Above and below the historic average
Every temperature figure on this page is measured against the 1850–1900 average — the "pre-industrial" baseline the Paris Agreement thresholds are defined on. That choice matters more than any other methodological decision here, and it is explained at the bottom of the page.
The full record, 1850–2024
of annual global mean surface temperature. The 10-year running mean is what the Paris thresholds actually refer to — a single year above 1.5 °C is not a breach.
Every year as one stripe
The same data with the axes removed. Blue years sat below the 1850–1900 average, red years above.
Warming is not evenly distributed
The global average of +1.5 °C is an average over a planet that is two-thirds ocean. Land has warmed roughly twice as fast as the sea, and the Arctic three to four times as fast as the global mean — which is why the number most people experience is higher than the number in the headlines.
Where it goes next
Four IPCC scenarios to 2100. The spread between them by the end of the century is wider than all the warming that has happened so far — which is the entire argument for the electrification data on the rest of this site.
Reading the scenarios
The SSP pathways are not forecasts and carry no probabilities. They are internally consistent stories about emissions, and the warming attached to each is the IPCC's best estimate with a substantial range either side.
Where today's policies actually point sits between SSP1-2.6 and SSP2-4.5. The Climate Action Tracker's assessment of current policies is around 2.6 °C by 2100; full delivery of every pledge and target on the books would be about 2.1 °C. Neither number has moved much in five years, because commitments and delivery have improved at roughly the same slow rate.
On baselines, and why figures disagree
Two well-sourced articles can give different anomalies for the same year without either being wrong. There are three reasons, and knowing them makes the rest of this dataset readable.
- The baseline. Both describe the same planet. Only the first is comparable to the Paris thresholds.
- The dataset. Copernicus ERA5 is a reanalysis with full global coverage; HadCRUT5 is built from station and ship observations with statistical infilling. ERA5 runs warmer than HadCRUT5 for the same year, and the two can order the most recent years differently. The WMO publishes a consolidated figure across six datasets, which is lower again.
- The averaging period. “1.5 °C of warming” in the Paris Agreement means a multi-decade average, not a calendar year.
This site uses 1850–1900 throughout, states which dataset each figure comes from, and never mixes baselines inside one chart. Where a figure is compiled rather than quoted, it is marked.
Latest coverage
Headlines from third-party publishers, linked in full. Nothing here feeds the figures above — it is context, not source data. All headlines.