The instrumental record · 1880–2025
One stripe per year. Colour is temperature. The right-hand edge is now.
2024 · +1.28 °C — warmest year in this series
GISTEMP v4 · NASA Goddard Institute for Space Studies · annual land–ocean anomaly against the 1951–1980 mean · retrieved 2026-08-08
01The threshold
In the Copernicus ERA5 reanalysis the 2023–2025 average is the first three-year period above 1.5 degrees Celsius; the World Meteorological Organization’s consolidated multi-dataset average for the same period is 1.48 degrees. Paris defined the threshold as a long-term average rather than a single year, and that distinction is what remains.
The Paris Agreement framed 1.5 degrees Celsius as a limit on long-term warming, not on any individual year. That distinction mattered while single hot years stood apart from the trend. It matters less now. The World Meteorological Organization puts the consolidated 2023–2025 average at 1.48 degrees above the 1850–1900 baseline, the first three-year period to sit at that level.
Which side of the line those three years fall on depends on the dataset. Copernicus records 2025 at 1.47 degrees above pre-industrial in the ERA5 reanalysis, ranking third-warmest: one hundredth of a degree below 2023, which it puts at 1.48, and 0.13 degrees below 2024, which it puts at 1.60. Those three ERA5 values average above 1.5; the WMO figure quoted here consolidates ERA5 with five other datasets and lands just below it. The ranking, and the half-hundredth either side of the threshold, are less informative than the floor beneath them. Every one of the eleven years from 2015 to 2025 now sits among the eleven warmest in the instrumental record.
A threshold crossed is not a cliff. The IPCC assessed warming as a near-linear function of cumulative carbon dioxide emissions, which means each additional tonne moves the number and no particular value triggers a discontinuity. What changes at 1.5 degrees is not the physics but the accounting: the margin between observed warming and the level governments agreed to hold.
02The record
Surface thermometers, ocean floats, satellite altimeters and polar-orbiting radiometers are separate measurement systems with separate error budgets. They agree.
Data summary: the annual anomaly begins at −0.18 °C in 1880, is −0.17 °C in 1950, +0.39 °C in 2000, peaks at +1.28 °C in 2024 and is +1.19 °C in 2025. All values are against the 1951–1980 mean.
| Year | Anomaly (°C vs 1951–1980) |
|---|---|
| 1880 | −0.18 |
| 1881 | −0.09 |
| 1882 | −0.11 |
| 1883 | −0.18 |
| 1884 | −0.28 |
| 1885 | −0.34 |
| 1886 | −0.32 |
| 1887 | −0.36 |
| 1888 | −0.18 |
| 1889 | −0.11 |
| 1890 | −0.36 |
| 1891 | −0.23 |
| 1892 | −0.27 |
| 1893 | −0.31 |
| 1894 | −0.30 |
| 1895 | −0.23 |
| 1896 | −0.12 |
| 1897 | −0.11 |
| 1898 | −0.28 |
| 1899 | −0.18 |
| 1900 | −0.09 |
| 1901 | −0.15 |
| 1902 | −0.28 |
| 1903 | −0.37 |
| 1904 | −0.48 |
| 1905 | −0.27 |
| 1906 | −0.23 |
| 1907 | −0.39 |
| 1908 | −0.43 |
| 1909 | −0.49 |
| 1910 | −0.44 |
| 1911 | −0.45 |
| 1912 | −0.37 |
| 1913 | −0.35 |
| 1914 | −0.16 |
| 1915 | −0.15 |
| 1916 | −0.36 |
| 1917 | −0.46 |
| 1918 | −0.30 |
| 1919 | −0.28 |
| 1920 | −0.28 |
| 1921 | −0.19 |
| 1922 | −0.29 |
| 1923 | −0.27 |
| 1924 | −0.27 |
| 1925 | −0.22 |
| 1926 | −0.11 |
| 1927 | −0.22 |
| 1928 | −0.20 |
| 1929 | −0.36 |
| 1930 | −0.16 |
| 1931 | −0.10 |
| 1932 | −0.16 |
| 1933 | −0.29 |
