
• Open JSON feeds. The parsed data behind this site is public: /api/enso/status, /api/enso/weekly, /api/enso/oni, /api/enso/soi, /api/enso/roni, /api/enso/bom-rnino, /api/enso/race and /api/enso/probabilities. Please cache responses and attribute NOAA or BOM as the underlying source per feed.

Disclaimers

We are an independent project, not affiliated with NOAA, NASA or the Bureau of Meteorology. Seasonal forecasts are probabilistic: they tilt odds, they do not promise outcomes. For warnings and life-safety decisions, use your national meteorological service. Our phase/advisory language follows NOAA usage; Australia's BOM applies its own thresholds on its own numbers (+0.8°C for its relative Niño 3.4, versus NOAA's ±0.5°C for the conventional reading and RONI), which is why declarations sometimes differ between agencies — see the three-number breakdown above rather than assuming one shared scale.