hypothesis about global warming

• EO Explorer

• Global Maps

Climate Q&A

If earth has warmed and cooled throughout history, what makes scientists think that humans are causing global warming now.

The first piece of evidence that the warming over the past few decades isn’t part of a natural cycle is how fast the change is happening. The biggest temperature swings our planet has experienced in the past million years are the ice ages. Based on a combination of paleoclimate data and models, scientists estimate that when ice ages have ended in the past, it has taken about 5,000 years for the planet to warm between 4 and 7 degrees Celsius. The warming of the past century—0.7 degrees Celsius—is roughly eight times faster than the ice-age-recovery warming on average.

The second reason that scientists think the current warming is not from natural influences is that, over the past century, scientists from all over the world have been collecting data on natural factors that influence climate—things like changes in the Sun’s brightness, major volcanic eruptions, and cycles such as El Niño and the Pacific Decadal Oscillation. These observations have failed to show any long-term changes that could fully account for the recent, rapid warming of Earth’s temperature.

Finally, scientists know that carbon dioxide is a greenhouse gas and that it is released into the air when coal and other fossil fuels burn. Paleoclimate data show that atmospheric carbon dioxide levels are higher than they have been in the past 800,000 years. There is no plausible explanation for why such high levels of carbon dioxide would not cause the planet to warm.

• Hegerl, G. C., Zwiers, F. W., Braconnot, P., Gillett, N. P., Luo, Y., Orsini, J. A., Nicholls, N., et al. (2007). Chapter 9: Understanding and attributing climate change. In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. [Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B. , Tignor, M., and Miller, H.L. (eds.)] Cambridge and New York: Cambridge University Press.
• Jansen, E., Overpeck, J., Briffa, K.R. , Duplessy, J.-C , Joos, F., Masson-Delmotte, V., Olgao, D., et al. (2007). Chapter 6: Paleoclimate. In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. [Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B. , Tignor, M., and Miller, H.L. (eds.)] Cambridge and New York: Cambridge University Press.
• Lean, J. L., & Rind, D. H. (2008). How natural and anthropogenic influences alter global and regional surface temperatures: 1889 to 2006. Geophysical Research Letters, 35(18).
• Lockwood, M., & Fröhlich, C. (2008). Recent oppositely directed trends in solar climate forcings and the global mean surface air temperature. II. Different reconstructions of the total solar irradiance variation and dependence on response time scale. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 464(2094), 1367-1385.
• Lüthi, D., Le Floch, M., Bereiter, B., Blunier, T., Barnola, J., Siegenthaler, U., Raynaud, D., et al. (2008). High-resolution carbon dioxide concentration record 650,000–800,000 years before present. Nature, 453(7193), 379-382. [Download 800,000-Year CO 2 Data]
• Steele, L. P., Krummel, P. B., & Langenfelds, R. L. (2007). Atmospheric CO2 concentrations from sites in the CSIRO Atmospheric Research GASLAB air sampling network (August 2007 version). In Trends: A Compendium of Data on Global Change, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, US Department of Energy, Oak Ridge, TN, USA.

This entry was posted on Tuesday, May 4th, 2010 at 4:26 pm and is filed under Climate: Human Impact , Global Warming: Causes . You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

Comments are closed.

• Climate: Greenhouse Gases
• Climate: Human Impact
• Climate: Sun-Earth
• Global Warming
• Global Warming: Causes
• Global Warming: Controlling
• Global Warming: Impacts
• Global Warming: Myths
• Observations

Climate change has changed the way I think about science. Here’s why

Research fellow, Australian National University

Disclosure statement

Sophie Lewis receives funding from the Australian Research Council. She is the author of the book discussed in this article, and has received remuneration for its publication but does not receive royalties.

Australian National University provides funding as a member of The Conversation AU.

View all partners

I’ve wanted to be a scientist since I was five years old.

My idea of a scientist was someone in a lab, making hypotheses and testing theories. We often think of science only as a linear, objective process. This is also the way that science is presented in peer reviewed journal articles – a study begins with a research question or hypothesis, followed by methods, results and conclusions.

It turns out that my work now as a climate scientist doesn’t quite gel with the way we typically talk about science and how science works.

Climate change, and doing climate change research, has changed the way I see and do science. Here are five points that explain why.

Read more : Australia needs dozens more scientists to monitor climage properly

1. Methods aren’t always necessarily falsifiable

Falsifiability is the idea that an assertion can be shown to be false by an experiment or an observation, and is critical to distinctions between “true science” and “ pseudoscience ”.

Climate models are important and complex tools for understanding the climate system. Are climate models falsifiable? Are they science? A test of falsifiability requires a model test or climate observation that shows global warming caused by increased human-produced greenhouse gases is untrue. It is difficult to propose a test of climate models in advance that is falsifiable.

This difficulty doesn’t mean that climate models or climate science are invalid or untrustworthy. Climate models are carefully developed and evaluated based on their ability to accurately reproduce observed climate trends and processes. This is why climatologists have confidence in them as scientific tools, not because of ideas around falsifiability.

2. There’s lots of ways to interpret data

Climate research is messy. I spent four years of my PhD reconstructing past changes in Australian and Indonesian rainfall over many thousands of years. Reconstructing the past is inherently problematic. It is riddled with uncertainty and subject to our individual interpretations.

During my PhD, I submitted a paper for publication detailing an interpretation of changes in Indonesian climates, derived from a stalagmite that formed deep in a cave.

My coauthors had disparate views about what, in particular, this stalagmite was telling us. Then, when my paper was returned from the process of peer review, seemingly in shreds, it turns out the two reviewers themselves had directly opposing views about the record.

What happens when everyone who looks at data has a different idea about what it means? (The published paper reflects a range of different viewpoints).

Another example of ambiguity emerged around the discussion of the hiatus in global warming. This was the temporary slowdown in the rate of global warming at the Earth’s surface occurring roughly over the 15 year period since 1997. Some sceptics were adamant that this was unequivocal proof that the world was not warming at all and that global warming was unfounded.

There was an avalanche of academic interest in the warming slowdown. It was attributed to a multitude of causes, including deep ocean processes, aerosols, measurement error and the end of ozone depletion.

Ambiguity and uncertainty are key parts of the natural world, and scientific exploration of it.

3. Sometimes the scientist matters as well as the results

I regularly present my scientific results at public lectures or community events. I used to show a photo depicting a Tasmanian family sheltering under a pier from a fire front. The sky is suffused with heat. In the ocean, a grandmother holds two children while their sister helps her brother cling to underside of the pier.

After a few talks, I had to remove the photo from my PowerPoint presentation because each time I turned around to discuss it, it would make me teary. I felt so strongly that the year we were living was a chilling taste of our world to come.

Just outside of Sydney, tinderbox conditions occurred in early spring of 2013, following a dry, warm winter. Bushfires raged far too early in the season. I was frightened of a world 1°C hotter than now (regardless of what the equilibrium climate sensitivity turns out to be).

At public lectures and community events, people want to know that I am frightened about bushfires. They want to know that I am concerned about the vulnerability of our elderly to increasing summer heat stress. People want to know that, among everything else, I remain optimistic about our collective resilience and desire to care for each other.

Read more: Distrust of experts happens when we forget they are human beings

Communicating how we connect with scientific results is also important part of the role of climate scientists. That photo of the family who survived the Tasmanian bushfire is now back in my presentations.

4. Society matters too

In November 2009, computer servers at the University of East Anglia were illegally hacked and email correspondence was stolen.

A selection of these emails was published publicly, focusing on quotes that purported to reveal dishonest practices that promoted the myth of global warming. The “climategate” scientists were exhaustively cleared of wrongdoing.

On the surface, the climategate emails were an unpleasant but unremarkable event. But delving a little deeper, this can be seen as a significant turning point in society’s expectations of science.

While numerous fastidious reviews of the scientists cleared them of wrongdoing, the strong and ongoing public interest in this matter demonstrates that society wants to know how science works, and who “does” science.

There is a great desire for public connection with the processes of science and the outcomes of scientific pursuits. The public is not necessarily satisfied by scientists working in universities and publishing their finding in articles obscured by pay walls, which cannot be publicly accessed.

A greater transparency of science is required. This is already taking off, with scientists communicating broadly through social and mainstream media and publishing in open access journals.

5. Non-experts can be scientists

Climate science increasingly recognises the value of citizen scientists .

Enlisting non-expert volunteers allows researchers to investigate otherwise very difficult problems, for example when the research would have been financially and logistically impossible without citizen participation.

Read more: Exoplanet discovery by an amateur astronomer shows the power of citizen science

The OzDocs project involved volunteers digitising early records of Australian weather from weather journals, government gazettes, newspapers and our earliest observatories. This project provided a better understanding of the climate history of southeastern Australia.

Personal computers also provide another great tool for citizen collaborators. In one ongoing project, climate scientists conduct experiments using publicly volunteered distributed computing. Participants agree to run experiments on their home or work computers and the results are fed back to the main server for analysis.

While we often think of scientists as trained experts working in labs and publishing in scholarly journals, the lines aren’t always so clear. Everyone has an opportunity to contribute to science.

My new book explores this space between the way science is discussed and the way it takes place.

This isn’t a criticism of science, which provides a useful way to explore and understand the natural world. It is a celebration of the richness, diversity and creativity of science that drives this exploration.

• Climate change
• Science communication
• Citizen science
• Scientists at work

Director of STEM

Community member - Training Delivery and Development Committee (Volunteer part-time)

Chief Executive Officer

Finance Business Partner

Head of Evidence to Action

The Climate Crisis: Why the Earth Is Warming

Study disproves two of three hypotheses, aims blame at man-made pollutants

In this weeklong series, BU researchers explore the science behind Earth’s environmental changes, and what they mean for our future.

B ruce Anderson didn’t set out to prove that the rise in global temperatures since the start of the Industrial Revolution is caused by human activity. And the five-year study that he and four colleagues then published in the October 2012 Journal of Climate doesn’t draw that conclusion, but it does suggest that man-made pollutants are to blame.

The study, which tested three hypotheses about causes of the warming trend, debunks alternative theories that have been floated in recent years. At the same time, says Anderson , a College of Arts & Sciences associate professor of earth and environment, the research strengthens the theory that humans are responsible for the phenomenon, in which carbon dioxide, methane, nitrous oxide, and the other gases we emit accumulate in the atmosphere, trapping the heat that radiates from the Earth.

“This was one of the most basic scientific experiments you could perform,” he says. “You have multiple hypotheses to explain a phenomenon—the increasing global temperatures over the last half-century—and instead of trying to prove any one of them, our main goal was to disprove those hypotheses that we could, such that we’re left with one remaining hypothesis that then becomes the reigning theory.”

The consensus among scientists is that global temperatures have risen about .8 degrees Celsius since the mid-1800s. Anderson believes that focusing on the heat content in the oceans can tell us why more effectively than the temperatures themselves. He notes that the oceans store and release nearly 100 times more heat than land surfaces and that nearly 95 percent of the heat added to the environment over the last 50 years has gone into the oceans.

For their study, Anderson and his colleagues—from Washington state, the United Kingdom, and Italy—looked at the heat content of the oceans from 1950 to 2000 and used complex computer models to test the three hypotheses.

First, Anderson calculated the expected increase in ocean heat resulting from the changes in levels of carbon dioxide and other chemicals. Then he compared those numbers with the observed, or measured, increases in ocean heat over the same period—data that’s typically collected from the top 700 meters (about 2,300 feet) of oceans across the globe by devices such as an XBT, or expendable bathythermograph. The two sets of numbers matched.

“What we find,” says Anderson, who led the study, “is that the heating of the interior oceans is fully consistent with what we’d expect, given what we know about how much energy we are retaining through our emissions of heat-trapping gases and what is needed to produce the overall increase in temperatures around the world. That wasn’t entirely new. What was new was that we were able, using the same set of data, to also test alternative hypotheses that other people hadn’t looked at.”

One hypothesis suggests that global warming is the result of what’s called internal climate variability, or changes in the interactions between the oceans and the atmosphere. In this case, it’s suggested, temperatures are rising because they are drawing heat from deep within the oceans.

That can certainly happen, he says, pointing to El Niño, a warming of the waters in the equatorial Pacific that affects climate around the world. “You can see it from year to year and from decade to decade,” he says. “Some people claim that you can see it over the course of multiple decades, and that it might be a contributor” to climate change.

Ocean temperatures would be much lower than they are if the theory of internal climate variability—or changes in the interactions between the oceans and the atmosphere, such as El Niño—were true. But as the graph above shows, the predicted heat content of the oceans under that theory is at odds with what scientists have observed over 50 years.

Anderson estimated how much energy would be needed to drive the increases in global temperatures, and as a result, how much that process would reduce the heat of the interior oceans. He compared those calculations to the measured heat content of the oceans over the five decades, plotting the numbers on a graph. The two sets of numbers did not match.

It turns out, he says, that for the hypothesis to hold water, there would have to be a significant drop in ocean heat in order to feed the increasing global temperatures. But that prediction is at odds with what scientists have actually seen: a steadily increasing amount of heat over a 50-year period.

“We can show that at most 10 percent of the warming could have been contributed by heat drawn from the interior oceans,” Anderson says.

The third theory, he says, was the trickiest one to test. “Every so often, particularly in nonscientific literature, where people aren’t required to test the hypothesis they propose, there will be a slew of hypotheses that are put out there: increases in the energy output of the sun, changes in gamma radiation, galactic cosmic rays.”

“The problem with testing these hypotheses is that they’re innumerable,” he says. “People throw them out willy-nilly.” And scientists don’t know how much heat would be generated by each mechanism.

For the study, Anderson assumed that there is some unknown source of heat, and proposed that the source added as much heat to the climate system as all of the greenhouse gases combined. How would the heat content of the oceans respond?

He calculated that response and compared it to the observed heat content of the oceans. Like the second hypothesis, the two sets of numbers failed to match. “We came to the conclusion that, at most, 15 percent of the warming over the last half-century could have been the result of some unknown mechanism for heating the planet,” he says.

“No theory is in and of itself immutable. Even the theory of gravity is constantly being adjusted, but at no point in time does that mean I’m going to choose to walk off a cliff just because some questions remain about it.”

Anderson believes that the study rejects “in one fell swoop” the idea that there are mechanisms other than human activity that cause climate change. At the same time, he says, the foundation for the theory that global warming is caused by human activity is becoming more robust. “No theory is in and of itself immutable,” he says. “Even the theory of gravity is constantly being adjusted, but at no point in time does that mean I’m going to choose to walk off a cliff just because some questions remain about it.”

Questions do remain about what will happen if we continue to ramp up the concentration of heat-trapping gases. Anderson is now looking into the way changes in the climate system—in the amount of water in the atmosphere, in the number, elevation, and location of clouds, in the amount of snow and ice cover in higher latitudes and during the winter—amplify or dampen the forces of heat-trapping gases.

“But with regard to mitigation, this takes us one step further away from the need to do more process-based and scientific studies and one step closer to actually saying this is a problem that we need to address,” Anderson says. “So in that sense, whether it’s a small step or a large step, it’s progress toward that transition from developing scientific questions to developing scientific policy.”

Related Stories

• Fulweiler’s Message in a Bottle
• Global Warming Research Heats Things Up

This Series

The Climate Crisis: Tracking Change, Predicting Trouble

The climate crisis.

March 27, 2013

The Climate Crisis: Measuring Boston’s Metabolism

March 28, 2013

The Climate Crisis: Breaking the Fossil Fuel Habit

March 29, 2013

The Climate Crisis: Your Job or Your Planet?

Series home

Explore Related Topics:

• Climate Change
• Sustainability
• Share this story
• 10 Comments Add

Cindy Buccini

Editor, bostonia twitter profile.

Cindy Buccini is editor of Bostonia , Boston University’s alumni magazine, and alumni publications. Before joining the BU staff in 2001, she was a local newspaper reporter for many years and was communications coordinator for the Massachusetts Board of Higher Education. She earned a bachelor’s degree in biology and a master’s degree in science journalism, both from BU.  Profile

Senior Video Producer

Devin Hahn creates video content for BU Today , Bostonia online, and The Brink . He is a producer, a cameraman, an editor, and, under duress, a writer. Profile

Joe Chan Profile

Comments & Discussion

Boston University moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (EST) and can only accept comments written in English. Statistics or facts must include a citation or a link to the citation.

There are 10 comments on The Climate Crisis: Why the Earth Is Warming

With 95 percent of the heat going into the ocean you can’t mitigate climate change without converting this heat to work.

Excellent account of a cogent and even elegant research project.

The article is well written. However, I would have liked a more succinct description of how the third theory was tackled. It seems like the author of this article did not quite get the science on this one.

In fact, after listening to the video I feel the following line “For the study, Anderson assumed that there is some unknown source of heat, and proposed that the source added as much heat to the climate system as all of the greenhouse gases combined. ”

should have read

“For the study, Anderson assumed that there is some unknown source of heat, and CALCULATED THE ADDITIONAL HEAT THAT WOULD BE ADDED TO THE OCEANS ON TOP OF WHAT IS ALREADY BEING ADDED BECAUSE OF GREENHOUSE GASSES.”

