El Nino and La Nina: The Oceanic Nino Index

The Oceanic Nino Index (ONI) is NOAA's three month running mean of the sea surface temperature anomaly in the Nino 3.4 region, a box of the central equatorial Pacific between 5 degrees north and south and 120 and 170 degrees west. An anomaly is the difference from a 30 year base period, and NOAA moves the base period forward every 5 years so that it stays centred on the data. A season is named by its three months (DJF is December to February) and this page dates it by its middle month, so DJF 1950 is January 1950. A difference of 1 degree C is 1.8 degrees F.

NOAA calls a period an El Nino episode when the index is at or above +0.5 C for at least 5 consecutive overlapping seasons, and a La Nina episode when it is at or below -0.5 C for as long. The index is rounded to one decimal, halves up, before it is compared with the threshold, which is how NOAA's own table is coloured; 0.46 therefore counts as 0.5. The strength classes are a common convention for the peak of an episode and are not part of NOAA's definition: weak (0.5 to 0.9 C), moderate (1.0 to 1.4), strong (1.5 to 1.9) and very strong (2.0 or more).

Since 1 February 2026 NOAA has used a Relative Oceanic Nino Index (RONI) for its official monitoring. It subtracts the average anomaly of the whole tropics from the Nino 3.4 anomaly and rescales the result, which makes it less sensitive to the choice of base period. NOAA keeps updating the ONI for users who need continuity, and this page's episodes are computed from the ONI so that the record is comparable back to 1950. The two can differ, and the table of recent seasons shows both. Values can change for up to two months after they are first posted, so the newest ones are estimates.

The index measures one region of the ocean. NOAA notes that other indices are needed to confirm that the atmosphere responded as well, so an index above the threshold is a sign of the ocean state and not a forecast of weather anywhere. An episode that is already under way in the first season of the series, DJF 1950, is marked in the table, because its start, length and peak may be incomplete.

This site measures the temperature of the long-record weather stations in its own database, 3,978 stations in 356 grid cells of 5 by 5 degrees. The series is the annual mean temperature of each station as a difference from that station's own 1961-1990 mean, area weighted. It covers land stations only, and the stations are not spread evenly over the world, so it is not a global mean temperature. The table lists the 6 strongest El Nino episodes by peak, the year in which the peak season falls (the year of its middle month) and the year after.

El Nino episodes with a peak of 1.5 C or more: the year after the peak was warmer than the peak year in 4 of 9, and the average change was +0.05 C (+0.08 F). Over all 74 pairs of consecutive years the average change was +0.02 C (+0.04 F).

La Nina episodes with a peak of 1.5 C or more: the year after the peak was warmer than the peak year in 2 of 7, and the average change was -0.20 C (-0.36 F). Over all 74 pairs of consecutive years the average change was +0.02 C (+0.04 F).

These are associations in a handful of events. The series also carries the long term warming, so a rise from one year to the next is expected on average whatever the phase of the Pacific, and a few events cannot separate the two. Both years are shown because the peak of a strong episode usually falls in the last months of one year or the first of the next (see the peak season in the table). Other influences, such as volcanic eruptions, are not accounted for here. The table does not show that El Nino caused any of the changes in it.

The counts come from this site's records against chance analysis: daily record highs and lows of the daily maximum and minimum set at the long-record stations, against the number a stable climate would produce by chance. They are not a count of every record in the world. A year with too few expected records to read is left blank.

Each calendar year from 1951 to 2025 is given the phase of the index in the DJF season (December to February) that opens it: El Nino when that value, rounded to one decimal, is +0.5 C or more, La Nina when it is -0.5 C or less, and neutral otherwise.

The opening winter is used because land temperatures follow the Pacific by a few months, so an episode that peaks in the northern winter goes with the calendar year that winter opens (the 3 highest DJF values of these years opened 2016, 1998 and 1983).

The counts are the daily record highs (of the daily maximum and of the daily minimum together) and record lows set at the long-record stations in this site's database, each divided by the number a stable climate would produce by chance, as on the records against chance page. That ratio has a long-term trend, so for each area and kind of record a straight line is fitted to the logarithm of the ratio against the year, and each year is measured as a percent above or below that line.

The average of a phase is taken on the same logarithmic scale and then shown as a percent, so a year 50 percent above the line and one a third below it average to zero. The intervals are 95 percent intervals from Student's t distribution, and the comparison of El Nino and La Nina years uses Welch's interval, which does not assume the two groups vary by the same amount.

A year is read when the stations were expected to set at least 1,000 record highs by chance for the world, or 200 for a continent (the sum of its countries). A continent is shown when it has at least 25 such years, and a phase needs 5 years for an average.

From 1951 to 2025, the stations set on average 1.9 percent more daily record highs than the long-term trend in the 26 years that began with El Nino, and 3.6 percent fewer in the 28 years that began with La Nina.

Against the trend, El Nino years set 5.7 percent more daily record highs than La Nina years from 1951 to 2025 (95 percent interval -2.6 to +14.8 percent). The interval includes zero, so these years are also consistent with no difference.

From 1951 to 2025, the stations set on average 5.4 percent fewer daily record lows than the long-term trend in the 26 years that began with El Nino, and 7.5 percent more in the 28 years that began with La Nina.

Against the trend, El Nino years set 12.0 percent fewer daily record lows than La Nina years from 1951 to 2025 (95 percent interval -19.8 to -3.4 percent). The interval does not include zero.

Of the 6 continents with enough years, the 95 percent interval of El Nino years against La Nina years in daily record highs lies above zero in Africa (1971 to 2025) and Oceania (1951 to 2025). It includes zero in Asia (1951 to 2025), Europe (1951 to 2025), North America (1951 to 2025) and South America (1986 to 2025).

Of the 6 continents with enough years, the 95 percent interval of El Nino years against La Nina years in daily record lows lies below zero in Africa (1971 to 2025) and North America (1951 to 2025). It includes zero in Asia (1951 to 2025), Europe (1951 to 2025), Oceania (1951 to 2025) and South America (1986 to 2025).

This is an association across 26 years that began with El Nino and 28 years that began with La Nina in the world series, 1951 to 2025. The phase of the Pacific is one of several influences on a year's temperatures, and nothing here says what caused the record count of any single year.

The intervals treat the years as independent of each other and the trend as a straight line on the logarithmic scale; neither holds exactly. The world counts every station, so it leans towards the continents with the most long-record stations, and a continent is only as complete as the stations this site holds for it.