Copernicus called 2025 the third warmest year at 1.47C, NASA called it a tie for second

The Copernicus Climate Change Service published its Global Climate Highlights for 2025 on 14 January 2026, and for the first time in three years the headline was a step down rather than a record. The service put the 2025 global average surface air temperature at 14.97C, which is 0.59C above the 1991 to 2020 average and 1.47C above the 1850 to 1900 level used to represent pre-industrial conditions. That made 2025 the third warmest year in the ERA5 reanalysis, 0.13C cooler than 2024 and 0.01C cooler than 2023.

NASA released its own annual figures the same day and did not agree. NASA's Goddard Institute for Space Studies put 2025 at 1.19C, or 2.14F, above its 1951 to 1980 baseline, and described the year as effectively tied with 2023, making it second warmest rather than third. NASA's release acknowledged the split directly, noting that analyses by the National Oceanic and Atmospheric Administration, Berkeley Earth, the Hadley Centre and Copernicus had all concluded 2025 was third warmest, and that these bodies use much of the same underlying temperature data but different methodologies and models.

The spread is documented in the Copernicus methods note, which states that the comparison used the latest versions of the six datasets available on 26 January 2026. Measured against 1850 to 1900, and assuming the same 0.88C offset between that period and 1991 to 2020 that Copernicus applies to its own reanalysis, the other centres place 2025 between 1.39C on HadCRUT5 and 1.45C on GISTEMPv4. Japan's JRA-3Q reanalysis gives 1.46C and an absolute figure of 14.98C, one hundredth of a degree above the ERA5 number. Copernicus itself revised the report on 28 January 2026, updating two figures and the associated text once all the global datasets for 2025 were available.

One number is not in dispute. The average for 2023 to 2025 is 1.52C above pre-industrial in ERA5 and 1.50C in JRA-3Q, which the report identifies as the first three-year average in the instrumental period to exceed 1.5C. Six months of 2025, January to April and again October and November, ran above 1.5C, closing an almost unbroken run of 21 such months that began in July 2023 with July 2024 the only exception. All eleven years from 2015 to 2025 now rank among the eleven warmest.

The ice record in the same report is harsher than the temperature one. Using the EUMETSAT OSI SAF Sea Ice Index version 2.3, Copernicus found the Arctic annual maximum in March 2025 was the lowest in the 47-year satellite record. In February, record low Arctic extent for the time of year coincided with a much below average Antarctic minimum, producing the lowest global sea ice cover of any month since satellite observations began in the late 1970s. The Antarctic minimum in February ranked fourth lowest and the September maximum third lowest. The Arctic September minimum was less extreme, ranking thirteenth lowest for the month.

On causes, the report attributes the 2023 to 2025 warmth to accelerating human-induced warming, driven partly by faster greenhouse gas build-up and reduced uptake by land and ocean sinks, and partly by reduced aerosol cooling including lower emissions from East Asia since 2010. It notes the 2023 to 2024 El Nino peaked at an Oceanic Nino Index of 2.0C, smaller than the 2015 to 2016, 1997 to 1998 and 1982 to 1983 events, so the warmth cannot be laid at its door alone.

Australia appears through its weather. The report lists Cyclone Zelia and Tropical Cyclone Alfred in February and Cyclone Fina in November, flooding from large low pressure systems in the austral late summer and autumn, and Australian wildfire activity that produced local emission records according to the Copernicus Atmosphere Monitoring Service. Most of Australia saw more days than average with extreme heat stress, while south western Australia recorded more days than average with at least strong cold stress.

What happens next is contested. The Copernicus global trend monitor, which assumes the warming rate of the past 30 years continues, points to crossing 1.5C in 2029. A Met Office forecast cited in the report expects 2026 to be a fourth consecutive year more than 1.4C above 1850 to 1900, likely below 2024 but close to 2023 and 2025. The World Meteorological Organization's 2025 to 2029 update, issued in June 2025, gave a 70 per cent chance that the five-year average would exceed 1.5C. Berkeley Earth's site now records July 2026 as the second warmest July and says a strengthening El Nino makes 2026 more likely than not to set a record.

Two things remain unsettled. The community has not agreed a single method for defining the current global surface warming level, and Copernicus offers three, which converge on about 1.4C by the end of 2025 without resolving the question. And the observational base is thinning: the United States National Snow and Ice Data Centre announced on 15 October 2025 that its Sea Ice Today service would run at reduced functionality because funding was not renewed, while continuing to issue the March and September maximum and minimum announcements.