Even then, it would have been nice to get some idea behind what the assumptions were behind the calculation.

Excellent piece of research! Digging deep into the interplay of factors controlling temperature variability is a first step and will have a broad impact on the understanding of how these factors actually tune the uncertainty in model projections of temperature.

The extra heat that warmed up the oceans was found over 5 years ago. I would suggest you look up the earth’s albedo variation. All the heat necessary to warm up the oceans as observed, with plenty left over to be radiated into space is there. CO2 may be having an effect, but it is relatively small.

The best and most elegant science always asks the most basic questions in a way that the hypotheses can be tested. I am very intrigued by this study and impressed with the results. Wish I would have thought to do it myself. Well done!!!

I am wondering how much life on earth is worth to the coal burning plant owners. Also how much it is worth to the people that say,” We don’t have global warming. ” We do, and if nothing is done the human race will be extinct in 200 years or sooner. Sad!

Global warming is not the issue of a single nation. If we didn’t respond to it in a proper way then our upcoming generations has to be struggle for their survival.

It’s an interesting piece of research but there is something that’s bothering me. Assume that the researchers are correct and that the increase in temperature is a direct effect of the CO2 in the atmosphere (which has already been proven before). There is no evidence, not in this study and not in any other studies, that this is a direct effect of human behavior. In fact, CO2 levels have been going up and down since the beginning of time, even without any human presence. I’m not saying that human activity is not contributing at all but there is a chance that we’re not able to change anything to the variations in CO2 at all. Therefore I would suggest we start looking at ways to deal with global warming instead of figuring out who is to blame.

Post a comment. Cancel reply

Your email address will not be published. Required fields are marked *

Latest from BU Today

Exploring boston’s cultural and culinary diversity, here’s where to find last-minute back-to-school shopping essentials, the six best restaurants near campus to celebrate national eat outside day, looking to purchase, or rent, textbooks here’s what you need to know, to do today: rollerama open skate at kendall common, fysop looks to expand the horizons of incoming first-years, fire program returns to college of engineering, bu names jennifer gunter king first associate university librarian for special collections, bu’s booth theatre helps boston public schools students and grads set the stage for success, to do today: summer exhibitions at isabella stewart gardner museum highlight lgbtqia+ photography, love insects this club’s for you, karen engelbourg, leader of bu’s fundraising efforts and alumni engagement, is stepping down, the terrier bucket list: 15 places you need to visit while living in boston, mechanical vs biomedical engineering: what’s the difference between the majors, chief investment officer lila hunnewell recognized for stewardship of bu’s endowment, 20 questions: can you guess the bu location, to do today: saint anthony’s feast in the north end, to do today: boston landmarks orchestra at the hatch shell, to do today: an evening of silent film with the tanglewood music center at the coolidge corner theatre, young geneiuses: summerlab gives teens a chance to learn about state-of-the art crispr technology.

Isn't there a lot of disagreement among climate scientists about global warming?

No. By a large majority, climate scientists agree that average global temperature today is warmer than in pre-industrial times and that human activity is the most significant factor. 

Today, there is no real disagreement among climate experts that humans are the primary cause of recent global warming. NOAA Climate.gov cartoon by Emily Greenhalgh. 

Consensus of experts

The United States' foremost scientific agencies and organizations have recognized global warming as a human-caused problem that should be addressed. The U.S. Global Change Research Program has published a series of scientific reports documenting the causes and impacts of global climate change. NOAA , NASA , the National Science Foundation , the National Research Council , and the Environmental Protection Agency have all published reports and fact sheets stating that Earth is warming mainly due to the increase in human-produced heat-trapping gases.

On their climate home page , the National Academies of Sciences, Engineering, and Medicines says, "Scientists have known for some time, from multiple lines of evidence, that humans are changing Earth’s climate, primarily through greenhouse gas emissions," and that "Climate change is increasingly affecting people’s lives." 

Soot from fires and air pollution contributes to global warming, and its impacts may be especially strong in the Arctic, where it darkens the snow and ice—as shown in this photo—and accelerates melting. Despite some uncertainty about just how much influence soot and other aerosol particles have played in climate change in the past century, there's little debate among climate scientists that the primary driver of recent global warming is carbon dioxide emissions. Photo from NOAA Ocean Today .

The American Meteorological Society (AMS) issued this position statement : "Scientific evidence indicates that the leading cause of climate change in the most recent half century is the anthropogenic increase in the concentration of atmospheric greenhouse gases, including carbon dioxide (CO 2 ), chlorofluorocarbons, methane, tropospheric ozone, and nitrous oxide." (Adopted April 15, 2019)

The American Geophysical Union (AGU) issued this position statement : "Human activities are changing Earth's climate, causing increasingly disruptive societal and ecological impacts. Such impacts are creating hardships and suffering now, and they will continue to do so into the future—in ways expected as well as potentially unforeseen. To limit these impacts, the world's nations have agreed to hold the increase in global average temperature to well below 2°C (3.6°F) above pre-industrial levels. To achieve this goal, global society must promptly reduce its greenhouse gas emissions." (Reaffirmed in November 2019)

The American Association for the Advancement of Science (AAAS) What We Know site states: "Based on the evidence, about 97 percent of climate scientists agree that human-caused climate change is happening."

Consensus of evidence

These scientific organizations have not issued statements in a void; they echo the findings of individual papers published in refereed scientific journals. The Institute for Scientific Information (ISI) maintains a database of over 8,500 peer-reviewed science journals, and multiple studies of this database show evidence of overwhelming agreement among climate scientists. In 2004, science historian Naomi Oreskes published the results of her examination of the ISI database in the journal Science . She reviewed 928 abstracts published between 1993 and 2003 related to human activities warming the Earth's surface, and stated, "Remarkably, none of the papers disagreed with the consensus position."

This finding hasn't changed with time. In 2016, a review paper summarized the results of several independent studies on peer-reviewed research related to climate. The authors found results consistent with a 97-percent consensus that human activity is causing climate change. A 2021 paper found a greater than 99-percent consensus.

Probably the most definitive assessments of global climate science come from the Intergovernmental Panel on Climate Change (IPCC). Founded by the United Nations in 1988, the IPCC releases periodic reports, and each major release includes three volumes: one on the science, one on impacts, and one on mitigation. Each volume is authored by a separate team of experts, who reviews, evaluates, and summarizes relevant research published since the prior report. Each IPCC report undergoes several iterations of expert and government review. The 2021 IPCC report, for instance, received and responded to more than 78,000 expert and government review comments.

Every five years, the Intergovernmental Panel on Climate Change convenes hundreds of international scientists and government representatives to review and assess peer-reviewed research on climate science. In each cycle, the panel publishes three key reports: one on the basic science , one on impacts , and one on mitigation .

The IPCC does not involve just a few scientists, or even just dozens of scientists. An IPCC press release explains: "Thousands of people from all over the world contribute to the work of the IPCC. For the assessment reports, IPCC scientists volunteer their time to assess the thousands of scientific papers published each year to provide a comprehensive summary of what is known about the drivers of climate change, its impacts and future risks, and how adaptation and mitigation can reduce those risks."

Governments and climate experts across the globe nominate scientists for IPCC authorship, and the IPCC works to find a mix of authors, from developed and developing countries, among men and women, and among authors who are experienced with the IPCC and new to the process. Published in 2021, the Sixth Assessment Report was assembled by 751 experts from more than 60 countries (31 coordinating authors, 167 lead authors, 36 review editors, and 517 contributing authors). Collectively, the authors cited more than 14,000 scientific papers. In other words, the IPCC reports themselves are a comprehensive, consensus statement on the state climate science.

In the headline statements from the Sixth Assessment report's Summary for Policymakers, the IPCC concluded:

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 scale of recent changes across the climate system as a whole – and the present state of many aspects of the climate system – are unprecedented over many centuries to many thousands of years. Human-induced climate change is already affecting many weather and climate extremes in every region across the globe. Evidence of observed changes in extremes such as heatwaves, heavy precipitation, droughts, and tropical cyclones, and, in particular, their attribution to human influence, has strengthened since [our last report].

Cook, J., D. Nuccitelli, S.A. Green, M. Richardson, B. Winkler, R. Painting, R. Way, P. Jacobs, and A. Skuce (2013). Quantifying the consensus on anthropogenic global warming in the scientific literature. Environmental Research Letters , 8, 024024. https://doi.org/10.1088/1748-9326/8/2/024024 .

Cook, J., Oreskes, N., Doran, P.T., Anderegg, W.R.L., Verheggen, B., Mailbach, E.W., Carlton, J.S., Lewandowsky, S., Skuce, A.G., Green, S.A., Nuccitelli, D., Jacobs, P., Richardson, M., Winkler, B., Painting, R., Rice, K. (2016). Consensus on consensus: A synthesis of consensus estimates on human-caused global warming. Environmental Research Letters , 11, 048002. https://doi.org/10.1088/1748-9326/11/4/048002 .

Doran, P., and M.K. Zimmerman (2009): Examining the Scientific Consensus on Climate Change. Eos , 90(3), 22–23.

IPCC. (2013). Factsheet: How does the IPCC select its authors? Accessed January 3, 2020.

IPCC. (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva Switzerland. Accessed January 22, 2020.

IPCC. (2021). Summary for Policymakers. In: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 3−32, doi:10.1017/9781009157896.001.

IPCC Sixth Assessment Report. (2021). https://www.ipcc.ch/report/ar6/wg1/

Lynas, M., Houlton, B.Z., Perry, S. (2021). Greater than 99% consensus on human caused climate change in the peer-reviewed scientific literature. Environmental Research Letters , 16, 114005. https://doi.org/10.1088/1748-9326/ac2966 .

Oreskes, N. (2004). The Scientific Consensus on Climate Change. Science , 306, 1686. https://doi.org/10.1126/science.1103618 .

Oreskes, N. (2018). The scientific consensus on climate change: How do we know we're not wrong? Climate Modelling , pp. 31–64. https://doi.org/10.1007/978-3-319-65058-6_2 .

Sherwood, S. (2011, May 10). Trust us, we're climate scientists: The case for the IPCC . The Conversation .

We value your feedback

Help us improve our content

Related Content

News & features, talking with ipcc vice-chair ko barrett: on climate change and consensus building, what's it like to be an author for the ipcc report, 2016 arctic heat would have been virtually impossible without global warming, national climate assessment map shows uneven impact of future global warming on u.s. energy spending, maps & data, climate forcing, future climate, climate models, teaching climate, toolbox for teaching climate & energy, the smap/globe partnership: citizen scientists measure soil moisture, climate youth engagement, climate resilience toolkit, climate change 2007: the physical science basis, advancing the science of climate change, the effects of climate change on agriculture, land resources, water resources, and biodiversity in the united states.

Scientists Agree: Global Warming is Happening and Humans are the Primary Cause

Published Aug 3, 2017 Updated Jan 9, 2018

The evidence is overwhelming. Record-breaking temperatures, humidity, and sea level rise, along with many other indicators, show that the Earth is warming fast, and that all the heat-trapping emissions we release into the atmosphere from burning fossil fuels is changing our climate.

The Climate Accountability Scorecard

The time to act is now. But action isn't easy: many powerful industry interests have hindered climate solutions and spread dangerous myths about climate change.

One of the preferred tactics these groups use to sow confusion is to promote studies that either ignore or misrepresent the evidence of thousands of articles published in well-established and well-respected scientific journals, which show that global warming is happening and that it is caused by humans.

No matter how much contrarians try to cloak reality, the evidence is not going away.

Widespread scientific consensus

Scientists worldwide agree that global warming is happening, and that human activity causes it.

The IPCC Fifth Assessment Report ( AR5 ), written by a panel of hundreds of climate experts and scientists from member countries of the World Meteorological Organization the United Nations Environmental Programme , plus a team of external reviewers, states unambiguously:

Human influence on the climate system is clear, and recent anthropogenic emissions of green-house gases are the highest in history. […] Warming of the climate system is unequivocal, and since the 1950s, many of the observed changes are unprecedented over decades to millennia. IPCC Fifth Assessment Report (AR5)

Building on two previous studies, a landmark 2013 peer-reviewed study evaluated 10,306 scientists to confirm that over 97 percent climate scientists agree, and over 97 percent of scientific articles find that global warming is real and largely caused by humans.

A 2016 peer-reviewed paper examined existing studies on consensus in climate research, and concluded that the 97 percent estimate is robust.

This level of consensus is equivalent to the level of agreement among scientists that smoking causes cancer – a statement that very few people, if any, contest today.

The US public also increasingly agrees that global warming is happening. A 2016 poll from Yale found that 70 percent of US residents believe global warming is happening, while record low numbers of people (12 percent) say the opposite.

A Gallup poll from 2017 showed that the number of people who worry “a great deal” about global warming has increased from 37 percent in 2016 to 45 percent in 2017.  The acceptance of human-caused emissions as the cause of warming is not keeping pace with those that believe it is happening, but it is at 53 percent.

There is no shortage of published research on the consensus of climate scientists and climate science when it comes to human-caused global warming. In addition to the references above, you can read about how the discussion on consensus developed over time in these studies .

Many different scientific societies in the United States and numerous national academies of science from around the world have also issued statements that verify the scientific claims about human-caused warming (see below).

Consensus and scientific uncertainty

Climate skeptics and deniers often misrepresent and aggrandize “ scientific uncertainty ” to undermine climate science consensus. When it comes to scientific consensus on global warming, it is important to clarify what type of uncertainty exists, and what type does not: there is strong certainty on the types of impacts that global warming is causing (or would be likely to cause under a given scenario for emissions), but less certainty on the exact timing and intensity of these impacts.  

When Rising Seas Hit Home

For instance, on the issue of sea level rise, we know with certainty that it will happen – it is already happening – and projections under different scenarios give us a range of possible rise.  We don’t know an exact value, however, for future sea level rise, because in large part it is dependent on the rate of future emissions, which is unknown.  

If emissions continue in a “business as usual” fashion, the sea level rise will be closer to the higher range of projections. But if we significantly reduce emissions, the rise will be closer to the lower levels of projections.

The same is true for how much warming will actually happen, or how much land-based ice and glaciers will melt. All these things are already happening, but future rates are not known because they, too, depend on the rate of future emissions. What scientists can calculate quite confidently is a narrow range of outcomes within a given scenario—meaning the likely highest and lowest values  if we continue on a certain path of emissions. This information is critical to making smart collective choices and for planning for the future.

Uncertainties are not a reason to delay action on climate change. Quite the contrary: those uncertainties are really a consequence of our collective choices, and a risk we must prepare for.

You can think about it like car insurance: everyone hopes they won’t be in a car accident but have accident insurance anyway, even though the likelihood is very low.

Climate adaptation and climate risk reduction are “insurance” against the effects of climate change, which in contrast are NOT low-probability events, but highly likely and predicted with high levels of certainty under specific conditions.  Being prepared for these scenarios is simply smart planning. Nobody wants to be caught unaware and unprepared.

Killer Heat in the United States

Consensus worldwide.

Many scientific societies and academies have released statements and studies that highlight the overwhelming consensus on climate change science.

American Association for the Advancement of Science: AAAS Reaffirms the Reality of Human-Caused Climate Change “Observations throughout the world make it clear that climate change is occurring, and rigorous scientific research concludes that the greenhouse gases emitted by human activities are the primary driver. This conclusion is based on multiple independent lines of evidence and the vast body of peer-reviewed science.” (June 2016)

American Chemical Society : Statement on Global Climate Change “The Earth’s climate is changing in response to increasing concentrations of greenhouse gases (GHGs) and particulate matter in the atmosphere, largely as the result of human activities. … Unmitigated climate change will lead to increases in extreme weather events and will cause significant sea level rise, causing property damage and population displacement. It also will continue to degrade ecosystems and natural resources, affecting food and water availability and human health, further burdening economies and societies. Continued uncontrolled GHG emissions will accelerate and compound the effects and risks of climate change well into the future.” (2016)

American Geophysical Union : Human-induced Climate Change Requires Urgent Action.

"Extensive, independent observations confirm the reality of global warming. These observations show large-scale increases in air and sea temperatures, sea level, and atmospheric water vapor; they document decreases in the extent of mountain glaciers, snow cover, permafrost, and Arctic sea ice. These changes are broadly consistent with long-understood physics and predictions of how the climate system is expected to respond to human-caused increases in greenhouse gases. The changes are inconsistent with explanations of climate change that rely on known natural influences.”(December 2003, revised and reaffirmed December 2007, February 2012, August 2013)

American Meteorological Society: Climate Change: An Information Statement of the American Meteorological Society "It is clear from extensive scientific evidence that the dominant cause of the rapid change in climate of the past half century is human-induced increases in the amount of atmospheric greenhouse gases, including carbon dioxide (CO2), chlorofluorocarbons, methane, and nitrous oxide.” (August 2012)

American Physical Society : Statement on Earth’s Changing Climate

"While natural sources of climate variability are significant, multiple lines of evidence indicate that human influences have had an increasingly dominant effect on global climate warming observed since the mid-twentieth century. Although the magnitudes of future effects are uncertain, human influences on the climate are growing." (November 2015)

Geological Society of America : Position Statement on Climate Change "Scientific advances have greatly reduced previous uncertainties about recent global warming. Ground-station measurements have shown a warming trend of ~0.85 °C since 1880, a trend consistent with (1) retreat of northern hemisphere snow and Arctic sea ice; (2) greater heat storage in the ocean; (3) retreat of most mountain glaciers; (4) an ongoing rise in global sea level; and (5) proxy reconstructions of temperature change over past centuries from archives that include ice cores, tree rings, lake sediments, boreholes, cave deposits, and corals." (October 2006; revised April 2010, March 2013, April 2015).

