Climate Muse

The average carbon dioxide (CO₂) concentration in the atmosphere, across the whole globe, was 426.8 parts per million (ppm) in 2025, with an uncertainty of ±0.2 ppm. That is the latest annual figure from the World Meteorological Organization (WMO), published on 27 October 2026 in its Greenhouse Gas Bulletin No. 22. The change matters more than the level. The concentration rose by 2.8 ppm from 2024 to 2025, less than the 3.6 ppm rise the year before and still more than the average of the last ten years. ...

When a temperature record is announced, the first question is “compared with what?”. Most of this blog’s posts are about the answer. This one is easier, because the figures Copernicus gives for the ocean’s surface are absolute temperatures in degrees Celsius, not anomalies. There is no baseline to argue about. The ocean surface outside the poles Copernicus averages sea surface temperature across the ocean between 60°S and 60°N, the extra-polar ocean, from ERA5. ERA5 is a reanalysis: a weather model continuously corrected by observations. Its record runs from 1979. ...

Everyone knows what a heat wave is, until you try to write it down. There is no universally accepted definition: some countries use a fixed temperature, some a number of days, some a percentile of local history. A heat wave in Phoenix and one in Oslo are not the same temperature, and nobody has a standard way to say how big one is. On 20 October Reuters published a proposal for one, and a new word to go with it. The word is heat storm. ...

Around the start of October, Antarctic sea ice was reported as being at a record low. That was true, on some days, in some files. It was not true on others, and the same article that said it also said something that sounds almost opposite. This is a post about why. The ice did not behave differently in different places. The files count it differently, and a “record for the date” is a statement about one file on one day. ...

On 9 September we published a table of what three sources said the Arctic sea-ice extent was on 8 September. One of them, EUMETSAT’s OSI SAF, said 5,073,817 km². We reported it correctly. Today OSI SAF’s file says 5,133,852 km² for the same day. Nothing was misquoted, and nothing was retracted. The source changed its own number. This post is about how that happened in public, and what it did to the minimum. ...

In August, two agencies ranked the same July differently. Copernicus called July 2026 the joint-second warmest on record. NOAA called it tied for the warmest. Both were right, for reasons that had to do with how each builds its global average. August is simpler. Three datasets looked at it, and all three say the same thing: August 2026 was the warmest August on record. Three records, three different numbers Dataset August 2026 anomaly Measured against Previous August record Copernicus (ERA5) +0.85 °C 1991–2020 average +0.71 °C, 2023 and 2024 tied NOAA NCEI +1.32 °C 20th-century average 2023 and 2024 tied, beaten by 0.08 °C NASA GISTEMP +1.40 °C 1951–1980 average +1.30 °C, 2024 Put those three in a row and it looks as though the datasets disagree by more than half a degree. They do not. Each anomaly is measured from a different starting line, and the difference between 0.85 and 1.40 is mostly the difference between 1991–2020, a recent and already-warm period, and 1951–1980. Move the baseline and the number moves, even though the temperature has not. ...

Update, 14 October 2026. The OSI SAF figure in the table below was a provisional one. OSI SAF has since rewritten its own values for those days: its file now gives 5,133,852 km² for 8 September, not 5,073,817. The gap between the highest and lowest of the three sources would now be about 648,000 km², not 588,000. NSIDC’s and JAXA’s values for that day have not changed. The table is left as it was retrieved on 9 September. What happened, and what it did to the minimum, is in The Number Moved After We Quoted It. ...

On 10 August, Copernicus published its July figures and called the month the joint-second warmest July on record. On 11 August, NOAA published its own and called July 2026 tied for the warmest July on record. Same month. Two agencies with nothing to gain from disagreeing. One day apart. Both are right, and the reasons are worth more than the headline. The numbers, with the thing that usually goes missing Here is what each actually said. ...

This blog went quiet for nearly two years. The 2024 posts are still here, and that is deliberate. Taking them down would have made the archive tidier and the record worse. They cover the warmest year that had been measured at the time, written as it happened, and a blog about climate data does not get to quietly delete its own time series because there is an awkward gap after it. ...

Great collection of graphical detail and sources from Zack Labe. Shoutout to Zack. Climate Change Indicators