Is Climate Change Real and Human-Caused?

The oldest objection to the temperature record is that it cannot be trusted: skewed by towns growing up around weather stations, or kept by a few institutions with a stake in the answer. It is in fact kept several times over. NASA and NOAA in the United States, the Met Office with the University of East Anglia in Britain, the independent nonprofit Berkeley Earth and the European Copernicus service each build a global series from their own choice of stations, ships, buoys and methods. Here are four of their figures for 2024, the WMO’s figure combining them with others, and the IPCC’s assessment of 2011 to 2020:

Sources: NASA’s and NOAA’s January 2025 releases; Copernicus’s Global Climate Highlights 2024, which also reports the WMO’s consolidated figure; Berkeley Earth’s report for 2024; IPCC AR6 WGI, SPM A.1.2, for the decade, an assessment of four records: the Met Office’s HadCRUT5, NOAAGlobalTemp, Berkeley Earth and Kadow et al. The same assessment puts 2001–2020 at 0.99 °C.

The 2024 figures run from 1.46 to 1.62 mostly because each analysis sets the 1850–1900 starting line in a slightly different place, which is how Copernicus explains the spread in its 2024 report. Agreement among teams using different stations and methods cannot exclude an error they all share. It does exclude an error in one team’s choices, which is what the objection usually alleges.

Whether 2024 was the first calendar year above 1.5 °C, the level the Paris Agreement names, depends on whose analysis you read: it was on Copernicus’s, Berkeley Earth’s and the WMO’s consolidated figure, and it fell a few hundredths short on NASA’s and NOAA’s. Berkeley’s report had already put 2023 above the line, at 1.54. 2025 came in at 1.44 °C on the WMO’s consolidated figure, third on six of the eight datasets it uses and second on NASA’s and one other; Berkeley Earth’s 2025 report also puts it at 1.44, third. The eleven warmest years on record are the last eleven, 2015 to 2025, on all eight datasets. (Berkeley’s 1.62 for 2024 is from its January 2025 report. Its 2025 report puts 2025 at 1.44 and 0.08 °C cooler than 2024, so its current figure for 2024 is about 1.52.)

Download this conclusion: the 1.5 °C line

This card sets out why 2024 is counted both above and below 1.5 °C: Copernicus, the WMO and Berkeley Earth put it above, NASA and NOAA just under, and the spread comes mostly from where each analysis sets the 1850–1900 starting line. Anyone may download it and post it.

Download the card (PNG, 1,200 × 630)

The urban version of the objection has been tested head on. Berkeley Earth compared the trends at all 39,028 of its sites with those at 16,132 “very rural” sites, picked from satellite imagery to lie well away from any town, and over 1950 to 2010 found the opposite of an urban warming effect, a small one inside its own margin of error (Wickham et al., as reported by Carbon Brief). Cities are warmer than the country around them, but about 99% of the Earth’s surface is not city. Satellites, which measure the lower atmosphere and have no weather stations for buildings to crowd, show the troposphere warming as well (Karl et al., 2006; Santer et al., 2013).

Carbon dioxide has been measured at Mauna Loa in Hawaii since March 1958, when Charles Keeling of the Scripps Institution of Oceanography took the first reading, 313 ppm. Scripps and NOAA now run separate measurements there, and the annual figures on this page are NOAA’s.

Fourier worked out in 1824 that the atmosphere keeps the surface warmer than sunlight alone could. In 1859 Tyndall found out which part of it, in his laboratory at the Royal Institution in London: oxygen, nitrogen and hydrogen let infrared radiation through, while water vapour, carbon dioxide and methane absorb it. In 1896 Arrhenius calculated what doubling the carbon dioxide would do to the temperature. The same absorption now shows from orbit, as a dip in the infrared leaving the planet in the bands where carbon dioxide absorbs (see the fingerprint table below).

Greenhouse warming and warming from a brighter sun would leave different marks. Greenhouse gases would cool the stratosphere, warm nights faster than days and cut the heat escaping in the carbon dioxide bands; a brighter sun would warm the stratosphere too and warm days fastest. The attribution section sets out which marks have been found.

Critics are right that carbon dioxide has varied naturally, between about 180 ppm in ice ages and 280 ppm in the warm periods between them, over the 800,000 years the ice cores cover. The difference now is speed. The annual mean has risen about 145 ppm above the pre-industrial 280, and NOAA puts today’s rate of increase at more than 100 times the rate at the end of the last ice age.

Most of the extra heat does not stay in the air. About 90% of it goes into the ocean (Cheng et al., 2024), so the heat stored in the upper 2,000 m tracks the energy the planet is gaining more closely than the surface temperature does.

Sources: Cheng et al.’s annual assessments in Advances in Atmospheric Sciences, which set the Chinese Academy of Sciences’ analysis beside NOAA NCEI’s and Copernicus Marine’s; Pan, Cheng et al. (2026) for 2025. A zettajoule (ZJ) is 10²¹ joules.