Intergovernmental Panel on Climate Change: Synthesis Report Summary for Policymakers

“Human influence on the climate system is clear, and recent anthropogenic emissions of green-house gases are the highest in history. Recent climate changes have had widespread impacts on human and natural systems. “Warming of the climate system is unequivocal, and since the 1950s, many of the observed changes are unprecedented over decades to millennia. The atmosphere and ocean have warmed, the amounts of snow and ice have diminished, and sea level has risen.” (2014)

International academies joint statement: Global response to climate change

“The world’s climate is changing, and the impacts are already being observed. Changing agricultural conditions, ocean warming and acidification, rising sea levels, and increased frequency and intensity of many extreme weather events are impacting infrastructure, environmental assets and human health.” (2018, African Academy of Sciences and the national academies of science of the United Kingdom, Australia, Bangladesh, Botswana, Canada, India, Mauritius, Mozambique, Nigeria, Pakistan, New Zealand, Cyprus, Singapore, Sri Lanka, South Africa, Scotland, Zimbabwe, Kenya, Zambia, Malaysia, Cameroon). 

US Global Change Research Program:  Highlights of the Findings of the U.S. Global Change Research Program Climate Science Special Report

“Based on extensive evidence, … it is extremely likely that human activities, especially emissions of greenhouse gases, are the dominant cause of the observed warming since the mid-20th century. For the warming over the last century, there is no convincing alternative explanation supported by the extent of the observational evidence.

“In addition to warming, many other aspects of global climate are changing, primarily in response to human activities. Thousands of studies conducted by researchers around the world have documented changes in surface, atmospheric, and oceanic temperatures; melting glaciers; diminishing snow cover; shrinking sea ice; rising sea levels; ocean acidification; and increasing atmospheric water vapor.” (November 2017)

U.S. National Academy of Sciences : Understanding and Responding to Climate Change "The scientific understanding of climate change is now sufficiently clear to begin taking steps to prepare for climate change and to slow it." (2008)

Related resources

Water, Water Every Where

Looming Deadlines for Coastal Resilience

Follow the Money

The Science and Climate Litigation Webinar Series

We use cookies to improve your experience. By continuing, you accept our use of cookies. Learn more .

Support our work

Other ways to give.

• Honor & memory
• Become a member
• Give monthly
• Make a planned gift
• Gift memberships

Your donation at work

• Student Opportunities

About Hoover

Located on the campus of Stanford University and in Washington, DC, the Hoover Institution is the nation’s preeminent research center dedicated to generating policy ideas that promote economic prosperity, national security, and democratic governance. 

• The Hoover Story
• Hoover Timeline & History
• Mission Statement
• Vision of the Institution Today
• Key Focus Areas
• About our Fellows
• Research Programs
• Annual Reports
• Hoover in DC
• Fellowship Opportunities
• Visit Hoover
• David and Joan Traitel Building & Rental Information
• Newsletter Subscriptions
• Connect With Us

Hoover scholars form the Institution’s core and create breakthrough ideas aligned with our mission and ideals. What sets Hoover apart from all other policy organizations is its status as a center of scholarly excellence, its locus as a forum of scholarly discussion of public policy, and its ability to bring the conclusions of this scholarship to a public audience.

• Peter Berkowitz
• Ross Levine
• Michael McFaul
• Timothy Garton Ash
• China's Global Sharp Power Project
• Economic Policy Group
• History Working Group
• Hoover Education Success Initiative
• National Security Task Force
• National Security, Technology & Law Working Group
• Middle East and the Islamic World Working Group
• Military History/Contemporary Conflict Working Group
• Renewing Indigenous Economies Project
• State & Local Governance
• Strengthening US-India Relations
• Technology, Economics, and Governance Working Group
• Taiwan in the Indo-Pacific Region

Books by Hoover Fellows

Economics Working Papers

Hoover Education Success Initiative | The Papers

• Hoover Fellows Program
• National Fellows Program
• Student Fellowship Program
• Veteran Fellowship Program
• Congressional Fellowship Program
• Media Fellowship Program
• Silas Palmer Fellowship
• Economic Fellowship Program

Throughout our over one-hundred-year history, our work has directly led to policies that have produced greater freedom, democracy, and opportunity in the United States and the world.

• Determining America’s Role in the World
• Answering Challenges to Advanced Economies
• Empowering State and Local Governance
• Revitalizing History
• Confronting and Competing with China
• Revitalizing American Institutions
• Reforming K-12 Education
• Understanding Public Opinion
• Understanding the Effects of Technology on Economics and Governance
• Energy & Environment
• Health Care
• Immigration
• International Affairs
• Key Countries / Regions
• Law & Policy
• Politics & Public Opinion
• Science & Technology
• Security & Defense
• State & Local
• Books by Fellows
• Published Works by Fellows
• Working Papers
• Congressional Testimony
• Hoover Press
• PERIODICALS
• The Caravan
• China's Global Sharp Power
• Economic Policy
• History Lab
• Hoover Education
• Global Policy & Strategy
• Middle East and the Islamic World
• Military History & Contemporary Conflict
• Renewing Indigenous Economies
• State and Local Governance
• Technology, Economics, and Governance

Hoover scholars offer analysis of current policy challenges and provide solutions on how America can advance freedom, peace, and prosperity.

• China Global Sharp Power Weekly Alert
• Email newsletters
• Hoover Daily Report
• Subscription to Email Alerts
• Periodicals
• California on Your Mind
• Defining Ideas
• Hoover Digest
• Video Series
• Uncommon Knowledge
• Battlegrounds
• GoodFellows
• Hoover Events
• Capital Conversations
• Hoover Book Club
• AUDIO PODCASTS
• Matters of Policy & Politics
• Economics, Applied
• Free Speech Unmuted
• Secrets of Statecraft
• Capitalism and Freedom in the 21st Century
• Libertarian
• Library & Archives

Support Hoover

Learn more about joining the community of supporters and scholars working together to advance Hoover’s mission and values.

What is MyHoover?

MyHoover delivers a personalized experience at  Hoover.org . In a few easy steps, create an account and receive the most recent analysis from Hoover fellows tailored to your specific policy interests.

Watch this video for an overview of MyHoover.

Log In to MyHoover

Forgot Password

Don't have an account? Sign up

Have questions? Contact us

• Support the Mission of the Hoover Institution
• Subscribe to the Hoover Daily Report
• Follow Hoover on Social Media

Make a Gift

Your gift helps advance ideas that promote a free society.

• About Hoover Institution
• Meet Our Fellows
• Focus Areas
• Research Teams
• Library & Archives

Library & archives

Events, news & press.

Flawed Climate Models

The relationship between CO 2 and temperature is more complicated than the polemics suggest.

The atmosphere is about 0.8˚ Celsius warmer than it was in 1850. Given that the atmospheric concentration of carbon dioxide has risen 40 percent since 1750 and that CO 2 is a greenhouse gas, a reasonable hypothesis is that the increase in CO 2 has caused, and is causing, global warming.

But a hypothesis is just that. We have virtually no ability to run controlled experiments, such as raising and lowering CO 2 levels in the atmosphere and measuring the resulting change in temperatures. What else can we do? We can build elaborate computer models that use physics to calculate how energy flows into, through, and out of our planet’s land, water, and atmosphere. Indeed, such models have been created and are frequently used today to make dire predictions about the fate of our Earth.

The problem is that these models have serious limitations that drastically limit their value in making predictions and in guiding policy. Specifically, three major problems exist. They are described below, and each one alone is enough to make one doubt the predictions. All three together deal a devastating blow to the forecasts of the current models.

• Measurement Error

Imagine that you’re timing a high school track athlete running 400 meters at the beginning of the school year, and you measure 56 seconds with your handheld stopwatch that reads to ±0.01 seconds. Imagine also that your reaction time is ±0.2 seconds. With your equipment, you can measure an improvement to 53 seconds by the end of the year. The difference between the two times is far larger than the resolution of the stopwatch combined with your imperfect reaction time, allowing you to conclude that the runner is indeed now faster. To get an idea of this runner’s improvement, you calculate a trend of 0.1 seconds per week (3 seconds in 30 weeks). But if you try to retest this runner after half a week, trying to measure the expected 0.05-second improvement, you will run into a problem. Can you measure such a small difference with the instrumentation at hand? No. There’s no point in even trying because you’ll have no way of discovering if the runner is faster: the size of what you are trying to measure is smaller than the size of the errors in your measurements.

Scientists present measurement error by describing the range around their measurements. They might, for example, say that a temperature is 20˚C ±0.5˚C. The temperature is probably 20.0˚C, but it could reasonably be as high as 20.5˚C or as low as 19.5˚C.

Now consider the temperatures that are recorded by weather stations around the world.

Patrick Frank is a scientist at the Stanford Synchrotron Radiation Lightsource (SSRL), part of the SLAC National Accelerator Laboratory at Stanford University. Frank has published papers that explain how the errors in temperatures recorded by weather stations have been incorrectly handled. Temperature readings, he finds, have errors over twice as large as generally recognized. Based on this, Frank stated, in a 2011 article in Energy & Environment , “…the 1856–2004 global surface air temperature anomaly with its 95% confidence interval is 0.8˚C ± 0.98˚C.” The error bars are wider than the measured increase. It looks as if there’s an upward temperature trend, but we can’t tell definitively. We cannot reject the hypothesis that the world’s temperature has not changed at all.

• The Sun’s Energy

Climate models are used to assess the CO 2 -global warming hypothesis and to quantify the human-caused CO 2 “fingerprint.”

How big is the human-caused CO 2 fingerprint compared to other uncertainties in our climate model? For tracking energy flows in our model, we use watts per square meter (Wm –2 ). The sun’s energy that reaches the Earth’s atmosphere provides 342 Wm –2 —an average of day and night, poles and equator—keeping it warm enough for us to thrive. The estimated extra energy from excess CO 2 —the annual anthropogenic greenhouse gas contribution—is far smaller, according to Frank, at 0.036 Wm –2 , or 0.01 percent of the sun’s energy. If our estimate of the sun’s energy were off by more than 0.01 percent, that error would swamp the estimated extra energy from excess CO 2 . Unfortunately, the sun isn’t the only uncertainty we need to consider.

• Cloud Errors

Clouds reflect incoming radiation and also trap it as it is outgoing. A world entirely encompassed by clouds would have dramatically different atmospheric temperatures than one devoid of clouds. But modeling clouds and their effects has proven difficult. The Intergovernmental Panel on Climate Change (IPCC), the established global authority on climate change, acknowledges this in its most recent Assessment report , from 2013:

The simulation of clouds in climate models remains challenging. There is very high confidence that uncertainties in cloud processes explain much of the spread in modelled climate sensitivity. [bold and italics in original]

What is the net effect of cloudiness? Clouds lead to a cooler atmosphere by reducing the sun’s net energy by approximately 28 Wm –2 . Without clouds, more energy would reach the ground and our atmosphere would be much warmer.  Why are clouds hard to model? They are amorphous; they reside at different altitudes and are layered on top of each other, making them hard to discern; they aren’t solid; they come in many different types; and scientists don’t fully understand how they form. As a result, clouds are modeled poorly. This contributes an average uncertainty of ±4.0 Wm –2 to the atmospheric thermal energy budget of a simulated atmosphere during a projection of global temperature. This thermal uncertainty is 110 times as large as the estimated annual extra energy from excess CO 2 . If our climate model’s calculation of clouds were off by just 0.9 percent—0.036 is 0.9 percent of 4.0—that error would swamp the estimated extra energy from excess CO 2 . The total combined errors in our climate model are estimated be about 150 Wm –2 , which is over 4,000 times as large as the estimated annual extra energy from higher CO 2 concentrations. Can we isolate such a faint signal?

In our track athlete example, this is equivalent to having a reaction time error of ±0.2 seconds while trying to measure a time difference of 0.00005 seconds between any two runs. How can such a slight difference in time be measured with such overwhelming error bars? How can the faint CO 2 signal possibly be detected by climate models with such gigantic errors?

Other Complications

Even the relationship between CO 2 concentrations and temperature is complicated.

The glacial record shows geological periods with rising CO 2 and global cooling and periods with low levels of atmospheric CO 2 and global warming. Indeed, according to a 2001 article in Climate Research by astrophysicist and geoscientist Willie Soon and his colleagues, “atmospheric CO 2 tends to follow rather than lead temperature and biosphere changes.”

A large proportion of the warming that occurred in the 20th century occurred in the first half of the century, when the amount of anthropogenic CO 2 in the air was one quarter of the total amount there now. The rate of warming then was very similar to the rate of warming recently. We can’t have it both ways. The current warming can’t be unambiguously caused by anthropogenic CO 2 emissions if an earlier period experienced the same type of warming without the offending emissions.

Climate Model Secret Sauce

It turns out that climate models aren’t “plug and chug.” Numerous inputs are not the direct result of scientific studies; researchers need to “discover” them through parameter adjustment, or tuning, as it is called. If a climate model uses a grid of 25x25-kilometer boxes to divide the atmosphere and oceans into manageable chunks, storm clouds and low marine clouds off the California coast will be too small to model directly. Instead, according to a 2016 Science article by journalist Paul Voosen, modelers need to tune for cloud formation in each key grid based on temperature, atmospheric stability, humidity, and the presence of mountains. Modelers continue tuning climate models until they match a known 20th century temperature or precipitation record. And yet, at that point, we will have to ask whether these models are more subjective than objective. If a model shows a decline in Arctic sea ice, for instance—and we know that Arctic sea ice has, in fact, declined—is the model telling us something new or just regurgitating its adjustments?

Climate Model Errors

Before we put too much credence in any climate model, we need to assess its predictions. The following points highlight some of the difficulties of current models.

Vancouver, British Columbia, warmed by a full degree in the first 20 years of the 20th century, then cooled by two degrees over the next 40 years, and then warmed to the end the century, ending almost where it started. None of the six climate models tested by the IPCC reproduced this pattern. Further, according to scientist Patrick Frank in a 2015 article in Energy & Environment , the projected temperature trends of the models, which all employed the same theories and historical data, were as far apart as 2.5˚C.

According to a 2002 article by climate scientists Vitaly Semenov and Lennart Bengtsson in Climate Dynamics , climate models have done a poor job of matching known global rainfall totals and patterns.

Climate models have been subjected to “perfect model tests,” in which the they were used to project a reference climate and then, with some minor tweaks to initial conditions, recreate temperatures in that same reference climate. This is basically asking a model to do the same thing twice, a task for which it should be ideally suited. In these tests, Frank found, the results in the first year correlated very well between the two runs, but years 2-9 showed such poor correlation that the results could have been random. Failing a perfect model test shows that the results aren’t stable and suggests a fundamental inability of the models to predict the climate.

The ultimate test for a climate model is the accuracy of its predictions. But the models predicted that there would be much greater warming between 1998 and 2014 than actually happened. If the models were doing a good job, their predictions would cluster symmetrically around the actual measured temperatures. That was not the case here; a mere 2.4 percent of the predictions undershot actual temperatures and 97.6 percent overshot, according to Cato Institute climatologist Patrick Michaels, former MIT meteorologist Richard Lindzen, and Cato Institute climate researcher Chip Knappenberger. Climate models as a group have been “running hot,” predicting about 2.2 times as much warming as actually occurred over 1998–2014. Of course, this doesn’t mean that no warming is occurring, but, rather, that the models’ forecasts were exaggerated.

Conclusions

If someone with a hand-held stopwatch tells you that a runner cut his time by 0.00005 seconds, you should be skeptical. If someone with a climate model tells you that a 0.036 Wm –2 CO 2 signal can be detected within an environment of 150 Wm –2 error, you should be just as skeptical.

As Willie Soon and his coauthors found, “Our current lack of understanding of the Earth’s climate system does not allow us to determine reliably the magnitude of climate change that will be caused by anthropogenic CO 2 emissions, let alone whether this change will be for better or for worse.”

View the discussion thread.

Join the Hoover Institution’s community of supporters in ideas advancing freedom.

56 Million Years Later: Could Today’s Warming Mirror a Past Climate Catastrophe?

A team from UNIGE analyzed 56 million-year-old sediments to determine the rise in soil erosion due to global warming, which is associated with significant flooding.

Around 56 million years ago, Earth underwent significant and swift climate warming, likely triggered by greenhouse gases emitted from volcanic eruptions. Researchers from the University of Geneva (UNIGE) have studied sediments from this era to evaluate the effects of this climate change on the environment, focusing particularly on soil erosion.

