Global Temperature Record
Global mean surface temperature is running about 1.4°C above its 1850–1900 baseline, with HadCRUT5 at 1.5°C in both 2024 and 2025 and human-induced warming estimated at 1.37°C in 2025. The five hottest years on record have all occurred in the last decade. Ocean heat content — a more stable indicator than air temperature — has risen even faster, with most of the planet's anthropogenic heat going into the upper ocean.
Key insights
Four independent records agree
The four main global temperature reconstructions — HadCRUT (UK Met Office), GISTEMP (NASA), NOAA NCEI, Berkeley Earth — use different methodologies and overlapping but distinct station networks. They agree on the magnitude and shape of warming since 1850 to within ~0.05°C. The agreement is one of the most thoroughly cross-validated results in observational climate science.
The ocean tells a smoother story
Ocean heat content (OHC) has risen monotonically since the 1970s when reliable global measurements began (Argo floats from 2005). Air temperature shows year-to-year variability from El Niño/La Niña and volcanic eruptions; the deep ocean smooths it out. OHC is the cleanest single indicator of the planet's energy imbalance.
Recent warming has accelerated
The 1980–2010 trend was roughly 0.18°C per decade. The 2010–2024 trend is roughly 0.27°C per decade. Some of the acceleration is forced (rising CO₂); some reflects declining aerosol pollution unmasking underlying warming; some is natural variability. The acceleration is real but the exact split is the subject of active research.
Global mean surface temperature anomaly 1850–2024
°C above 1850–1900 baseline (HadCRUT5 annual averages)
Show data table
| Anomaly (°C) | |
|---|---|
| 1850 | -0.06 |
| 1880 | 0.04 |
| 1900 | 0.12 |
| 1920 | 0.06 |
| 1940 | 0.43 |
| 1960 | 0.24 |
| 1980 | 0.55 |
| 2000 | 0.69 |
| 2010 | 1.04 |
| 2020 | 1.28 |
| 2024 | 1.54 |
Key Finding: Three phases — warming 1900–1940, a pause 1940–1975 (aerosol-masked), and accelerating warming from 1975. HadCRUT5 reached +1.54°C in 2024.
Atmospheric CO₂ concentration vs temperature
CO₂ in ppm (left axis equivalent), temperature anomaly °C
Show data table
| CO₂ concentration (ppm) | Temperature anomaly ×100 | |
|---|---|---|
| 1850 | 285 | -6 |
| 1900 | 295 | 12 |
| 1950 | 311 | 13 |
| 1970 | 325 | 27 |
| 1990 | 354 | 72 |
| 2000 | 369 | 69 |
| 2010 | 390 | 104 |
| 2020 | 414 | 128 |
| 2024 | 425 | 154 |
Key Finding: CO₂ has risen from 285 ppm (1850) to 425 ppm (2024, Mauna Loa); temperature has risen about 1.4°C on the long-term trend, tracking cumulative emissions, with 2024 itself at +1.54°C.
Methodology & caveats
Baseline period choice
Different temperature products use different baselines (1850–1900 for IPCC; 1951–1980 for GISTEMP; 1991–2020 for NOAA). The choice changes the headline anomaly but not the warming rate. Most policy discussion uses 1850–1900 as the pre-industrial baseline — the period for which we have continuous records before substantial anthropogenic CO₂.
Station coverage
Pre-1900 records are sparse — many stations are in Europe and the eastern US. Modern records combine ~25,000 land stations with ocean ship-track and buoy data and satellite-derived surface temperatures (post-1979). Modern reconstructions explicitly handle changing coverage; uncertainty bands widen in earlier decades but the broad trend is robust.
Ocean heat content as the gold standard
The ocean is the planet's main heat reservoir. Argo floats (3,800 worldwide since 2005) measure temperature and salinity to 2,000m every 10 days. OHC has risen at near-constant rate since 1970s; it is the best single index of Earth's energy imbalance. Surface air temperature is what we feel, but ocean heat content is what physically moves.
Why datasets differ by tenths in single years
HadCRUT5, GISTEMP, NOAA and Berkeley Earth treat the Arctic, sparse early data and the change from ship to buoy measurements differently. HadCRUT5 reports 2024 at about 1.54°C above 1850-1900, and others differ by up to about 0.1°C. Their long-term trends agree closely. Note the difference between a single year and the long-term level: 2024 and 2025 were near 1.5°C, but El Niño helped, and the Paris limit refers to a multi-decade average, not one year.
Observed and human-induced warming are different numbers
Observed warming is what thermometers show. Human-induced warming is an attribution estimate that removes natural swings such as El Niño and volcanic eruptions, and stood at 1.37°C in 2025. Both are valid, and both use 1850-1900 as a stand-in for pre-industrial, although some warming had already occurred by then. The 2023-25 jump in temperatures exceeded what many models expected, and how much is due to reduced shipping aerosols is still being studied.
Cite this page
GlobeChart. “Global Temperature Record.” Last updated 2026-10-31. https://globechart.com/co2-emissions/temperature-record/
Please also credit the original publisher named under each chart: UK Met Office Hadley Centre, HadCRUT5; NOAA Global Monitoring Laboratory, HadCRUT5. Spotted an error? Tell us.