| 1934 | −0.13 |
| 1935 | −0.20 |
| 1936 | −0.15 |
| 1937 | −0.03 |
| 1938 | 0.00 |
| 1939 | −0.02 |
| 1940 | +0.12 |
| 1941 | +0.18 |
| 1942 | +0.06 |
| 1943 | +0.09 |
| 1944 | +0.20 |
| 1945 | +0.09 |
| 1946 | −0.07 |
| 1947 | −0.03 |
| 1948 | −0.11 |
| 1949 | −0.11 |
| 1950 | −0.17 |
| 1951 | −0.07 |
| 1952 | +0.01 |
| 1953 | +0.08 |
| 1954 | −0.13 |
| 1955 | −0.14 |
| 1956 | −0.19 |
| 1957 | +0.05 |
| 1958 | +0.06 |
| 1959 | +0.03 |
| 1960 | −0.02 |
| 1961 | +0.06 |
| 1962 | +0.03 |
| 1963 | +0.05 |
| 1964 | −0.20 |
| 1965 | −0.11 |
| 1966 | −0.06 |
| 1967 | −0.02 |
| 1968 | −0.08 |
| 1969 | +0.05 |
| 1970 | +0.03 |
| 1971 | −0.08 |
| 1972 | +0.01 |
| 1973 | +0.16 |
| 1974 | −0.07 |
| 1975 | −0.01 |
| 1976 | −0.10 |
| 1977 | +0.18 |
| 1978 | +0.07 |
| 1979 | +0.16 |
| 1980 | +0.25 |
| 1981 | +0.32 |
| 1982 | +0.14 |
| 1983 | +0.31 |
| 1984 | +0.15 |
| 1985 | +0.12 |
| 1986 | +0.18 |
| 1987 | +0.32 |
| 1988 | +0.39 |
| 1989 | +0.27 |
| 1990 | +0.45 |
| 1991 | +0.41 |
| 1992 | +0.22 |
| 1993 | +0.23 |
| 1994 | +0.31 |
| 1995 | +0.44 |
| 1996 | +0.33 |
| 1997 | +0.46 |
| 1998 | +0.61 |
| 1999 | +0.38 |
| 2000 | +0.39 |
| 2001 | +0.53 |
| 2002 | +0.63 |
| 2003 | +0.61 |
| 2004 | +0.53 |
| 2005 | +0.68 |
| 2006 | +0.64 |
| 2007 | +0.66 |
| 2008 | +0.54 |
| 2009 | +0.66 |
| 2010 | +0.72 |
| 2011 | +0.61 |
| 2012 | +0.65 |
| 2013 | +0.68 |
| 2014 | +0.75 |
| 2015 | +0.90 |
| 2016 | +1.01 |
| 2017 | +0.91 |
| 2018 | +0.85 |
| 2019 | +0.98 |
| 2020 | +1.01 |
| 2021 | +0.85 |
| 2022 | +0.89 |
| 2023 | +1.17 |
| 2024 | +1.28 |
| 2025 | +1.19 |
NASA’s GISTEMP analysis, which measures against a 1951–1980 baseline rather than a pre-industrial one, places 2024 at 1.28 degrees and 2025 at 1.19 degrees. The two baselines answer different questions and should not be compared directly. Against either reference the shape is the same: a record that hovers near zero until the 1970s and then climbs without returning.
Most of the additional energy does not stay in the air. The WMO reports that upper-2000-metre ocean heat content rose by roughly 23 zettajoules between 2024 and 2025, and that the ocean has been absorbing on the order of eighteen times annual human energy use each year for two decades. Warming water expands, which is one of the two reasons sea level is rising.
The other reason is land ice. NOAA’s satellite altimetry record, running from late 1992, shows global mean sea level roughly 103 millimetres higher in 2024 than in 1993, the first full year of the record, against a stated trend of 3.17 millimetres per year before any glacial isostatic adjustment. The rise is not uniform across coastlines, and regional differences can exceed the global figure.
Arctic sea ice tells the most legible version of the story because it is measured in square kilometres rather than degrees. The figure plotted here is extent, the area of ocean at least fifteen percent covered by ice, averaged over September. It stood at 7.05 million square kilometres in 1979, the first satellite year. In 2025 it was 4.75 million, though that final value is retrieved with a different algorithm from the years before it and is flagged as such on the chart. Antarctic sea ice, which held steady far longer, reached its third-lowest annual maximum on record in September 2025 and an annual minimum in March 2025 tied for second lowest, both per the National Snow and Ice Data Center.