The study revealed a four-fold increase in soil erosion due to heavy rainfall and river flooding. These results suggest that current warming could have a similar effect over time, significantly increasing flood risks. They are published in the journal Geology .

Because of its similarities to current warming, the Paleocene-Eocene Thermal Maximum is closely studied to understand how the Earth’s environment reacts to a global rise in temperature. Occurring 56 million years ago, this episode saw the Earth warm by 5 to 8°C within 20,000 years, a very short time at the geological scale. It lasted for 200,000 years, causing major disruption to flora and fauna. According to recent IPCC reports, the Earth is now on the brink of a similar warming.

Scientists are analyzing sediments from this period to obtain a more accurate ‘‘picture’’ of this past warming and its consequences and to make predictions for the future. These natural deposits are the result of soil erosion by water and wind. They were carried by rivers into the oceans. Now preserved in rocks, these geological archives provide valuable information about our past, but also our future.

Four times more erosion

‘‘Our starting hypothesis was that, during such a period of warming, the seasonality and intensity of rainfall increases. This alters the dynamics of river flooding and intensifies sediment transport from the mountains to the oceans. Our objective is to test this hypothesis and, above all, to better quantify this change,’’ explains Marine Prieur, a doctoral student in the Section of Earth and Environmental Sciences at the UNIGE Faculty of Science, and first author of this study funded by the European Union’s Horizon 2020 program.

The research team studied a specific type of sediment, Microcodium grains, collected in the Pyrenees (around 20kg). These prisms of calcite, no more than a millimeter in size, were specifically formed at this period around the roots of plants, in the soil. However, they are also found in marine sediments, proving their erosion on the continent. Therefore, Microcodium grains are a good indicator of the intensity of soil erosion on the continents.

‘‘By quantifying the abundance of Microcodium grains in marine sediments, based on samples taken from the Spanish Pyrenees, which were submerged during the Palaeocene-Eocene, we have shown a four-fold increase in soil erosion on the continent during the climate change that occurred 56 million years ago,’’ reveals Sébastien Castelltort, a full professor in the Section of Earth and Environmental Sciences at the UNIGE Faculty of Science, who led the study.

Human action will exacerbate the phenomenon

This discovery highlights the significant impact of global warming on soil erosion through the intensification of rainfall during storm events and the increase in river flooding. This is an indicator of heavy flooding. ‘‘These results relate specifically to this area of the Pyrenees, and each geographical zone is dependent on certain unique factors. However, increased sedimentary input in the Paleocene-Eocene strata is observed worldwide. It is, therefore, a global phenomenon, on an Earth-wide scale, during a significant warming event,’’ points out Marine Prieur.

These results provide new information that can be incorporated into predictions about our future climate. In particular, to better assess the risks of flooding and soil collapse in populated areas. ‘‘We need to bear in mind that this increase in erosion has occurred naturally, under the effect of global warming alone. Today, to predict what lies ahead, we must also consider the impact of human action, such as deforestation, which amplifies various phenomena, including erosion,’’ conclude the scientists.

Reference: “Fingerprinting enhanced floodplain reworking during the Paleocene–Eocene Thermal Maximum in the Southern Pyrenees (Spain): Implications for channel dynamics and carbon burial” by Marine Prieur, Alexander C. Whittaker, Perach Nuriel, Rocío Jaimes-Gutierrez, Eduardo Garzanti, Marta Roigé, Tor O. Sømme, Fritz Schlunegger and Sébastien Castelltort, 23 May 2024, Geology . DOI: 10.1130/G52180.1

Related Articles

Arctic sea ice loss is impacting atlantic ocean water circulation system, temperature of ancient seas may shape global climate, siberian traps likely triggered mass extinction, nasa study reveals oceans temporarily hide global warming, surfer-shaped waves discovered at the boundaries of near-earth space, nasa study provides new estimates for the global water cycle, mit research reveals natural impediment to long-term sequestration of carbon dioxide, millions of tons of seafloor methane could be released, nasa to launch spacecraft to keep track of global carbon dioxide.

The other side of the coin is that the warm Eocene epoch was a time of rapid evolution of mammals, with diversification and increase in size, including the development of primates, from which humans are derived. The PETM is blamed on vulcanism, yet during the following epoch, the Oligocene, it was clearly volcanically more active, with volcanoes burying the Eocene landscape in the western USA in ash and lava, and an extinction event happened at the boundary between the Eocene and Oligocene, resulting in part from the Oligocene cooling. The Eocene in California is characterized by the development of deep (200-300′) lateritic soils obtained by chemical weathering, and several south-draining rivers, some rivaling the present day Mississippi. The large rivers had low grades and were meandering like the modern Mississippi, moving primarily fine-grained material. The animals and plants living on the flood plains were obviously adapted to that environment. Worrying about what might be in the future, using a model of a salubrious climate that fostered biological diversity does not result in a convincing argument for concern. When climate or landforms change, there is always disruption to the status quo, which drives evolution. Unless one can present a convincing argument that evolution is bad for the biosphere, and humans should do whatever they can to prevent it, the concern falls flat. The only thing that has been constant on Earth is change.

How many millions? Is this data posted as fact, observed, replicable, and documented? Looks to me like the clang of the Uniformitarian bell, the presuppositional god of evolutionary ancient age pandering. The truth is an Occams Razor reality. Here, change is inevitable, both slow and rapid. Man want change over millions of years, not supported by observed evidence.

Funny that. All this time the scientific documentaries kept telling us that historic volcanic eruptions caused global cooling. Guess they will have to burn the BBC archive now.

I suspect that the effects are variable, depending on where the eruptions take place, how high the ejecta reach, and the amount of H2O, SO2, and CO2 that are present, which varies with the kind of rock the magma is composed of.

Save my name, email, and website in this browser for the next time I comment.

Type above and press Enter to search. Press Esc to cancel.

What’s the difference between climate change and global warming?

The terms “global warming” and “climate change” are sometimes used interchangeably, but "global warming" is only one aspect of climate change.

“Global warming” refers to the long-term warming of the planet. Global temperature shows a well-documented rise since the early 20th century and most notably since the late 1970s. Worldwide since 1880, the average surface temperature has risen about 1 ° C (about 2 ° F), relative to the mid-20th century baseline (of 1951-1980). This is on top of about an additional 0.15 ° C of warming from between 1750 and 1880.

“Climate change” encompasses global warming, but refers to the broader range of changes that are happening to our planet. These include rising sea levels; shrinking mountain glaciers; accelerating ice melt in Greenland, Antarctica and the Arctic; and shifts in flower/plant blooming times. These are all consequences of warming, which is caused mainly by people burning fossil fuels and putting out heat-trapping gases into the air.

• Overview: Weather, Global Warming and Climate Change

Discover More Topics From NASA

Explore Earth Science

Earth Science in Action

Earth Science Data

Facts About Earth

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 22 July 2024

Neural general circulation models for weather and climate

• Dmitrii Kochkov   ORCID: orcid.org/0000-0003-3846-4911 1   na1 ,
• Janni Yuval   ORCID: orcid.org/0000-0001-7519-0118 1   na1 ,
• Ian Langmore 1   na1 ,
• Peter Norgaard 1   na1 ,
• Jamie Smith 1   na1 ,
• Griffin Mooers 1 ,
• Milan Klöwer 2 ,
• James Lottes 1 ,
• Stephan Rasp 1 ,
• Peter Düben   ORCID: orcid.org/0000-0002-4610-3326 3 ,
• Sam Hatfield 3 ,
• Peter Battaglia 4 ,
• Alvaro Sanchez-Gonzalez 4 ,
• Matthew Willson   ORCID: orcid.org/0000-0002-8730-1927 4 ,
• Michael P. Brenner 1 , 5 &
• Stephan Hoyer   ORCID: orcid.org/0000-0002-5207-0380 1   na1  

Nature volume  632 ,  pages 1060–1066 ( 2024 ) Cite this article

53k Accesses

3 Citations

678 Altmetric

Metrics details

• Atmospheric dynamics
• Climate and Earth system modelling
• Computational science

General circulation models (GCMs) are the foundation of weather and climate prediction 1 , 2 . GCMs are physics-based simulators that combine a numerical solver for large-scale dynamics with tuned representations for small-scale processes such as cloud formation. Recently, machine-learning models trained on reanalysis data have achieved comparable or better skill than GCMs for deterministic weather forecasting 3 , 4 . However, these models have not demonstrated improved ensemble forecasts, or shown sufficient stability for long-term weather and climate simulations. Here we present a GCM that combines a differentiable solver for atmospheric dynamics with machine-learning components and show that it can generate forecasts of deterministic weather, ensemble weather and climate on par with the best machine-learning and physics-based methods. NeuralGCM is competitive with machine-learning models for one- to ten-day forecasts, and with the European Centre for Medium-Range Weather Forecasts ensemble prediction for one- to fifteen-day forecasts. With prescribed sea surface temperature, NeuralGCM can accurately track climate metrics for multiple decades, and climate forecasts with 140-kilometre resolution show emergent phenomena such as realistic frequency and trajectories of tropical cyclones. For both weather and climate, our approach offers orders of magnitude computational savings over conventional GCMs, although our model does not extrapolate to substantially different future climates. Our results show that end-to-end deep learning is compatible with tasks performed by conventional GCMs and can enhance the large-scale physical simulations that are essential for understanding and predicting the Earth system.

Similar content being viewed by others

Accurate medium-range global weather forecasting with 3D neural networks

Deep learning for twelve hour precipitation forecasts

Skilful predictions of the Asian summer monsoon one year ahead

Solving the equations for Earth’s atmosphere with general circulation models (GCMs) is the basis of weather and climate prediction 1 , 2 . Over the past 70 years, GCMs have been steadily improved with better numerical methods and more detailed physical models, while exploiting faster computers to run at higher resolution. Inside GCMs, the unresolved physical processes such as clouds, radiation and precipitation are represented by semi-empirical parameterizations. Tuning GCMs to match historical data remains a manual process 5 , and GCMs retain many persistent errors and biases 6 , 7 , 8 . The difficulty of reducing uncertainty in long-term climate projections 9 and estimating distributions of extreme weather events 10 presents major challenges for climate mitigation and adaptation 11 .

Recent advances in machine learning have presented an alternative for weather forecasting 3 , 4 , 12 , 13 . These models rely solely on machine-learning techniques, using roughly 40 years of historical data from the European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis v5 (ERA5) 14 for model training and forecast initialization. Machine-learning methods have been remarkably successful, demonstrating state-of-the-art deterministic forecasts for 1- to 10-day weather prediction at a fraction of the computational cost of traditional models 3 , 4 . Machine-learning atmospheric models also require considerably less code, for example GraphCast 3 has 5,417 lines versus 376,578 lines for the National Oceanic and Atmospheric Administration’s FV3 atmospheric model 15 (see Supplementary Information section  A for details).

Nevertheless, machine-learning approaches have noteworthy limitations compared with GCMs. Existing machine-learning models have focused on deterministic prediction, and surpass deterministic numerical weather prediction in terms of the aggregate metrics for which they are trained 3 , 4 . However, they do not produce calibrated uncertainty estimates 4 , which is essential for useful weather forecasts 1 . Deterministic machine-learning models using a mean-squared-error loss are rewarded for averaging over uncertainty, producing unrealistically blurry predictions when optimized for multi-day forecasts 3 , 13 . Unlike physical models, machine-learning models misrepresent derived (diagnostic) variables such as geostrophic wind 16 . Furthermore, although there has been some success in using machine-learning approaches on longer timescales 17 , 18 , these models have not demonstrated the ability to outperform existing GCMs.

Hybrid models that combine GCMs with machine learning are appealing because they build on the interpretability, extensibility and successful track record of traditional atmospheric models 19 , 20 . In the hybrid model approach, a machine-learning component replaces or corrects the traditional physical parameterizations of a GCM. Until now, the machine-learning component in such models has been trained ‘offline’, by learning parameterizations independently of their interaction with dynamics. These components are then inserted into an existing GCM. The lack of coupling between machine-learning components and the governing equations during training potentially causes serious problems, such as instability and climate drift 21 . So far, hybrid models have mostly been limited to idealized scenarios such as aquaplanets 22 , 23 . Under realistic conditions, machine-learning corrections have reduced some biases of very coarse GCMs 24 , 25 , 26 , but performance remains considerably worse than state-of-the-art models.

Here we present NeuralGCM, a fully differentiable hybrid GCM of Earth’s atmosphere. NeuralGCM is trained on forecasting up to 5-day weather trajectories sampled from ERA5. Differentiability enables end-to-end ‘online training’ 27 , with machine-learning components optimized in the context of interactions with the governing equations for large-scale dynamics, which we find enables accurate and stable forecasts. NeuralGCM produces physically consistent forecasts with accuracy comparable to best-in-class models across a range of timescales, from 1- to 15-day weather to decadal climate prediction.

Neural GCMs

A schematic of NeuralGCM is shown in Fig. 1 . The two key components of NeuralGCM are a differentiable dynamical core for solving the discretized governing dynamical equations and a learned physics module that parameterizes physical processes with a neural network, described in full detail in Methods , Supplementary Information sections  B and C , and Supplementary Table 1 . The dynamical core simulates large-scale fluid motion and thermodynamics under the influence of gravity and the Coriolis force. The learned physics module (Supplementary Fig. 1 ) predicts the effect of unresolved processes, such as cloud formation, radiative transport, precipitation and subgrid-scale dynamics, on the simulated fields using a neural network.

a , Overall model structure, showing how forcings F t , noise z t (for stochastic models) and inputs y t are encoded into the model state x t . The model state is fed into the dynamical core, and alongside forcings and noise into the learned physics module. This produces tendencies (rates of change) used by an implicit–explicit ordinary differential equation (ODE) solver to advance the state in time. The new model state x t +1 can then be fed back into another time step, or decoded into model predictions. b , The learned physics module, which feeds data for individual columns of the atmosphere into a neural network used to produce physics tendencies in that vertical column.

The differentiable dynamical core in NeuralGCM allows an end-to-end training approach, whereby we advance the model multiple time steps before employing stochastic gradient descent to minimize discrepancies between model predictions and reanalysis (Supplementary Information section  G.2 ). We gradually increase the rollout length from 6 hours to 5 days (Supplementary Information section  G and Supplementary Table 5 ), which we found to be critical because our models are not accurate for multi-day prediction or stable for long rollouts early in training (Supplementary Information section  H.6.2 and Supplementary Fig. 23 ). The extended back-propagation through hundreds of simulation steps enables our neural networks to take into account interactions between the learned physics and the dynamical core. We train deterministic and stochastic NeuralGCM models, each of which uses a distinct training protocol, described in full detail in Methods and Supplementary Table 4 .

We train a range of NeuralGCM models at horizontal resolutions with grid spacing of 2.8°, 1.4° and 0.7° (Supplementary Fig. 7 ). We evaluate the performance of NeuralGCM at a range of timescales appropriate for weather forecasting and climate simulation. For weather, we compare against the best-in-class conventional physics-based weather models, ECMWF’s high-resolution model (ECMWF-HRES) and ensemble prediction system (ECMWF-ENS), and two of the recent machine-learning-based approaches, GraphCast 3 and Pangu 4 . For climate, we compare against a global cloud-resolving model and Atmospheric Model Intercomparison Project (AMIP) runs.

Medium-range weather forecasting

Our evaluation set-up focuses on quantifying accuracy and physical consistency, following WeatherBench2 12 . We regrid all forecasts to a 1.5° grid using conservative regridding, and average over all 732 forecasts made at noon and midnight UTC in the year 2020, which was held-out from training data for all machine-learning models. NeuralGCM, GraphCast and Pangu compare with ERA5 as the ground truth, whereas ECMWF-ENS and ECMWF-HRES compare with the ECMWF operational analysis (that is, HRES at 0-hour lead time), to avoid penalizing the operational forecasts for different biases than ERA5.

Model accuracy

We use ECMWF’s ensemble (ENS) model as a reference baseline as it achieves the best performance across the majority of lead times 12 . We assess accuracy using (1) root-mean-squared error (RMSE), (2) root-mean-squared bias (RMSB), (3) continuous ranked probability score (CRPS) and (4) spread-skill ratio, with the results shown in Fig. 2 . We provide more in-depth evaluations including scorecards, metrics for additional variables and levels and maps in Extended Data Figs. 1 and 2 , Supplementary Information section  H and Supplementary Figs. 9 – 22 .

a , c , RMSE ( a ) and RMSB ( c ) for ECMWF-ENS, ECMWF-HRES, NeuralGCM-0.7°, NeuralGCM-ENS, GraphCast 3 and Pangu 4 on headline WeatherBench2 variables, as a percentage of the error of ECMWF-ENS. Deterministic and stochastic models are shown in solid and dashed lines respectively. e , g , CRPS relative to ECMWF-ENS ( e ) and spread-skill ratio for the ENS and NeuralGCM-ENS models ( g ). b , d , f , h , Spatial distributions of RMSE ( b ), bias ( d ), CRPS ( f ) and spread-skill ratio ( h ) for NeuralGCM-ENS and ECMWF-ENS models for 10-day forecasts of specific humidity at 700 hPa. Spatial plots of RMSE and CRPS show skill relative to a probabilistic climatology 12 with an ensemble member for each of the years 1990–2019. The grey areas indicate regions where climatological surface pressure on average is below 700 hPa.