Since 1993 global sea level has been measured from space by a chain of five satellites, from TOPEX/Poseidon to Sentinel-6 Michael Freilich. Before that the record comes from tide gauges, which NASA’s reconstruction takes back to about 1900 (Frederikse et al., 2020).

The rate matters as much as the total, because it has more than doubled over the satellite record. In recent years about two-thirds of the rise came from ice melting on land and a third from seawater expanding as it warms; in 2024 the shares reversed, with thermal expansion behind two-thirds of it (NASA, 2025).

The National Snow and Ice Data Center (NSIDC) has tracked Arctic sea ice extent using satellite passive microwave data continuously since 1979, a 47-year record with no surface stations in it. The 2025 minimum, 4.60 million square kilometres on 10 September, tied for the tenth lowest; the last 19 years are the 19 lowest in the record, and NSIDC notes that within those 19 years there has been no significant trend.

The World Glacier Monitoring Service (WGMS) keeps the records of about 60 reference glaciers with more than 30 years of continuous measurement, and 142 glaciers reported for 2024. Mass balance, the difference between snow accumulation and melt, has been negative for the reference network in every year since 1988, 37 years running, and the losses have grown each decade.

m w.e. is metres of water equivalent lost from the glacier surface. The decade figures are the reference network’s averages as reported by Mauri Pelto in the State of the Climate series; the 2022 to 2024 figures are WGMS’s regionalised global averages.

Ice sheets are a different category from mountain glaciers: the continental ice over Greenland and Antarctica. The satellite-era reconciliation, IMBIE, combines estimates from three independent techniques (altimetry, which measures the height of the surface; gravimetry, which weighs the ice from orbit; and the input-output method, which sets snowfall against the ice flowing out) and finds that the two sheets together added 21.0 ± 1.9 mm to sea level from 1992 to 2020, with the loss rising from 105 billion tonnes a year in 1992–1996 to 372 billion in 2016–2020 (Otosaka et al., 2023). The techniques agree closely on Greenland. On East Antarctica they do not even agree on the sign, and IMBIE puts it close to balance; the Antarctic loss comes from West Antarctica and the Peninsula.

Attribution asks how much of the observed warming each cause produced. The IPCC’s 2021 physical-science report, which assessed more than 14,000 publications, gives these figures for 2010–2019 against 1850–1900.

The table sets each cause’s predicted marks against what the records show.

Total solar irradiance has been measured from satellites since late 1978, by a chain of overlapping instruments (Nimbus-7, the ACRIM series, ERBS, SOHO, SORCE). The record shows the 11-year cycle and little else: the two rival composites disagree about a two-year gap in 1989 to 1991, so one (ACRIM) shows a slight rise from 1980 to 2000 and the other (PMOD) a slight decline, but both show a decline since about 2000, the period in which the surface warmed fastest. The IPCC puts the whole natural contribution to warming since 1850–1900, solar and volcanic together, between −0.1 and +0.1 °C.

Most coverage now gets the core conclusions right; the false-balance studies below measure how far that has changed. The consensus figure comes from surveys of papers rather than of scientists, and it holds. Cook et al. (2013) found that 97% of the abstracts from 1991 to 2012 that took a position endorsed human causation; Lynas et al. (2021) found four sceptical papers in a random sample of 3,000 from the 88,125 climate papers published from 2012 to 2020, with 69% of the sample taking no position and more than 99.9% of the rest endorsing it.

In earlier decades much coverage practised false balance, setting the scientific position and its contrarian opponents side by side as if they carried equal weight. Boykoff and Boykoff (Global Environmental Change, 2004) read 636 articles on the human contribution from the New York Times, the Washington Post, the Los Angeles Times and the Wall Street Journal, 1988 to 2002: 52.7% gave roughly equal attention to the two positions, 35.3% emphasised the human role while presenting both, 6.2% emphasised doubt and 5.8% carried only the scientific position. A later study of 4,856 articles from 17 papers in five countries, 2005 to 2019, found 90% represented the science accurately, with accuracy improving over the period (McAllister et al., Environmental Research Letters, 2021).

Brulle (Climatic Change, 2014) counted 91 U.S. organisations opposing limits on emissions, with a combined income of just over $900 million a year in 2003 to 2010, of which $64 million a year was traceable foundation money; the total is their whole income, not their spending on climate, as Brulle himself noted. Oreskes and Conway, Brulle and Supran and Oreskes (2017) have documented concerted and funded efforts to manufacture doubt about climate science, drawing parallels to tobacco industry tactics. ExxonMobil’s own scientists, in internal memos and published papers between 1977 and 2003, projected warming of 0.20 ± 0.04 °C a decade, which is what has happened, while the company’s public statements disputed the science; the projections were assessed from the company’s records by Supran, Rahmstorf and Oreskes (Science, 2023).