Data summary: September Arctic sea ice extent was 7.05 million square kilometres in 1979 and 4.75 million in 2025, the lowest value in the series being 3.57 million. The 2025 value is derived from source dataset NSIDC-0803 and the years 1979 to 2024 from NSIDC-0051, so the final value is not strictly homogeneous with the rest of the series.
| Year | Extent (million km²) |
|---|---|
| 1979 | 7.05 |
| 1980 | 7.67 |
| 1981 | 7.14 |
| 1982 | 7.30 |
| 1983 | 7.39 |
| 1984 | 6.81 |
| 1985 | 6.70 |
| 1986 | 7.41 |
| 1987 | 7.28 |
| 1988 | 7.37 |
| 1989 | 7.01 |
| 1990 | 6.14 |
| 1991 | 6.47 |
| 1992 | 7.47 |
| 1993 | 6.40 |
| 1994 | 7.14 |
| 1995 | 6.08 |
| 1996 | 7.58 |
| 1997 | 6.69 |
| 1998 | 6.54 |
| 1999 | 6.12 |
| 2000 | 6.25 |
| 2001 | 6.73 |
| 2002 | 5.83 |
| 2003 | 6.12 |
| 2004 | 5.98 |
| 2005 | 5.50 |
| 2006 | 5.86 |
| 2007 | 4.27 |
| 2008 | 4.69 |
| 2009 | 5.26 |
| 2010 | 4.87 |
| 2011 | 4.56 |
| 2012 | 3.57 |
| 2013 | 5.21 |
| 2014 | 5.22 |
| 2015 | 4.62 |
| 2016 | 4.53 |
| 2017 | 4.82 |
| 2018 | 4.79 |
| 2019 | 4.36 |
| 2020 | 4.00 |
| 2021 | 4.95 |
| 2022 | 4.90 |
| 2023 | 4.38 |
| 2024 | 4.35 |
| 2025 | 4.75 |
Data summary: global mean sea level rose from −19.5 millimetres in 1993 to +83.3 millimetres in 2024, a rise of 102.8 millimetres, against a stated trend of 3.17 millimetres per year.
| Year | Anomaly (mm) |
|---|---|
| 1993 | −19.5 |
| 1994 | −14.8 |
| 1995 | −9.2 |
| 1996 | −5.8 |
| 1997 | −1.9 |
| 1998 | −2.7 |
| 1999 | −3.6 |
| 2000 | +0.6 |
| 2001 | +6.0 |
| 2002 | +7.8 |
| 2003 | +10.9 |
| 2004 | +12.8 |
| 2005 | +16.1 |
| 2006 | +18.8 |
| 2007 | +18.7 |
| 2008 | +22.8 |
| 2009 | +26.5 |
| 2010 | +27.9 |
| 2011 | +26.2 |
| 2012 | +36.6 |
| 2013 | +38.9 |
| 2014 | +41.9 |
| 2015 | +50.6 |
| 2016 | +54.2 |
| 2017 | +55.1 |
| 2018 | +57.9 |
| 2019 | +64.3 |
| 2020 | +67.0 |
| 2021 | +71.4 |
| 2022 | +74.1 |
| 2023 | +79.9 |
| 2024 | +83.3 |
03The cause
The case that human emissions drive the warming rests on isotopic composition, atmospheric physics and a closed energy budget, not on the coincidence of two rising curves.
Data summary: the annual mean rises from 315.98 ppm in 1959 to 427.35 ppm in 2025, an increase of 111.37 ppm, with no year lower than the one before it.