Deterministic models that produce a single weather forecast for given initial conditions can be compared effectively using RMSE skill at short lead times. For the first 1–3 days, depending on the atmospheric variable, RMSE is minimized by forecasts that accurately track the evolution of weather patterns. At this timescale we find that NeuralGCM-0.7° and GraphCast achieve best results, with slight variations across different variables (Fig. 2a ). At longer lead times, RMSE rapidly increases owing to chaotic divergence of nearby weather trajectories, making RMSE less informative for deterministic models. RMSB calculates persistent errors over time, which provides an indication of how models would perform at much longer lead times. Here NeuralGCM models also compare favourably against previous approaches (Fig. 2c ), with notably much less bias for specific humidity in the tropics (Fig. 2d ).

Ensembles are essential for capturing intrinsic uncertainty of weather forecasts, especially at longer lead times. Beyond about 7 days, the ensemble means of ECMWF-ENS and NeuralGCM-ENS forecasts have considerably lower RMSE than the deterministic models, indicating that these models better capture the average of possible weather. A better metric for ensemble models is CRPS, which is a proper scoring rule that is sensitive to full marginal probability distributions 28 . Our stochastic model (NeuralGCM-ENS) running at 1.4° resolution has lower error compared with ECMWF-ENS across almost all variables, lead times and vertical levels for ensemble-mean RMSE, RSMB and CRPS (Fig. 2a,c,e and Supplementary Information section  H ), with similar spatial patterns of skill (Fig. 2b,f ). Like ECMWF-ENS, NeuralGCM-ENS has a spread-skill ratio of approximately one (Fig. 2d ), which is a necessary condition for calibrated forecasts 29 .

An important characteristic of forecasts is their resemblance to realistic weather patterns. Figure 3 shows a case study that illustrates the performance of NeuralGCM on three types of important weather phenomenon: tropical cyclones, atmospheric rivers and the Intertropical Convergence Zone. Figure 3a shows that all the machine-learning models make significantly blurrier forecasts than the source data ERA5 and physics-based ECMWF-HRES forecast, but NeuralCGM-0.7° outperforms the pure machine-learning models, despite its coarser resolution (0.7° versus 0.25° for GraphCast and Pangu). Blurry forecasts correspond to physically inconsistent atmospheric conditions and misrepresent extreme weather. Similar trends hold for other derived variables of meteorological interest (Supplementary Information section  H.2 ). Ensemble-mean predictions, from both NeuralGCM and ECMWF, are closer to ERA5 in an average sense, and thus are inherently smooth at long lead times. In contrast, as shown in Fig. 3 and in Supplementary Information section  H.3 , individual realizations from the ECMWF and NeuralGCM ensembles remain sharp, even at long lead times. Like ECMWF-ENS, NeuralGCM-ENS produces a statistically representative range of future weather scenarios for each weather phenomenon, despite its eight-times-coarser resolution.

All forecasts are initialized at 2020-08-22T12z, chosen to highlight Hurricane Laura, the most damaging Atlantic hurricane of 2020. a , Specific humidity at 700 hPa for 1-day, 5-day and 10-day forecasts over North America and the Northeast Pacific Ocean from ERA5 14 , ECMWF-HRES, NeuralGCM-0.7°, ECMWF-ENS (mean), NeuralGCM-ENS (mean), GraphCast 3 and Pangu 4 . b , Forecasts from individual ensemble members from ECMWF-ENS and NeuralGCM-ENS over regions of interest, including predicted tracks of Hurricane Laura from each of the 50 ensemble members (Supplementary Information section  I.2 ). The track from ERA5 is plotted in black.

We can quantify the blurriness of different forecast models via their power spectra. Supplementary Figs. 17 and 18 show that the power spectra of NeuralCGM-0.7° is consistently closer to ERA5 than the other machine-learning forecast methods, but is still blurrier than ECMWF’s physical forecasts. The spectra of NeuralGCM forecasts is also roughly constant over the forecast period, in stark contrast to GraphCast, which worsens with lead time. The spectrum of NeuralGCM becomes more accurate with increased resolution (Supplementary Fig. 22 ), which suggests the potential for further improvements of NeuralGCM models trained at higher resolutions.

Water budget

In NeuralGCM, advection is handled by the dynamical core, while the machine-learning parameterization models local processes within vertical columns of the atmosphere. Thus, unlike pure machine-learning methods, local sources and sinks can be isolated from tendencies owing to horizontal transport and other resolved dynamics (Supplementary Fig. 3 ). This makes our results more interpretable and facilitates the diagnosis of the water budget. Specifically, we diagnose precipitation minus evaporation (Supplementary Information section  H.5 ) rather than directly predicting these as in machine-learning-based approaches 3 . For short weather forecasts, the mean of precipitation minus evaporation has a realistic spatial distribution that is very close to ERA5 data (Extended Data Fig. 4c–e ). The precipitation-minus-evaporation rate distribution of NeuralGCM-0.7° closely matches the ERA5 distribution in the extratropics (Extended Data Fig. 4b ), although it underestimates extreme events in the tropics (Extended Data Fig. 4a ). It is noted that the current version of NeuralGCM directly predicts tendencies for an atmospheric column, and thus cannot distinguish between precipitation and evaporation.

Geostrophic wind balance

We examined the extent to which NeuralGCM, GraphCast and ECMWF-HRES capture the geostrophic wind balance, the near-equilibrium between the dominant forces that drive large-scale dynamics in the mid-latitudes 30 . A recent study 16 highlighted that Pangu misrepresents the vertical structure of the geostrophic and ageostrophic winds and noted a deterioration at longer lead times. Similarly, we observe that GraphCast shows an error that worsens with lead time. In contrast, NeuralGCM more accurately depicts the vertical structure of the geostrophic and ageostrophic winds, as well as their ratio, compared with GraphCast across various rollouts, when compared against ERA5 data (Extended Data Fig. 3 ). However, ECMWF-HRES still shows a slightly closer alignment to ERA5 data than NeuralGCM does. Within NeuralGCM, the representation of the geostrophic wind’s vertical structure only slightly degrades in the initial few days, showing no noticeable changes thereafter, particularly beyond day 5.

Generalizing to unseen data

Physically consistent weather models should still perform well for weather conditions for which they were not trained. We expect that NeuralGCM may generalize better than machine-learning-only atmospheric models, because NeuralGCM employs neural networks that act locally in space, on individual vertical columns of the atmosphere. To explore this hypothesis, we compare versions of NeuralCGM-0.7° and GraphCast trained to 2017 on 5 years of weather forecasts beyond the training period (2018–2022) in Supplementary Fig. 36 . Unlike GraphCast, NeuralGCM does not show a clear trend of increasing error when initialized further into the future from the training data. To extend this test beyond 5 years, we trained a NeuralGCM-2.8° model using only data before 2000, and tested its skill for over 21 unseen years (Supplementary Fig. 35 ).

Climate simulations

Although our deterministic NeuralGCM models are trained to predict weather up to 3 days ahead, they are generally capable of simulating the atmosphere far beyond medium-range weather timescales. For extended climate simulations, we prescribe historical sea surface temperature (SST) and sea-ice concentration. These simulations feature many emergent phenomena of the atmosphere on timescales from months to decades.

For climate simulations with NeuralGCM, we use 2.8° and 1.4° deterministic models, which are relatively inexpensive to train (Supplementary Information section  G.7 ) and allow us to explore a larger parameter space to find stable models. Previous studies found that running extended simulations with hybrid models is challenging due to numerical instabilities and climate drift 21 . To quantify stability in our selected models, we run multiple initial conditions and report how many of them finish without instability.

Seasonal cycle and emergent phenomena

To assess the capability of NeuralGCM to simulate various aspects of the seasonal cycle, we run 2-year simulations with NeuralGCM-1.4°. for 37 different initial conditions spaced every 10 days for the year 2019. Out of these 37 initial conditions, 35 successfully complete the full 2 years without instability; for case studies of instability, see Supplementary Information section  H.7 , and Supplementary Figs. 26 and 27 . We compare results from NeuralGCM-1.4° for 2020 with ERA5 data and with outputs from the X-SHiELD global cloud-resolving model, which is coupled to an ocean model nudged towards reanalysis 31 . This X-SHiELD run has been used as a target for training machine-learning climate models 24 . For comparison, we evaluate models after regridding predictions to 1.4° resolution. This comparison slightly favours NeuralGCM because NeuralGCM was tuned to match ERA5, but the discrepancy between ERA5 and the actual atmosphere is small relative to model error.

Figure 4a shows the temporal variation of the global mean temperature to 2020, as captured by 35 simulations from NeuralGCM, in comparison with the ERA5 reanalysis and standard climatology benchmarks. The seasonality and variability of the global mean temperature from NeuralGCM are quantitatively similar to those observed in ERA5. The ensemble-mean temperature RMSE for NeuralGCM stands at 0.16 K when benchmarked against ERA5, which is a significant improvement over the climatology’s RMSE of 0.45 K. We find that NeuralGCM accurately simulates the seasonal cycle, as evidenced by metrics such as the annual cycle of the global precipitable water (Supplementary Fig. 30a ) and global total kinetic energy (Supplementary Fig. 30b ). Furthermore, the model captures essential atmospheric dynamics, including the Hadley circulation and the zonal-mean zonal wind (Supplementary Fig. 28 ), as well as the spatial patterns of eddy kinetic energy in different seasons (Supplementary Fig. 31 ), and the distinctive seasonal behaviours of monsoon circulation (Supplementary Fig. 29 ; additional details are provided in Supplementary Information section  I.1 ).

a , Global mean temperature for ERA5 14 (orange), 1990–2019 climatology (black) and NeuralGCM-1.4° (blue) for 2020 using 35 simulations initialized every 10 days during 2019 (thick line, ensemble mean; thin lines, different initial conditions). b , Yearly global mean temperature for ERA5 (orange), mean over 22 CMIP6 AMIP experiments 34 (violet; model details are in Supplementary Information section  I.3 ) and NeuralGCM-2.8° for 22 AMIP-like simulations with prescribed SST initialized every 10 days during 1980 (thick line, ensemble mean; thin lines, different initial conditions). c , The RMSB of the 850-hPa temperature averaged between 1981 and 2014 for 22 NeuralGCM-2.8° AMIP runs (labelled NGCM), 22 CMIP6 AMIP experiments (labelled AMIP) and debiased 22 CMIP6 AMIP experiments (labelled AMIP*; bias was removed by removing the 850-hPa global temperature bias). In the box plots, the red line represents the median. The box delineates the first to third quartiles; the whiskers extend to 1.5 times the interquartile range (Q1 − 1.5IQR and Q3 + 1.5IQR), and outliers are shown as individual dots. d , Vertical profiles of tropical (20° S–20° N) temperature trends for 1981–2014. Orange, ERA5; black dots, Radiosonde Observation Correction using Reanalyses (RAOBCORE) 41 ; blue dots, mean trends for NeuralGCM; purple dots, mean trends from CMIP6 AMIP runs (grey and black whiskers, 25th and 75th percentiles for NeuralGCM and CMIP6 AMIP runs, respectively). e – g , Tropical cyclone tracks for ERA5 ( e ), NeuralGCM-1.4° ( f ) and X-SHiELD 31 ( g ). h – k , Mean precipitable water for ERA5 ( h ) and the precipitable water bias in NeuralGCM-1.4° ( i ), initialized 90 days before mid-January 2020 similarly to X-SHiELD, X-SHiELD ( j ) and climatology ( k ; averaged between 1990 and 2019). In d – i , quantities are calculated between mid-January 2020 and mid-January 2021 and all models were regridded to a 256 × 128 Gaussian grid before computation and tracking.

Next, we compare the annual biases of a single NeuralGCM realization with a single realization of X-SHiELD (the only one available), both initiated in mid-October 2019. We consider 19 January 2020 to 17 January 2021, the time frame for which X-SHiELD data are available. Global cloud-resolving models, such as X-SHiELD, are considered state of the art, especially for simulating the hydrological cycle, owing to their resolution being capable of resolving deep convection 32 . The annual bias in precipitable water for NeuralGCM (RMSE of 1.09 mm) is substantially smaller than the biases of both X-SHiELD (RMSE of 1.74 mm) and climatology (RMSE of 1.36 mm; Fig. 4i–k ). Moreover, NeuralGCM shows a lower temperature bias in the upper and lower troposphere than X-SHiELD (Extended Data Fig. 6 ). We also indirectly compare precipitation bias in X-SHiELD with precipitation-minus-evaporation bias in NeuralGCM-1.4°, which shows slightly larger bias and grid-scale artefacts for NeuralGCM (Extended Data Fig. 5 ).

Finally, to assess the capability of NeuralGCM to generate tropical cyclones in an annual model integration, we use the tropical cyclone tracker TempestExtremes 33 , as described in Supplementary Information section   I.2 , Supplementary Fig. 34 and Supplementary Table 6 . Figure 4e–g shows that NeuralGCM, even at a coarse resolution of 1.4°, produces realistic trajectories and counts of tropical cyclone (83 versus 86 in ERA5 for the corresponding period), whereas X-SHiELD, when regridded to 1.4° resolution, substantially underestimates the tropical cyclone count (40). Additional statistical analyses of tropical cyclones can be found in Extended Data Figs. 7 and 8 .

Decadal simulations

To assess the capability of NeuralGCM to simulate historical temperature trends, we conduct AMIP-like simulations over a duration of 40 years with NeuralGCM-2.8°. Out of 37 different runs with initial conditions spaced every 10 days during the year 1980, 22 simulations were stable for the entire 40-year period, and our analysis focuses on these results. We compare with 22 simulations run with prescribed SST from the Coupled Model Intercomparison Project Phase 6 (CMIP6) 34 , listed in Supplementary Information section  I.3 .

We find that all 40-year simulations of NeuralGCM, as well as the mean of the 22 AMIP runs, accurately capture the global warming trends observed in ERA5 data (Fig. 4b ). There is a strong correlation in the year-to-year temperature trends with ERA5 data, suggesting that NeuralGCM effectively captures the impact of SST forcing on climate. When comparing spatial biases averaged over 1981–2014, we find that all 22 NeuralGCM-2.8° runs have smaller bias than the CMIP6 AMIP runs, and this result remains even when removing the global temperature bias in CMIP6 AMIP runs (Fig. 4c and Supplementary Figs. 32 and 33 ).

Next, we investigated the vertical structure of tropical warming trends, which climate models tend to overestimate in the upper troposphere 35 . As shown in Fig. 4d , the trends, calculated by linear regression, of NeuralGCM are closer to ERA5 than those of AMIP runs. In particular, the bias in the upper troposphere is reduced. However, NeuralGCM does show a wider spread in its predictions than the AMIP runs, even at levels near the surface where temperatures are typically more constrained by prescribed SST.

Lastly, we evaluated NeuralGCM’s capability to generalize to unseen warmer climates by conducting AMIP simulations with increased SST (Supplementary Information section  I.4.2 ). We find that NeuralGCM shows some of the robust features of climate warming response to modest SST increases (+1 K and +2 K); however, for more substantial SST increases (+4 K), NeuralGCM’s response diverges from expectations (Supplementary Fig. 37 ). In addition, AMIP simulations with increased SST show climate drift, underscoring NeuralGCM’s limitations in this context (Supplementary Fig. 38 ).

NeuralGCM is a differentiable hybrid atmospheric model that combines the strengths of traditional GCMs with machine learning for weather forecasting and climate simulation. To our knowledge, NeuralGCM is the first machine-learning-based model to make accurate ensemble weather forecasts, with better CRPS than state-of-the-art physics-based models. It is also, to our knowledge, the first hybrid model that achieves comparable spatial bias to global cloud-resolving models, can simulate realistic tropical cyclone tracks and can run AMIP-like simulations with realistic historical temperature trends. Overall, NeuralGCM demonstrates that incorporating machine learning is a viable alternative to building increasingly detailed physical models 32 for improving GCMs.

Compared with traditional GCMs with similar skill, NeuralGCM is computationally efficient and low complexity. NeuralGCM runs at 8- to 40-times-coarser horizontal resolution than ECMWF’s Integrated Forecasting System and global cloud-resolving models, which enables 3 to 5 orders of magnitude savings in computational resources. For example, NeuralGCM-1.4° simulates 70,000 simulation days in 24 hours using a single tensor-processing-unit versus 19 simulated days on 13,824 central-processing-unit cores with X-SHiELD (Extended Data Table 1 ). This can be leveraged for previously impractical tasks such as large ensemble forecasting. NeuralGCM’s dynamical core uses global spectral methods 36 , and learned physics is parameterized with fully connected neural networks acting on single vertical columns. Substantial headroom exists to pursue higher accuracy using advanced numerical methods and machine-learning architectures.

Our results provide strong evidence for the disputed hypothesis 37 , 38 , 39 that learning to predict short-term weather is an effective way to tune parameterizations for climate. NeuralGCM models trained on 72-hour forecasts are capable of realistic multi-year simulation. When provided with historical SSTs, they capture essential atmospheric dynamics such as seasonal circulation, monsoons and tropical cyclones. However, we will probably need alternative training strategies 38 , 39 to learn important processes for climate with subtle impacts on weather timescales, such as a cloud feedback.

The NeuralGCM approach is compatible with incorporating either more physics or more machine learning, as required for operational weather forecasts and climate simulations. For weather forecasting, we expect that end-to-end learning 40 with observational data will allow for better and more relevant predictions, including key variables such as precipitation. Such models could include neural networks acting as corrections to traditional data assimilation and model diagnostics. For climate projection, NeuralGCM will need to be reformulated to enable coupling with other Earth-system components (for example, ocean and land), and integrating data on the atmospheric chemical composition (for example, greenhouse gases and aerosols). There are also research challenges common to current machine-learning-based climate models 19 , including the capability to simulate unprecedented climates (that is, generalization), adhering to physical constraints, and resolving numerical instabilities and climate drift. NeuralGCM’s flexibility to incorporate physics-based models (for example, radiation) offers a promising avenue to address these challenges.

Models based on physical laws and empirical relationships are ubiquitous in science. We believe the differentiable hybrid modelling approach of NeuralGCM has the potential to transform simulation for a wide range of applications, such as materials discovery, protein folding and multiphysics engineering design.

Differentiable atmospheric model

NeuralGCM combines components of the numerical solver and flexible neural network parameterizations. Simulation in time is carried out in a coordinate system suitable for solving the dynamical equations of the atmosphere, describing large-scale fluid motion and thermodynamics under the influence of gravity and the Coriolis force.

Our differentiable dynamical core is implemented in JAX, a library for high-performance code in Python that supports automatic differentiation 42 . The dynamical core solves the hydrostatic primitive equations with moisture, using a horizontal pseudo-spectral discretization and vertical sigma coordinates 36 , 43 . We evolve seven prognostic variables: vorticity and divergence of horizontal wind, temperature, surface pressure, and three water species (specific humidity, and specific ice and liquid cloud water content).

Our learned physics module uses the single-column approach of GCMs 2 , whereby information from only a single atmospheric column is used to predict the impact of unresolved processes occurring within that column. These effects are predicted using a fully connected neural network with residual connections, with weights shared across all atmospheric columns (Supplementary Information section  C.4 ).

The inputs to the neural network include the prognostic variables in the atmospheric column, total incident solar radiation, sea-ice concentration and SST (Supplementary Information section  C.1 ). We also provide horizontal gradients of the prognostic variables, which we found improves performance 44 . All inputs are standardized to have zero mean and unit variance using statistics precomputed during model initialization. The outputs are the prognostic variable tendencies scaled by the fixed unconditional standard deviation of the target field (Supplementary Information section  C.5 ).

To interface between ERA5 14 data stored in pressure coordinates and the sigma coordinate system of our dynamical core, we introduce encoder and decoder components (Supplementary Information section  D ). These components perform linear interpolation between pressure levels and sigma coordinate levels. We additionally introduce learned corrections to both encoder and decoder steps (Supplementary Figs. 4–6 ), using the same column-based neural network architecture as the learned physics module. Importantly, the encoder enables us to eliminate the gravity waves from initialization shock 45 , which otherwise contaminate forecasts.

Figure 1a shows the sequence of steps that NeuralGCM takes to make a forecast. First, it encodes ERA5 data at t  =  t 0 on pressure levels to initial conditions on sigma coordinates. To perform a time step, the dynamical core and learned physics (Fig. 1b ) then compute tendencies, which are integrated in time using an implicit–explicit ordinary differential equation solver 46 (Supplementary Information section  E and Supplementary Table 2 ). This is repeated to advance the model from t  =  t 0 to t  =  t final . Finally, the decoder converts predictions back to pressure levels.

The time-step size of the ODE solver (Supplementary Table 3 ) is limited by the Courant–Friedrichs–Lewy condition on dynamics, and can be small relative to the timescale of atmospheric change. Evaluating learned physics is approximately 1.5 times as expensive as a time step of the dynamical core. Accordingly, following the typical practice for GCMs, we hold learned physics tendencies constant for multiple ODE time steps to reduce computational expense, typically corresponding to 30 minutes of simulation time.

Deterministic and stochastic models

We train deterministic NeuralGCM models using a combination of three loss functions (Supplementary Information section  G.4 ) to encourage accuracy and sharpness while penalizing bias. During the main training phase, all losses are defined in a spherical harmonics basis. We use a standard mean squared error loss for prompting accuracy, modified to progressively filter out contributions from higher total wavenumbers at longer lead times (Supplementary Fig. 8 ). This filtering approach tackles the ‘double penalty problem’ 47 as it prevents the model from being penalized for predicting high-wavenumber features in incorrect locations at later times, especially beyond the predictability horizon. A second loss term encourages the spectrum to match the training data using squared loss on the total wavenumber spectrum of prognostic variables. These first two losses are evaluated on both sigma and pressure levels. Finally, a third loss term discourages bias by adding mean squared error on the batch-averaged mean amplitude of each spherical harmonic coefficient. For analysis of the impact that various loss functions have, refer to Supplementary Information section  H.6.1 , and Supplementary Figs. 23 and 24 . The combined action of the three training losses allow the resulting models trained on 3-day rollouts to remain stable during years-to-decades-long climate simulations. Before final evaluations, we perform additional fine-tuning of just the decoder component on short rollouts of 24 hours (Supplementary Information section  G.5 ).

Stochastic NeuralGCM models incorporate inherent randomness in the form of additional random fields passed as inputs to neural network components. Our stochastic loss is based on the CRPS 28 , 48 , 49 . CRPS consists of mean absolute error that encourages accuracy, balanced by a similar term that encourages ensemble spread. For each variable we use a sum of CRPS in grid space and CRPS in the spherical harmonic basis below a maximum cut-off wavenumber (Supplementary Information section  G.6 ). We compute CRPS on rollout lengths from 6 hours to 5 days. As illustrated in Fig. 1 , we inject noise to the learned encoder and the learned physics module by sampling from Gaussian random fields with learned spatial and temporal correlation (Supplementary Information section  C.2 and Supplementary Fig. 2 ). For training, we generate two ensemble members per forecast, which suffices for an unbiased estimate of CRPS.

Data availability

For training and evaluating the NeuralGCM models, we used the publicly available ERA5 dataset 14 , originally downloaded from https://cds.climate.copernicus.eu/ and available via Google Cloud Storage in Zarr format at gs://gcp-public-data-arco-era5/ar/full_37-1h-0p25deg-chunk-1.zarr-v3. To compare NeuralGCM with operational and data-driven weather models, we used forecast datasets distributed as part of WeatherBench2 12 at https://weatherbench2.readthedocs.io/en/latest/data-guide.html , to which we have added NeuralGCM forecasts for 2020. To compare NeuralGCM with atmospheric models in climate settings, we used CMIP6 data available at https://catalog.pangeo.io/browse/master/climate/ , as well as X-SHiELD 24 outputs available on Google Cloud storage in a ‘requester pays’ bucket at gs://ai2cm-public-requester-pays/C3072-to-C384-res-diagnostics. The Radiosonde Observation Correction using Reanalyses (RAOBCORE) V1.9 that was used as reference tropical temperature trends was downloaded from https://webdata.wolke.img.univie.ac.at/haimberger/v1.9/ . Base maps use freely available data from https://www.naturalearthdata.com/downloads/ .

Code availability

The NeuralGCM code base is separated into two open source projects: Dinosaur and NeuralGCM, both publicly available on GitHub at https://github.com/google-research/dinosaur (ref. 50 ) and https://github.com/google-research/neuralgcm (ref. 51 ). The Dinosaur package implements a differentiable dynamical core used by NeuralGCM, whereas the NeuralGCM package provides machine-learning models and checkpoints of trained models. Evaluation code for NeuralGCM weather forecasts is included in WeatherBench2 12 , available at https://github.com/google-research/weatherbench2 (ref. 52 ).

Bauer, P., Thorpe, A. & Brunet, G. The quiet revolution of numerical weather prediction. Nature 525 , 47–55 (2015).

Article   ADS   CAS   PubMed   Google Scholar  

Balaji, V. et al. Are general circulation models obsolete? Proc. Natl Acad. Sci. USA 119 , e2202075119 (2022).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Lam, R. et al. Learning skillful medium-range global weather forecasting. Science 382 , 1416–1421 (2023).

Article   ADS   MathSciNet   CAS   PubMed   Google Scholar  

Bi, K. et al. Accurate medium-range global weather forecasting with 3D neural networks. Nature 619 , 533–538 (2023).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Hourdin, F. et al. The art and science of climate model tuning. Bull. Am. Meteorol. Soc. 98 , 589–602 (2017).

Article   ADS   Google Scholar  

Bony, S. & Dufresne, J.-L. Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett. 32 , L20806 (2005).

Webb, M. J., Lambert, F. H. & Gregory, J. M. Origins of differences in climate sensitivity, forcing and feedback in climate models. Clim. Dyn. 40 , 677–707 (2013).

Article   Google Scholar  

Sherwood, S. C., Bony, S. & Dufresne, J.-L. Spread in model climate sensitivity traced to atmospheric convective mixing. Nature 505 , 37–42 (2014).

Article   ADS   PubMed   Google Scholar  

Palmer, T. & Stevens, B. The scientific challenge of understanding and estimating climate change. Proc. Natl Acad. Sci. USA 116 , 24390–24395 (2019).

Fischer, E. M., Beyerle, U. & Knutti, R. Robust spatially aggregated projections of climate extremes. Nat. Clim. Change 3 , 1033–1038 (2013).

Field, C. B. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation: Special Report of the Intergovernmental Panel on Climate Change (Cambridge Univ. Press, 2012).

Rasp, S. et al. WeatherBench 2: A benchmark for the next generation of data-driven global weather models. J. Adv. Model. Earth Syst. 16 , e2023MS004019 (2024).

Keisler, R. Forecasting global weather with graph neural networks. Preprint at https://arxiv.org/abs/2202.07575 (2022).

Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 146 , 1999–2049 (2020).

Zhou, L. et al. Toward convective-scale prediction within the next generation global prediction system. Bull. Am. Meteorol. Soc. 100 , 1225–1243 (2019).

Bonavita, M. On some limitations of current machine learning weather prediction models. Geophys. Res. Lett. 51 , e2023GL107377 (2024).

Weyn, J. A., Durran, D. R. & Caruana, R. Improving data-driven global weather prediction using deep convolutional neural networks on a cubed sphere. J. Adv. Model. Earth Syst. 12 , e2020MS002109 (2020).

Watt-Meyer, O. et al. ACE: a fast, skillful learned global atmospheric model for climate prediction. Preprint at https://arxiv.org/abs/2310.02074 (2023).

Bretherton, C. S. Old dog, new trick: reservoir computing advances machine learning for climate modeling. Geophys. Res. Lett. 50 , e2023GL104174 (2023).

Reichstein, M. et al. Deep learning and process understanding for data-driven Earth system science. Nature 566 , 195–204 (2019).

Brenowitz, N. D. & Bretherton, C. S. Spatially extended tests of a neural network parametrization trained by coarse-graining. J. Adv. Model. Earth Syst. 11 , 2728–2744 (2019).

Rasp, S., Pritchard, M. S. & Gentine, P. Deep learning to represent subgrid processes in climate models. Proc. Natl Acad. Sci. USA 115 , 9684–9689 (2018).

Yuval, J. & O’Gorman, P. A. Stable machine-learning parameterization of subgrid processes for climate modeling at a range of resolutions. Nat. Commun. 11 , 3295 (2020).

Kwa, A. et al. Machine-learned climate model corrections from a global storm-resolving model: performance across the annual cycle. J. Adv. Model. Earth Syst. 15 , e2022MS003400 (2023).

Arcomano, T., Szunyogh, I., Wikner, A., Hunt, B. R. & Ott, E. A hybrid atmospheric model incorporating machine learning can capture dynamical processes not captured by its physics-based component. Geophys. Res. Lett. 50 , e2022GL102649 (2023).

Han, Y., Zhang, G. J. & Wang, Y. An ensemble of neural networks for moist physics processes, its generalizability and stable integration. J. Adv. Model. Earth Syst. 15 , e2022MS003508 (2023).

Gelbrecht, M., White, A., Bathiany, S. & Boers, N. Differentiable programming for Earth system modeling. Geosci. Model Dev. 16 , 3123–3135 (2023).

Gneiting, T. & Raftery, A. E. Strictly proper scoring rules, prediction, and estimation. J. Am. Stat. Assoc. 102 , 359–378 (2007).

Article   MathSciNet   CAS   Google Scholar  

Fortin, V., Abaza, M., Anctil, F. & Turcotte, R. Why should ensemble spread match the RMSE of the ensemble mean? J. Hydrometeorol. 15 , 1708–1713 (2014).

Holton, J. R. An introduction to Dynamic Meteorology 5th edn (Elsevier, 2004).

Cheng, K.-Y. et al. Impact of warmer sea surface temperature on the global pattern of intense convection: insights from a global storm resolving model. Geophys. Res. Lett. 49 , e2022GL099796 (2022).

Stevens, B. et al. DYAMOND: the dynamics of the atmospheric general circulation modeled on non-hydrostatic domains. Prog. Earth Planet. Sci. 6 , 61 (2019).

Ullrich, P. A. et al. TempestExtremes v2.1: a community framework for feature detection, tracking, and analysis in large datasets. Geosc. Model Dev. 14 , 5023–5048 (2021).

Eyring, V. et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev. 9 , 1937–1958 (2016).

Mitchell, D. M., Lo, Y. E., Seviour, W. J., Haimberger, L. & Polvani, L. M. The vertical profile of recent tropical temperature trends: persistent model biases in the context of internal variability. Environ. Res. Lett. 15 , 1040b4 (2020).

Bourke, W. A multi-level spectral model. I. Formulation and hemispheric integrations. Mon. Weather Rev. 102 , 687–701 (1974).

Ruiz, J. J., Pulido, M. & Miyoshi, T. Estimating model parameters with ensemble-based data assimilation: a review. J. Meteorol. Soc. Jpn Ser. II 91 , 79–99 (2013).

Schneider, T., Lan, S., Stuart, A. & Teixeira, J. Earth system modeling 2.0: a blueprint for models that learn from observations and targeted high-resolution simulations. Geophys. Res. Lett. 44 , 12–396 (2017).

Schneider, T., Leung, L. R. & Wills, R. C. J. Opinion: Optimizing climate models with process knowledge, resolution, and artificial intelligence. Atmos. Chem. Phys. 24 , 7041–7062 (2024).

Sutskever, I., Vinyals, O. & Le, Q. V. Sequence to sequence learning with neural networks. Adv. Neural Inf. Process. Syst. 27 , 3104–3112 (2014).

Haimberger, L., Tavolato, C. & Sperka, S. Toward elimination of the warm bias in historic radiosonde temperature records—some new results from a comprehensive intercomparison of upper-air data. J. Clim. 21 , 4587–4606 (2008).

Bradbury, J. et al. JAX: composable transformations of Python+NumPy programs. GitHub http://github.com/google/jax (2018).

Durran, D. R. Numerical Methods for Fluid Dynamics: With Applications to Geophysics Vol. 32, 2nd edn (Springer, 2010).

Wang, P., Yuval, J. & O’Gorman, P. A. Non-local parameterization of atmospheric subgrid processes with neural networks. J. Adv. Model. Earth Syst. 14 , e2022MS002984 (2022).

Daley, R. Normal mode initialization. Rev. Geophys. 19 , 450–468 (1981).

Whitaker, J. S. & Kar, S. K. Implicit–explicit Runge–Kutta methods for fast–slow wave problems. Mon. Weather Rev. 141 , 3426–3434 (2013).

Gilleland, E., Ahijevych, D., Brown, B. G., Casati, B. & Ebert, E. E. Intercomparison of spatial forecast verification methods. Weather Forecast. 24 , 1416–1430 (2009).

Rasp, S. & Lerch, S. Neural networks for postprocessing ensemble weather forecasts. Month. Weather Rev. 146 , 3885–3900 (2018).

Pacchiardi, L., Adewoyin, R., Dueben, P. & Dutta, R. Probabilistic forecasting with generative networks via scoring rule minimization. J. Mach. Learn. Res. 25 , 1–64 (2024).

Smith, J. A., Kochkov, D., Norgaard, P., Yuval, J. & Hoyer, S. google-research/dinosaur: 1.0.0. Zenodo https://doi.org/10.5281/zenodo.11376145 (2024).

Kochkov, D. et al. google-research/neuralgcm: 1.0.0. Zenodo https://doi.org/10.5281/zenodo.11376143 (2024).

Rasp, S. et al. google-research/weatherbench2: v0.2.0. Zenodo https://doi.org/10.5281/zenodo.11376271 (2023).

Download references

Acknowledgements

We thank A. Kwa, A. Merose and K. Shah for assistance with data acquisition and handling; L. Zepeda-Núñez for feedback on the paper; and J. Anderson, C. Van Arsdale, R. Chemke, G. Dresdner, J. Gilmer, J. Hickey, N. Lutsko, G. Nearing, A. Paszke, J. Platt, S. Ponda, M. Pritchard, D. Rothenberg, F. Sha, T. Schneider and O. Voicu for discussions.