| Year | CO₂ (ppm) |
|---|---|
| 1959 | 315.98 |
| 1960 | 316.91 |
| 1961 | 317.64 |
| 1962 | 318.45 |
| 1963 | 318.99 |
| 1964 | 319.62 |
| 1965 | 320.04 |
| 1966 | 321.37 |
| 1967 | 322.18 |
| 1968 | 323.05 |
| 1969 | 324.62 |
| 1970 | 325.68 |
| 1971 | 326.32 |
| 1972 | 327.46 |
| 1973 | 329.68 |
| 1974 | 330.19 |
| 1975 | 331.13 |
| 1976 | 332.03 |
| 1977 | 333.84 |
| 1978 | 335.41 |
| 1979 | 336.84 |
| 1980 | 338.76 |
| 1981 | 340.12 |
| 1982 | 341.48 |
| 1983 | 343.15 |
| 1984 | 344.87 |
| 1985 | 346.35 |
| 1986 | 347.61 |
| 1987 | 349.31 |
| 1988 | 351.69 |
| 1989 | 353.20 |
| 1990 | 354.45 |
| 1991 | 355.70 |
| 1992 | 356.54 |
| 1993 | 357.21 |
| 1994 | 358.96 |
| 1995 | 360.97 |
| 1996 | 362.74 |
| 1997 | 363.88 |
| 1998 | 366.84 |
| 1999 | 368.54 |
| 2000 | 369.71 |
| 2001 | 371.32 |
| 2002 | 373.45 |
| 2003 | 375.98 |
| 2004 | 377.70 |
| 2005 | 379.98 |
| 2006 | 382.09 |
| 2007 | 384.02 |
| 2008 | 385.83 |
| 2009 | 387.64 |
| 2010 | 390.10 |
| 2011 | 391.85 |
| 2012 | 394.06 |
| 2013 | 396.74 |
| 2014 | 398.81 |
| 2015 | 401.01 |
| 2016 | 404.41 |
| 2017 | 406.76 |
| 2018 | 408.72 |
| 2019 | 411.65 |
| 2020 | 414.21 |
| 2021 | 416.41 |
| 2022 | 418.53 |
| 2023 | 421.08 |
| 2024 | 424.61 |
| 2025 | 427.35 |
The IPCC’s Sixth Assessment Report states it without hedging: “It is unequivocal that human influence has warmed the atmosphere, ocean and land. Widespread and rapid changes in the atmosphere, ocean, cryosphere and biosphere have occurred.” The report quantifies the claim rather than asserting it. Human-caused warming from 1850–1900 to 2010–2019 has a best estimate of 1.07 degrees, with a likely range of 0.8 to 1.3 degrees.
That estimate is a sum of parts, and the parts do not all point the same way. Well-mixed greenhouse gases contributed 1.0 to 2.0 degrees of warming. Other human drivers, principally aerosol pollution, contributed 0.0 to 0.8 degrees of cooling. Natural drivers moved global surface temperature by no more than a tenth of a degree in either direction, and internal variability by no more than two tenths.
The fingerprint is chemical as well as statistical. Carbon released from fossil fuels is depleted in carbon-13 and contains no carbon-14, so as concentrations rise the isotopic signature of the atmosphere shifts in the direction that fossil carbon predicts and volcanic or oceanic carbon does not. Concurrently, the stratosphere cooled while the troposphere warmed, a pattern produced by greenhouse forcing and not by a brighter sun.
The concentrations themselves are measured directly. The WMO Greenhouse Gas Bulletin put globally averaged carbon dioxide at 423.9 parts per million in 2024, methane at 1942 parts per billion and nitrous oxide at 338.0 parts per billion. These are increases of 52, 166 and 25 percent respectively above pre-1750 levels. NOAA’s Mauna Loa record, continuous since 1958, reached an annual mean of 427.35 parts per million in 2025.
Data summary: gross fossil CO₂ emissions rose from 8.86 to 38.6 billion tonnes per year between 1959 and 2024.