Author information

These authors contributed equally: Dmitrii Kochkov, Janni Yuval, Ian Langmore, Peter Norgaard, Jamie Smith, Stephan Hoyer

Authors and Affiliations

Google Research, Mountain View, CA, USA

Dmitrii Kochkov, Janni Yuval, Ian Langmore, Peter Norgaard, Jamie Smith, Griffin Mooers, James Lottes, Stephan Rasp, Michael P. Brenner & Stephan Hoyer

Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USA

Milan Klöwer

European Centre for Medium-Range Weather Forecasts, Reading, UK

Peter Düben & Sam Hatfield

Google DeepMind, London, UK

Peter Battaglia, Alvaro Sanchez-Gonzalez & Matthew Willson

School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA

Michael P. Brenner

You can also search for this author in PubMed   Google Scholar

Contributions

D.K., J.Y., I.L., P.N., J.S. and S. Hoyer contributed equally to this work. D.K., J.Y., I.L., P.N., J.S., G.M., J.L. and S. Hoyer wrote the code. D.K., J.Y., I.L., P.N., G.M. and S. Hoyer trained models and analysed the data. M.P.B. and S. Hoyer managed and oversaw the research project. M.K., S.R., P.D., S. Hatfield, P.B. and M.P.B. contributed technical advice and ideas. M.W. ran experiments with GraphCast for comparison with NeuralGCM. A.S.-G. assisted with data preparation. D.K., J.Y., I.L., P.N. and S. Hoyer wrote the paper. All authors gave feedback and contributed to editing the paper.

Corresponding authors

Correspondence to Dmitrii Kochkov , Janni Yuval or Stephan Hoyer .

Ethics declarations

Competing interests.

D.K., J.Y., I.L., P.N., J.S., J.L., S.R., P.B., A.S.-G., M.W., M.P.B. and S. Hoyer are employees of Google. S. Hoyer, D.K., I.L., J.Y., G.M., P.N., J.S. and M.B. have filed international patent application PCT/US2023/035420 in the name of Google LLC, currently pending, relating to neural general circulation models.

Peer review

Peer review information.

Nature thanks Karthik Kashinath and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended data fig. 1 maps of bias for neuralgcm-ens and ecmwf-ens forecasts..

Bias is averaged over all forecasts initialized in 2020.

Extended Data Fig. 2 Maps of spread-skill ratio for NeuralGCM-ENS and ECMWF-ENS forecasts.

Spread-skill ratio is averaged over all forecasts initialized in 2020.

Extended Data Fig. 3 Geostrophic balance in NeuralGCM, GraphCast 3 and ECMWF-HRES.

Vertical profiles of the extratropical intensity (averaged between latitude 30°–70° in both hemispheres) and over all forecasts initialized in 2020 of (a,d,g) geostrophic wind, (b,e,h) ageostrophic wind and (c,f,i) the ratio of the intensity of ageostrophic wind over geostrophic wind for ERA5 (black continuous line in all panels), (a,b,c) NeuralGCM-0.7°, (d,e,f) GraphCast and (g,h,i) ECMWF-HRES at lead times of 1 day, 5 days and 10 days.

Extended Data Fig. 4 Precipitation minus evaporation calculated from the third day of weather forecasts.

(a) Tropical (latitudes −20° to 20°) precipitation minus evaporation (P minus E) rate distribution, (b) Extratropical (latitudes 30° to 70° in both hemispheres) P minus E, (c) mean P minus E for 2020 ERA5 14 and (d) NeuralGCM-0.7° (calculated from the third day of forecasts and averaged over all forecasts initialized in 2020), (e) the bias between NeuralGCM-0.7° and ERA5, (f-g) Snapshot of daily precipitation minus evaporation for 2020-01-04 for (f) NeuralGCM-0.7° (forecast initialized on 2020-01-02) and (g) ERA5.

Extended Data Fig. 5 Indirect comparison between precipitation bias in X-SHiELD and precipitation minus evaporation bias in NeuralGCM-1.4°.

Mean precipitation calculated between 2020-01-19 and 2021-01-17 for (a) ERA5 14 (c) X-SHiELD 31 and the biases in (e) X-SHiELD and (g) climatology (ERA5 data averaged over 1990-2019). Mean precipitation minus evaporation calculated between 2020-01-19 and 2021-01-17 for (b) ERA5 (d) NeuralGCM-1.4° (initialized in October 18th 2019) and the biases in (f) NeuralGCM-1.4° and (h) climatology (data averaged over 1990–2019).

Extended Data Fig. 6 Yearly temperature bias for NeuralGCM and X-SHiELD 31 .

Mean temperature between 2020-01-19 to 2020-01-17 for (a) ERA5 at 200hPa and (b) 850hPa. (c,d) the bias in the temperature for NeuralGCM-1.4°, (e,f) the bias in X-SHiELD and (g,h) the bias in climatology (calculated from 1990–2019). NeuralGCM-1.4° was initialized in 18th of October (similar to X-SHiELD).

Extended Data Fig. 7 Tropical Cyclone densities and annual regional counts.

(a) Tropical Cyclone (TC) density from ERA5 14 data spanning 1987–2020. (b) TC density from NeuralGCM-1.4° for 2020, generated using 34 different initial conditions all initialized in 2019. (c) Box plot depicting the annual number of TCs across different regions, based on ERA5 data (1987–2020), NeuralGCM-1.4° for 2020 (34 initial conditions), and orange markers show ERA5 for 2020. In the box plots, the red line represents the median; the box delineates the first to third quartiles; the whiskers extend to 1.5 times the interquartile range (Q1 − 1.5IQR and Q3 + 1.5IQR), and outliers are shown as individual dots. Each year is defined from January 19th to January 17th of the following year, aligning with data availability from X-SHiELD. For NeuralGCM simulations, the 3 initial conditions starting in January 2019 exclude data for January 17th, 2021, as these runs spanned only two years.

Extended Data Fig. 8 Tropical Cyclone maximum wind distribution in NeuralGCM vs. ERA5 14 .

Number of Tropical Cyclones (TCs) as a function of maximum wind speed at 850hPa across different regions, based on ERA5 data (1987–2020; in orange), and NeuralGCM-1.4° for 2020 (34 initial conditions; in blue). Each year is defined from January 19th to January 17th of the following year, aligning with data availability from X-SHiELD. For NeuralGCM simulations, the 3 initial conditions starting in January 2019 exclude data for January 17th, 2021, as these runs spanned only two years.

Supplementary information

Supplementary information.

Supplementary Information (38 figures, 6 tables): (A) Lines of code in atmospheric models; (B) Dynamical core of NeuralGCM; (C) Learned physics of NeuralGCM; (D) Encoder and decoder of NeuralGCM; (E) Time integration; (F) Evaluation metrics; (G) Training; (H) Additional weather evaluations; (I) Additional climate evaluations.

Peer Review File

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Kochkov, D., Yuval, J., Langmore, I. et al. Neural general circulation models for weather and climate. Nature 632 , 1060–1066 (2024). https://doi.org/10.1038/s41586-024-07744-y

Download citation

Received : 13 November 2023

Accepted : 15 June 2024

Published : 22 July 2024

Issue Date : 29 August 2024

DOI : https://doi.org/10.1038/s41586-024-07744-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Google ai predicts long-term climate trends and weather — in minutes.

• Helena Kudiabor

Nature (2024)

Weather and climate predicted accurately — without using a supercomputer

• Oliver Watt-Meyer

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

Fact Check: Global temperature has risen, not fallen, since medieval period 

• Medium Text

TWICE THE RISE SINCE 1850

Historic warming.

Our Standards: The Thomson Reuters Trust Principles. , opens new tab

Fact Check: Fake ‘Kamal Aroush’ narrative used to discredit Harris’ eligibility for office

U.S. Vice President Kamala Harris’ presidential bid has revived false allegations online about her eligibility to run for office, including a years-old disinformation narrative that she is a man named “Kamal Aroush.”

Your complimentary articles

You’ve read one of your four complimentary articles for this month.

You can read four articles free per month. To have complete access to the thousands of philosophy articles on this site, please

Climate Science & Falsifiability

Richard lawson shows how karl popper can help settle the climate debate..

Policymakers worldwide face a major headache relating to energy strategy. On the one hand, most climate scientists are warning that we must make a radical change away from reliance on carbon-based fossil fuels in order to avoid a catastrophic long-term change in global climate. On the other hand, the politicians are intensively lobbied by, and sometimes financed by, immensely wealthy and powerful fossil fuel corporations. Moreover, many popular newspapers and influential commentators are sceptical of prevailing climate science, and there is an active and noisy campaign against climate theory by contrarians in the social media. The electorate is of two minds; they do not like the apparently abnormal kinds of weather they are experiencing, but neither do they like the idea of the higher fuel bills or taxes that may be forced by decarbonisation. Small wonder that politicians have gone quiet on climate change.

Philosophers may not find this a particularly attractive arena to step into, but we have a moral duty to help unlock the truth about climate change if we can. And we do possess a key, in the form of the principle of falsifiability set out by Karl Popper in his book The Logic of Scientific Discovery (1934).

Earlier, logical positivists such as A.J. Ayer had argued that for a proposition to be meaningful, it must in principle be capable of proof (‘verification’). But Popper argued that the hallmark of a genuinely scientific proposition is not that it can be verified (because no number of observations can conclusively prove a hypothesis), but that the proposition can in principle be disproved (‘falsified’). For example, the proposition ‘All swans are white’ cannot be proved no matter how many swans you see; but it can be disproved by seeing just one black swan.

Popper & Kuhn

Is Popper still popular? W.V.O. Quine was happy to endorse falsification, although he modified the principle to make allowances for the quality of data purporting to overthrow a hypothesis, and to allow that new findings might simply force a modification of a hypothesis rather than its absolute refutation – with which idea Imre Lakatos agreed. But Thomas Kuhn in The Structure of Scientific Revolutions (1962) instead described the historic processes science follows: a consensus position or paradigm prevails in some area of science, but will eventually be overthrown by some radically new paradigm.

Both sides of the climate debate have availed themselves of Kuhn’s ideas. Climate contrarians have tried to present themselves as overthrowers of the prevailing climate science paradigm. However, there is more to being a Galileo than merely objecting to the prevailing consensus. It is necessary to possess a convincing and powerful alternative theory, and with the contrarians this is not the case, as we will see.

Indeed, the prevailing climate science might itself be characterised as having overthrown an old way of thinking. Soon after Svante Arrhenius first raised the possibility of an anthropogenically-enhanced greenhouse effect in 1896, Anders Angstrom argued on the basis of a simple laboratory experiment that the effect of carbon dioxide in absorbing infra red energy was very limited. Angstrom’s view prevailed until the 1960s, when it became understood that convection processes within the atmosphere means that his argument does not hold up. So it can be said that the contrarians represent the old paradigm; and indeed, Angstrom’s argument is still a core talking point in the contrarian community.

Not that this has much to do with the validity or otherwise of either side’s argument. In essence, Kuhn’s theory has more to do with the sociology of science than with its content. However, Kuhn did emphasise criteria for choosing one scientific theory over another: accuracy, consistency, broad scope, simplicity, and fruitfulness. Popperian falsifiability is implied by Kuhn’s first criterion – accuracy.

It is of course only too true that social and psychological factors influence what scientists accept and believe, but this is peripheral to the core of science. It’s what scientists might call ‘noise’. The central matter, the ‘signal’, is our changing understanding of objective reality, which scientists encounter as data, and data is still what they have to deal with. We can conclude that the refutation of theories through contrary data remains at the heart of the scientific method.

The Paradox of Science

Popper’s falsifiability principle implies that, contrary to popular misunderstanding, there is no such thing as scientific ‘proof’. The best status that even the best scientific theory can attain is ‘not-yet-disproven’. Even the most durable and revered laws, such as Newton’s laws of motion, may find extreme conditions where they no longer apply. (This does not necessarily mean that the old law is completely overthrown, but rather that its area of application becomes circumscribed.)

This absence of final, definitive proof creates a paradox: science, which we rightly regard as the most certain form of knowledge of the world, exists in a continuous state of uncertainty. In their daily lives scientists are perfectly happy with this uncertainty, not least because each new research paper can truthfully be concluded “More research is needed” – hopefully ensuring a continued supply of funding, and certainly ensuring a continued arena in which they can exercise their curiosity.

The paradox of science can be exploited in the media by opponents of any scientific case, who are able to challenge unwelcome scientific knowledge by saying, “ Prove it to me! There you are, you see, you cannot!” The scientist can put forward his evidence; but unfortunately most scientists are trained to be meticulous, and meticulous exposition does not sit easily with the standard two minute popular media discussion.

This then is the predicament in which climate scientists find themselves. They can put forward the evidence, but they cannot force their audience to agree with them. They can point to the fact that carbon dioxide is a greenhouse gas, that its levels in the atmosphere have risen by 40% since the Industrial Revolution, and that we can only account for the recent rise in global temperatures by including the enhanced greenhouse effect alongside known natural factors such as solar variability and ocean currents. They can point to the observed patterns of warming as consistent with warming due to greenhouse gases in contrast to other possible causes of warming. But in the end, the reasoning is inductive, not deductive. It is not proof . To be persuaded, the listener has to recognise a pattern that satisfies a number of questions and agrees with a large number of different lines of evidence. But in the end, no person can be compelled to make an inductive judgement that he or she does not wish to make. A creationist cannot be compelled to believe the evidence of evolution, for example.

This leaves climatologists with a problem. They can make a coherent and reasonable case, presenting the facts to decision-makers. But the decision-makers can also imagine the angry cries of the climate contrarians, and the sound of strife frightens them. They ask the scientists for proof, and all the scientists can say is that further research will narrow down the uncertainties. The only way for the politician to stay in his comfort zone is by deferring the decision to act against carbon-based energy – thus allowing the situation to get worse, forcing the next set of decision-makers to face even more difficult decisions. So the controversy rolls on, with the machinery of climate contrarianism every week putting forward another question to create doubt in the minds of the public, journalists and politicians. A few climatologists and activists devote their free time to answering them, but it is like fighting a Hydra: answer one question, and two more spring up to take its place.

Disproving the Doubters

Is this how it has to be? Can a handful of dissidents, using the megaphone of mass media, sustain inaction on a process of global warming that might well end in disaster for the human species?

Philosophy has revealed the means to resolve this problem. Science may not do proof, but it certainly does do disproof. So although it may not be possible for climatologists to prove their case conclusively, it is possible to look at the contrary hypothesis and refute it. And the contrarians do have a hypothesis: it is that man-made carbon dioxide will not have a severe effect on global climate. This angle transforms the debate into a question about the degree to which the global climate will change given the known increase in greenhouse gases.

There is no reasonable doubt that, ignoring feedback mechanisms, a doubling of carbon dioxide will raise the planet’s surface temperature by about 1.2°C, because this fact is derived from calculations based on universally accepted textbook physics, and is accepted by climatologists and reasonable contrarians alike. The real debate is about climate sensitivity – or what will result from this 1.2°C rise. The Earth’s climate is a complex system of interrelated energy flows, and any warming will result in an array of changes in the system. Most of these changes provide positive feedbacks – that is, they will further increase the initial warming. A number of different lines of evidence drawing from known or deduced changes in global temperature, recent and palaeological, all converge on an eventual temperature rise of between 1.5-4.5°C, with the most likely value being 3°C. Against this, classical climate contrarians put forward a value of 0.5-1°C as their figure for the final temperature increase resulting from a doubling of atmospheric carbon dioxide concentrations. This is their hypothesis; and it is refutable through measuring and calculating the known positive feedbacks – increase in atmospheric water vapour, changes in ice and snow albedo (reflectivity), changes in vegetation, and from secondary releases of carbon dioxide and methane from soil and ocean. The main negative feedbacks (temperature reducers) are a change in heat distribution in the atmosphere, which can be calculated as slightly reducing the positive water vapour feedback, and an increase in total energy radiation from the warming Earth (a feedback which probably sets a limit to extreme planetary overheating). Several attempts have been made by sceptical climate scientists to substantiate their 0.5-1°C warming hypothesis, but each of these has ended in failure. Contrarian scientists placed their faith in clouds to provide a strong negative feedback, for instance, but recently, measurements by Andrew Dessler have shown that the net effect of clouds is more positive than negative (see Science , Vol.330, 10 December 2010).

Perhaps as a result of realising the unsustainability of the idea of ultra-low climate sensitivity, a small sub-set of climate sceptics has emerged recently, the ‘lukewarmers’, who argue for a figure somewhere below that of the consensus view but above that of the classical contrarians. However, given that their evidence base is much smaller than the evidence for higher climate sensitivity, this group is in a very weak position to claim that there is no need decarbonise the global energy supply.