| Year | Emissions (Gt CO₂ per year) |
|---|---|
| 1959 | 8.86 |
| 1960 | 9.39 |
| 1961 | 9.41 |
| 1962 | 9.75 |
| 1963 | 10.27 |
| 1964 | 10.82 |
| 1965 | 11.31 |
| 1966 | 11.86 |
| 1967 | 12.24 |
| 1968 | 12.91 |
| 1969 | 13.76 |
| 1970 | 14.90 |
| 1971 | 15.50 |
| 1972 | 16.22 |
| 1973 | 17.08 |
| 1974 | 17.00 |
| 1975 | 16.99 |
| 1976 | 17.86 |
| 1977 | 18.36 |
| 1978 | 19.02 |
| 1979 | 19.51 |
| 1980 | 19.41 |
| 1981 | 18.88 |
| 1982 | 18.73 |
| 1983 | 18.90 |
| 1984 | 19.46 |
| 1985 | 20.14 |
| 1986 | 20.44 |
| 1987 | 21.11 |
| 1988 | 21.92 |
| 1989 | 22.20 |
| 1990 | 22.73 |
| 1991 | 23.21 |
| 1992 | 22.52 |
| 1993 | 22.75 |
| 1994 | 22.97 |
| 1995 | 23.52 |
| 1996 | 24.23 |
| 1997 | 24.38 |
| 1998 | 24.30 |
| 1999 | 24.84 |
| 2000 | 25.51 |
| 2001 | 25.69 |
| 2002 | 26.27 |
| 2003 | 27.65 |
| 2004 | 28.61 |
| 2005 | 29.60 |
| 2006 | 30.59 |
| 2007 | 31.50 |
| 2008 | 32.05 |
| 2009 | 31.51 |
| 2010 | 33.32 |
| 2011 | 34.48 |
| 2012 | 34.95 |
| 2013 | 35.28 |
| 2014 | 35.47 |
| 2015 | 35.40 |
| 2016 | 35.39 |
| 2017 | 35.97 |
| 2018 | 36.73 |
| 2019 | 37.09 |
| 2020 | 35.16 |
| 2021 | 36.87 |
| 2022 | 37.53 |
| 2023 | 38.09 |
| 2024 | 38.60 |
“It is unequivocal that human influence has warmed the atmosphere, ocean and land.”
IPCC Sixth Assessment Report, Working Group I, Summary for Policymakers, statement A.1 · published 2021 · ipcc.ch
04The consequence
Warming does not arrive as a uniform increase in comfortable temperatures. It arrives as shifts in the tails of distributions, and it lands hardest where adaptive capacity is lowest.
The IPCC’s Working Group II assessment estimates that approximately 3.3 to 3.6 billion people live in contexts that are highly vulnerable to climate change. The consequence of that vulnerability is measurable. Between 2010 and 2020, human mortality from floods, droughts and storms was fifteen times higher in highly vulnerable regions than in regions assessed as having very low vulnerability.
Heat is the most direct pathway. Human thermoregulation depends on evaporative cooling, which fails when humidity is high enough that sweat does not evaporate, so the risk is governed by combined heat and moisture rather than by dry-bulb temperature alone. The populations least able to avoid exposure are those working outdoors and those without reliable cooling or electricity.
Water and food arrive on a slower clock. Warming shifts precipitation patterns and intensifies both drought and heavy rainfall, often in the same region in different seasons. Glacier-fed river systems that supply irrigation across South and Central Asia depend on ice that is being drawn down. Marine fisheries face a projected loss of catch that scales with the temperature reached.
Compounding is the part that resists simple accounting. A drought that dries vegetation raises fire risk; fire removes the root structure that holds slopes; the next heavy rain produces debris flows. Each link is individually well understood. The joint probability of several arriving together is what strains insurance markets, grids and public health systems past their design assumptions.
Data summary: over 2010 to 2020, human mortality from floods, droughts and storms was 15 times higher in highly vulnerable regions than in regions assessed as having very low vulnerability.
05The fork
The gap between 1.5 and 2 degrees sounds like a rounding error. In the assessed impacts it is frequently the difference between a system that persists in reduced form and one that does not.
The observed series shown on this diagram is the same GISTEMP annual series tabulated under the first chart on this page. The two dashed paths are schematic and have no tabulated values.
The IPCC’s Special Report on 1.5 degrees was written to quantify exactly this comparison. Coral reefs are projected to decline by a further 70 to 90 percent at 1.5 degrees, with losses greater than 99 percent at 2 degrees. A sea-ice-free Arctic summer is projected roughly once per century at 1.5 degrees; at 2 degrees the frequency rises to at least once per decade.
Range loss follows the same pattern. Of 105,000 species studied, 6 percent of insects, 8 percent of plants and 4 percent of vertebrates are projected to lose over half their climatically determined range at 1.5 degrees. At 2 degrees those figures rise to 18 percent, 16 percent and 8 percent. For insects the risk roughly triples across half a degree.
Sea level responds more slowly and therefore commits further ahead. Global mean sea level rise by 2100 is projected to be about 0.1 metres lower at 1.5 degrees than at 2 degrees, a difference that implies up to 10 million fewer people exposed to related risks. Because ocean and ice sheet response continues for centuries, the choice made this decade sets conditions long after it.