In conclusion, despite the complexity and ongoing uncertainty in understanding the future effects of greenhouse gases on the climate system, one thing is certain: the hypothesis that the effect of carbon dioxide and other greenhouse gases is trivial and warrants no action simply does not hold up. It does not match the facts. It has been refuted. Journalists may not be able to understand science or the philosophy of science to any great depth, but they can understand the concept of ‘disproven’, and climate scientists can indeed disprove the contrarian hypothesis that greenhouse gases will have no significant effect on the global climate.

© Dr Richard Lawson 2014

Richard Lawson is a retired general practitioner and psychiatrist, a blogger, and a veteran environmental activist.

This site uses cookies to recognize users and allow us to analyse site usage. By continuing to browse the site with cookies enabled in your browser, you consent to the use of cookies in accordance with our privacy policy . X

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

Long-term variation patterns of precipitations driven by climate change in china from 1901 to 2022.

1. Introduction

2. materials and methods, 2.1. study area, 2.2. data source and preprocessing, 2.3. research methods, 2.3.1. technical flowchart, 2.3.2. mann–kendall trend test, 2.3.3. mann–kendall mutation test, 2.3.4. center of gravity model, 2.3.5. validation of worldclim data, 3.1. spatial distribution of precipitation during 1901–2022, 3.2. temporal variations in precipitation during 1901–2022, 3.3. distribution characteristics of the centers of gravity in different basins, 4. discussion, 5. conclusions, author contributions, institutional review board statement, informed consent statement, data availability statement, conflicts of interest.

• Yao, H.X.; Li, Q.Q.; Zhao, L.; Wu, X.Q.; Shen, X.Y.; Duan, C.F.; Li, C. Evolution characteristics of compound drought and heat events during the warm season in the Huaihe River Basin and their relationship with climate and vegetation. Acta Ecol. Sin. 2024 , 44 , 5596–5608. [ Google Scholar ]
• Zhao, Z.C.; Luo, Y.; Huang, J.B. Global climate indicators, climatic-impact drivers and global warming. Clim. Chang. Res. 2024 , 20 , 384–388. [ Google Scholar ]
• Liang, P.D.; Liu, A.X.; Duan, L.Y.; Zhou, M.S.; Zhou, L.D. The Relationship between the Persistent Anomaly of Spring and Summer Atmospheric Circulation in the Middle Part of Asia and the Summer Drought/Waterlogging in Eastern China. Chin. J. Atmos. Sci. 2008 , 32 , 1174–1186. [ Google Scholar ]
• Xu, D.P.; Li, J.M.; Zhou, Z.H.; Liu, J.J.; Yan, Z.Q.; Wang, D.X. Study on the spatial and temporal distribution of precipitation characteristics in China from 1956 to 2018. Water Resour. Hydropower Eng. 2020 , 51 , 20–27. [ Google Scholar ]
• Chen, S.; Zhao, W.W.; Han, Y. Spatio-temporal variation of vegetation precipitation use efficiency and influencing factors in arid and semi-arid areas of China. Acta Ecol. Sin. 2023 , 43 , 10295–10307. [ Google Scholar ]
• Yang, Z.H.; Tuo, Y.; Yang, J.; Wu, Y.Z.; Gong, Z.Q.; Feng, G.L. Integrated prediction of summer in China based on multi dynamic-statistic methods. Chin. J. Geophys. 2024 , 67 , 982–996. [ Google Scholar ]
• Liu, X.P.; Tong, X.H.; Jia, Q.Y.; Xin, Z.H.; Yang, J.R. Precipitation concentration characteristics in China during 1960–2017. Adv. Water Sci. 2021 , 32 , 10–19. [ Google Scholar ]
• Wang, L.P.; Wang, Y.Z.; Xiang, X.; Sun, H.; Lian, Z.H. Research on classification of short-duration heavy rain based on Gumbel-Linear Moment and K-Means algorithm. Chin. J. Geophys. 2023 , 66 , 3171–3184. [ Google Scholar ]
• Hu, Y.Y.; Xiao, Y.; Dai, S.P.; Luo, H.X.; Li, Y.P.; Li, M.F. Spatio-temporal Variation Characteristics of Precipitation and Temperature in Tropical China from 1961 to 2015. Southwest China J. Agric. Sci. 2021 , 34 , 1788–1795. [ Google Scholar ]
• Lu, S.; Hu, Z.Y.; Wang, B.P.; Qin, P.; Wang, L. Spatio-temporal Patterns of Extreme Precipitation Events over China in Recent 56 Years. Plateau Meteorol. 2020 , 39 , 683–693. [ Google Scholar ]
• Deng, H.J.; Guo, B.; Cao, Y.Q.; Chen, Z.S.; Zhang, Y.Q.; Chen, X.W.; Gao, L.; Chen, Y.; Liu, M.B. Spatial and temporal patterns of daytime and nighttime precipitation in China during 1961–2016. Geogr. Res. 2020 , 39 , 2415–2426. [ Google Scholar ]
• Zhao, Z.H.; Luo, Z.J.; Huang, L.X.; Xing, L.T.; Sun, H.J.; Chen, H.L.; Sun, B. Analysis of Precipitation and Groundwater Variation Based on STL and Mann-Kendall Methods in Jinan City. J. China Hydrol. 2022 , 42 , 73–77. [ Google Scholar ]
• Oliver, J.E. Monthly precipitation distribution: A comparative index. Prof. Geogr. 1980 , 32 , 300–309. [ Google Scholar ] [ CrossRef ]
• Michiels, P.; Gabriels, D.; Hartmann, R. Using the seasonal and temporal Precipitation concentration index for characterizing the monthly rainfall distribution in Spain. Catena 1992 , 19 , 43–58. [ Google Scholar ] [ CrossRef ]
• Javier, M. Spatial distribution of a daily precipitation concentration index in peninsular Spain. Int. J. Climatol. 2004 , 24 , 959–971. [ Google Scholar ]
• Kong, F.; Shi, P.J.; Fang, J.; Lu, L.L.; Fang, J.Y.; Guo, J.P. Advances and Prospects of Spatiotemporal Pattern Variation of Extreme Precipitation and its Affecting Factors under the Background of Global Climate Change. J. Catastrophol. 2017 , 32 , 165–174. [ Google Scholar ]
• Liu, K.; Nie, G.G.; Zhang, S. Study on the spatiotemporal evolution of temperature and precipitation in China from 1951 to 2018. Adv. Earth Sci. 2020 , 35 , 1113–1126. [ Google Scholar ]
• He, Q.X.; Chen, C.F.; Wang, Y.H.; Sun, Y.N.; Liu, Y.T.; Hu, B.J. Fusion Method for Multi-Source Remote Sensing Daily Precipitation Data: Random Forest Model Considering Spatial Autocorrelation. J. Geo-Inf. Sci. 2024 , 26 , 1517–1530. [ Google Scholar ]
• Zhang, R.N.; Sun, C.H.; Li, W.J. Relationship between the interannual variations of Arctic sea ice and summer Eurasian teleconnection and associated on summer precipitation over China. Chin. J. Geophys. 2018 , 61 , 91–105. [ Google Scholar ]
• Zhang, L.J.; Qian, Y.F. Annual distribution features of precipitation in China and their interannual variations. Acta Meteorol. Sin. 2003 , 17 , 146–163. [ Google Scholar ]
• Peng, S.Z. 1-km monthly precipitation dataset for China (1901–2023). Natl. Tibet. Plateau/Third Pole Environ. Data Cent. 2020 , 12 , 15–24. [ Google Scholar ]
• Zhang, D. Study on the Characteristics of Nighttime Light Distribution and Urban Gravity Center Migration in Urumqi. J. Xinjiang Norm. Univ. 2021 , 40 , 29–34+51. [ Google Scholar ]
• Yan, J.; Li, H.; Wang, J.N. Prediction of the Spatio-Temporal Variation Characteristics of Precipitation in Central Asia Based on the CMIP6 Model. Sci. Technol. Inf. 2024 , 22 , 244–248. [ Google Scholar ]
• Gao, X.T.; Dong, J.; Yang, R.R. Temporal and Spatial Variation of Precipitation in North China During the Last 50 Years. Henan Sci. 2016 , 34 , 596–600. [ Google Scholar ]
• Liu, M.R.; Liu, S.T.; Ma, C.Y.; Li, H.; Chang, S.H.; Hou, F.J.; Liu, Y.J. Research progress of the responses of grassland plants and soil to the variation of temperature and precipitation. Chin. J. Ecol. 2023 , 8 , 125–137. [ Google Scholar ]
• Zhang, X.; Li, P.; Li, D. Spatiotemporal variations of precipitation in the southern part of the Heihe river basin (China), 1984–2014. Water 2018 , 10 , 410. [ Google Scholar ] [ CrossRef ]
• Lv, J.X.; Liu, C.M.; Liang, K.; Tian, W.; Bai, P.; Zhang, Y.H. Spatiotemporal variations of extreme precipitation in the Yellow River Basin based on water resources regionalization. Resour. Sci. 2022 , 44 , 261–273. [ Google Scholar ] [ CrossRef ]
• Wang, L.Z.; Miao, J.F.; Han, F.R. Overview of Impact of Topography on Precipitation in China over Last 10 Years. Meteorol. Sci. Technol. 2018 , 46 , 64–75. [ Google Scholar ]
• Du, Y.B.; Li, S.S.; Feng, D.; Xiao, Y.Q.; Chen, X.T.; Huang, S.N.; Du, L.L. Characteristics of Extreme Summer Precipitation and Large-Scale Cir-culation in Shaanxi Province under Global Warming. Plateau Meteorol. 2024 , 43 , 342–352. [ Google Scholar ]
• Wang, Q.F.; Chen, M.; Qi, X.H. Analysis of Multi-Year Precipitation and Evaporation Changes in Fujin City from 1992 to 2020. Water Resour. Dev. Manag. 2024 , 10 , 70–73+69. [ Google Scholar ]
• Zhang, Y.Y.; Xin, C.L.; Guo, X.Y.; Zhang, B.; Chen, N.; Shi, Y.F. Characteristics of Stable Isotopes in Precipitation and Its Moisture Sources in the Inland Regions of Northwest China. Environ. Sci. 2024 , 45 , 2080–2095. [ Google Scholar ]
• Lang, L.C.; Tang, C.; Gao, X.; Li, Z.H.; Wu, F. Spatial interpolation of high-resolution daily precipitation over complex terrains. Acta Geogr. Sin. 2023 , 78 , 101–120. [ Google Scholar ]
• Tang, W.R.; Fang, K.Y.; Mei, Z.P.; Wu, H.; Zhou, F.F.; Chen, Y.; Li, W.L. The influence of climate change on social evolution in Eurasia over the past millennium. Chin. Sci. Bull. 2024 , 1–12. [ Google Scholar ]
• Li, X.E.; Liu, Y.; Jiang, L.M.; Huang, R.G.; Zhou, Z.W.; Pang, X.G. A Review on Remote Sensing Monitoring of Glacial Lake Change and Glacial Lake Outburst Floods in Mountain Glacier Regions. J. Geo-Inf. Sci. 2024 , 26 , 1019–1039. [ Google Scholar ]
• Zhang, Y.R.; Zhou, D.G.; Guo, X.F. Regional climate response to global warming in the source region of the Yellow River and its impact on runoff. Sci. China Earth Sci. 2024 , 54 , 862–873. [ Google Scholar ] [ CrossRef ]
• Liu, J.X.; Zhu, Z.W.; Lu, R.; Li, J. Objective Classification of Sea Surface Temperature Evolution diversity of ENSO Cycle. Plateau Meteorol. 2024 , 4 , 87–102. [ Google Scholar ]
• Yang, B.L. Analysis of Runoff Change Characteristics and Driving Factors in the Upper Reaches of the Meijiang River Basin. Guangdong Water Resour. Hydropower 2022 , 7 , 43–47. [ Google Scholar ]
• Chen, L.J.; Wang, Y.Y.; Li, W.J.; Sun, L.H.; Li, X.; Zhang, D.Q. Review of the Influence and Application of SST Anomaly to Flood Season Precipitation in China. J. Appl. Meteorol. Sci. 2024 , 35 , 129–141. [ Google Scholar ]
• Hao, Z.C.; Chen, Y. Research progresses and prospects of multi-sphere compound extremes from the Earth System perspective. Sci. China Earth Sci. 2024 , 67 , 343–374. [ Google Scholar ] [ CrossRef ]
• Wang, J.W.; Ran, J.L.; Wu, W.; Shang, R.Y.; Zhang, Z.M.; Zhang, M.X.; Liu, J.K. Scaling Effects of Anthropogenic and Climate Change Impacts on Runoff in the Haihe River Basin. J. Ecol. Rural. Environ. 2024 , 40 , 757–765. [ Google Scholar ]

Click here to enlarge figure

Share and Cite

Han, J.; Zhang, R.; Guo, B.; Han, B.; Xu, T.; Guo, Q. Long-Term Variation Patterns of Precipitations Driven by Climate Change in China from 1901 to 2022. Sustainability 2024 , 16 , 7283. https://doi.org/10.3390/su16177283

Han J, Zhang R, Guo B, Han B, Xu T, Guo Q. Long-Term Variation Patterns of Precipitations Driven by Climate Change in China from 1901 to 2022. Sustainability . 2024; 16(17):7283. https://doi.org/10.3390/su16177283

Han, Jing, Rui Zhang, Bing Guo, Baomin Han, Tianhe Xu, and Qiang Guo. 2024. "Long-Term Variation Patterns of Precipitations Driven by Climate Change in China from 1901 to 2022" Sustainability 16, no. 17: 7283. https://doi.org/10.3390/su16177283

Article Metrics

Article access statistics, further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals

Featured Topics

Featured series.

A series of random questions answered by Harvard experts.

Explore the Gazette

Read the latest.

The answer to your search may depend on where you live

Putting human past on the MAPS

Does AI help humans make better decisions?

From a plant-free place, clues about how to help plants survive as planet warms.

Bonneville Salt Flats in Nevada.

Chris Boswell/iStock for Getty Images

Anne J. Manning

Harvard Staff Writer

Data from salt flats suggest dry soil is worse than rising temperature

The warming climate is having ripple effects across ecosystems, including plants, which have evolved clever mechanisms to conserve water when stressed by drought.

But are plants likelier to defend themselves against dry air or dry soil? This question is hotly debated among climate scientists, and the distinction matters: While there’s consensus on the trajectory of temperature rise over coming decades, less is known about how global warming will affect soil moisture. Understanding this dynamic may help decide the most effective ways to ensure the survival of robust plant life.

A team led by Kaighin McColl , assistant professor in the Department of Earth and Planetary Sciences and the John A. Paulson School of Engineering and Applied Sciences, have new research in Nature Water indicating that plant drought-defense mechanisms, which involve closing tiny pores on leaves called stomata to limit photosynthesis and conserve water, are more likely triggered by dry soil than by dry air.

Their results challenge recently held views and were derived from a place with no plants at all — the barren salt flats of Utah and Nevada.

Previous research had found that plants are likelier to have closed stomata in the presence of dry air, rather than dry soils, so it was assumed that aridity triggered the drought response. But McColl and colleagues suspected these results did not tell the whole story about plant vulnerability to drier environments.

Kaighin McColl went to the salt flats in the Western U.S. desert to conduct his research.

Alex Griswold/Harvard University Center for the Environment

“The problem with this argument is that correlation does not imply causation; when plants close their stomata, that could actually be causing the air to get drier, rather than the other way around,” McColl said.

To investigate their opposing hypothesis, McColl and lead author Lucas Vargas Zeppetello , a Harvard postdoctoral researcher who starts at the University of California, Berkeley, in January, used as their natural laboratory one of the only places on Earth that has a vigorous water cycle but doesn’t grow any plants — salt flats in the Western U.S. desert.

Using salt flats data provided by collaborators in Nevada and Utah, the researchers reproduced other researchers’ studies that had calculated the relationship between air dryness and moisture flux, or movement (in this case through evaporation), from the land surface and had attributed those values to plants closing their stomata to conserve water. The Harvard team found their calculations lined up almost perfectly with those previous studies, but with no plants in the salt flats, they knew there had to be another explanation.

In that plant-free environment, evaporation responds only to soil dryness. McColl and Vargas Zeppetello concluded that plant responses to lack of humidity may have been exaggerated in previous studies. They think instead that plants respond most acutely to dry soil, an environmental stressor that is known to reduce transpiration and photosynthesis.

What does this mean? Soil dryness matters more than air dryness when it comes to global plant ecosystems.

“Our findings put emphasis on projections for water in the future,” Vargas Zeppetello said. “People talk about consensus on climate change, but that really has to do with global temperatures. There’s much less of a consensus on what regional changes to the water cycle are going to look like.”

The research was supported in part by the National Science Foundation.

Share this article

You might like.

Researchers find ‘language bias’ in various site algorithms, raising concerns about fallout for social divisions among nations

Harvard digital atlas plots patterns from history ancient and modern

One judge’s track record — with and without algorithm — surprises researchers

The way forward for Democrats — and the country

Danielle Allen is more worried about identity politics and gaps in civic education than the power of delegates

Good genes are nice, but joy is better

Harvard study, almost 80 years old, has proved that embracing community helps us live longer, and be happier

Between bright light and a good mood, plenty of sleep

Researchers outline path to lower risk of depression