Three degrees is not a symmetrical extension of the same reasoning. It is the region where the IPCC’s confidence in orderly adaptation weakens and where several assessed risks move from high to very high. Current policy trajectories matter here more than pledges, and the distinction between what has been announced and what has been implemented is the material one.
06What changes it
The physical system does not carry a large delayed penalty once emissions reach zero. That is the single most consequential and least widely known finding in the recent assessment literature.
AR6 states that reaching net zero anthropogenic carbon dioxide emissions is a requirement for stabilising human-induced warming at any level, and that emissions pathways reaching and sustaining net zero greenhouse gas emissions are projected to result in a decline in surface temperature after an earlier peak. Earlier framings implied decades of additional warming already locked in. The assessed science does not support that reading.
This does not make the problem smaller; it relocates the difficulty. Because warming tracks cumulative emissions, the total emitted before reaching zero determines the peak. AR6 assessed a remaining budget of 500 billion tonnes of carbon dioxide from the start of 2020 for a 50 percent chance of holding 1.5 degrees. The Global Carbon Project’s 2025 update puts what remains at 170 billion tonnes, which Pierre Friedlingstein, who led the study, said “will be gone before 2030 at current emission rate”.
Against that budget, the Global Carbon Budget 2025 projects fossil carbon dioxide emissions of 38.1 billion tonnes for the year, a rise of 1.1 percent, alongside 4.1 billion tonnes from land-use change. That headline is measured net of the cement carbonation sink; on the same net basis 2024 was 37.78 billion tonnes, against the 38.60 billion tonnes of the gross series charted in section 03. Some things nonetheless work. The same report identifies 35 countries that have reduced emissions while growing their economies, twice as many as a decade ago, which establishes that the coupling between output and emissions is not fixed.
The leverage is asymmetric. Decisions about electricity generation, industrial heat, land use and building standards sit with governments and firms, and individual choices operate mostly through the political and market signals they aggregate into. Warming below 1.5 degrees is, in Friedlingstein’s assessment, no longer plausible, yet every tenth of a degree above it still corresponds to measurable differences in coral survival, range loss and heat exposure. The question is how far past the line the peak is allowed to go.
07Sources & method
Climate data is revised as methods improve and records lengthen. Each value above carries the period it describes and the body that published it, so it can be checked against the source and superseded when the source updates.
GISTEMP v4
Global land–ocean surface temperature index, annual mean (the J–D column).
Mauna Loa CO₂
Annual mean atmospheric carbon dioxide, dry-air mole fraction, Mauna Loa Observatory.
Global mean CO₂
Globally averaged marine surface annual mean carbon dioxide. Held for reference; the Mauna Loa series is the one plotted.
Sea Ice Index v4.0
Arctic September monthly mean sea ice extent — the annual minimum month.
Sea level (altimetry)
Global mean sea level anomaly, 66°S–66°N, annual signals removed.
Global Carbon Budget 2025
Gross fossil CO₂ emissions from combustion and industrial processes, excluding cement carbonation.
No interpolation, no smoothing and no re-baselining beyond what each publisher provides. Annual values are taken as published. Where a series is plotted, the plotted years are exactly the complete years present in the retrieved file; partial years are excluded and the exclusion is stated with the dataset above.
Colour is a diverging scale anchored at the publisher's own baseline: parchment is 0.00 °C against the 1951–1980 GISTEMP mean, blue below it, ember and crimson above. Stops are interpolated in the Oklab colour space at build time into a fixed 256-entry table. The mapping never rescales to the data, so the same temperature is the same colour in the hero, in the charts, in the ribbon and in the icon.
Two baselines appear on this page and they are not interchangeable. Figures quoted against 1850–1900 are pre-industrial-referenced and come from the World Meteorological Organization and Copernicus. Figures quoted against 1951–1980 come from NASA GISTEMP. Each figure states its own baseline; none are compared across baselines.
On the fork diagram, the solid line is measured data. The two dashed paths are schematic: they are drawn to show the shape of the two outcomes the IPCC describes, they are not a projected dataset, and the only numeric values marked on them are the 1.5 °C and 2.0 °C thresholds, both of which are taken from the cited reports.
This site sets no cookies, runs no analytics, loads nothing from a third-party host and collects nothing. There is no form, no newsletter and no tracker. Everything is served as static files from a single origin.