research paper in climate change

Accessibility Links

• Skip to content
• Skip to search IOPscience
• Skip to Journals list
• Accessibility help
• Accessibility Help

Click here to close this panel.

Purpose-led Publishing is a coalition of three not-for-profit publishers in the field of physical sciences: AIP Publishing, the American Physical Society and IOP Publishing.

Together, as publishers that will always put purpose above profit, we have defined a set of industry standards that underpin high-quality, ethical scholarly communications.

We are proudly declaring that science is our only shareholder.

Greater than 99% consensus on human caused climate change in the peer-reviewed scientific literature

Mark Lynas 4,1 , Benjamin Z Houlton 2 and Simon Perry 3

Published 19 October 2021 • © 2021 The Author(s). Published by IOP Publishing Ltd Environmental Research Letters , Volume 16 , Number 11 Citation Mark Lynas et al 2021 Environ. Res. Lett. 16 114005 DOI 10.1088/1748-9326/ac2966

You need an eReader or compatible software to experience the benefits of the ePub3 file format .

Article metrics

326336 Total downloads

Share this article

Author e-mails.

[email protected]

Author affiliations

1 Visiting Fellow, Cornell University, Global Development, Alliance for Science, B75 Mann Library, Ithaca, NY 14850, United States of America

2 Cornell University, Department of Ecology and Evolutionary Biology and Department of Global Development, Cornell University, Ithaca, NY 14850, United States of America

3 Alliance for Science, Ithaca, NY 14850, United States of America

Author notes

4 Author to whom any correspondence should be addressed.

• Received 7 June 2021
• Accepted 23 September 2021
• Published 19 October 2021

Peer review information

Method : Double-anonymous Revisions: 2 Screened for originality? Yes

Buy this article in print

While controls over the Earth's climate system have undergone rigorous hypothesis-testing since the 1800s, questions over the scientific consensus of the role of human activities in modern climate change continue to arise in public settings. We update previous efforts to quantify the scientific consensus on climate change by searching the recent literature for papers sceptical of anthropogenic-caused global warming. From a dataset of 88125 climate-related papers published since 2012, when this question was last addressed comprehensively, we examine a randomized subset of 3000 such publications. We also use a second sample-weighted approach that was specifically biased with keywords to help identify any sceptical peer-reviewed papers in the whole dataset. We identify four sceptical papers out of the sub-set of 3000, as evidenced by abstracts that were rated as implicitly or explicitly sceptical of human-caused global warming. In our sample utilizing pre-identified sceptical keywords we found 28 papers that were implicitly or explicitly sceptical. We conclude with high statistical confidence that the scientific consensus on human-caused contemporary climate change—expressed as a proportion of the total publications—exceeds 99% in the peer reviewed scientific literature.

Export citation and abstract BibTeX RIS

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 license . Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

1. Introduction

The extent of the scientific consensus on human-caused climate change is of great interest to society. If there remains substantial genuine scientific doubt about whether modern climate change is human-caused, then the case for mitigation of greenhouse gas emissions is weakened. By contrast, a widely-held consensus view in the peer-reviewed literature invalidates alternative arguments which claim that there is still significant debate in the scientific community about the reality of anthropogenic climate change (ACC).

The question of the cause of observed and predicted global warming and precipitation change is still highly politically salient. A Gallup poll published in April 2021 found that there has been a deepening of the partisan divide in American politics on whether observed increases in the planet's temperature since the Industrial Revolution are primarily caused by humans [ 1 ]. Among elected U.S. politicians the divide is similarly stark: according to the Center for American Progress there were 139 elected officials in the 117th Congress (sitting in 2021), including 109 representatives and 30 senators, 'who refuse to acknowledge the scientific evidence of human-caused climate change' [ 2 ]. In 2016 Pew Research found that only 27% of U.S. adults believed that 'almost all' scientists agreed that climate change is due to human activity [ 3 ].

Many efforts have been made over the years to quantify the extent of the scientific consensus on ACC [ 4 , 5 ]. These are comprehensively reviewed in a paper published in 2016 entitled 'Consensus on consensus' [ 6 ]. It has additionally been argued that perception of scientific consensus is a 'gateway belief' motivating wider public support for mitigation of climate change [ 7 ]. While scientific consensus does not sensu stricto prove a statement about the physical world, science-based anlaysis and hypothesis testing is capable of disproving alternative constructs, which could explain a given observation, either absolutely or relatively [ 8 ]. Hence, quantifying the scientific consensus clarifies the extent of any dissent in the scientific community in the process of disproval, and the plausible validity of alternative hypotheses in the face of scientific scrutiny, observations, and testing over time.

The most recent well-known effort to quantify the consensus was published in 2013, encompassing papers appearing in the peer-reviewed literature between 1991 and 2012, and sparked the famous headline that 97% of the world's science supported the climate change consensus [ 9 ]. The '97% consensus' view (published by Cook et al 2013, referred to hereafter as C13) had a big impact on global awareness of the scientific consensus on the role of greenhouse gases in causing climate change and was extensively covered in the media. Our primary motivation for this current study was to re-examine the literature published since 2012 to ascertain whether any change in the scientific consensus on climate change is discernible.

Previous attempts to quantify the consensus on climate change have employed many different methodologies, varying from expert elicitation to examination of abstracts returned by a keyword search. We base our methodology on C13 with some important refinements. We searched the Web of Science for English language 'articles' added between the dates of 2012 and November 2020 with the keywords 'climate change', 'global climate change' and 'global warming'. C13 used the latter two phrases but not 'climate change' without the preceding 'global'. (As discussed below, this was justified post-facto in our study because the majority of sceptical papers we found would not have been returned had we used the same search phrases as C13.) This wider set of search terms yielded a total of 88125 papers, whereas C13 identified a total of 11944 abstracts from papers published over the years 1991 and 2011. (Using our expanded search terms over the same 1991–2011 time period as C13 would have yielded 30627 results.)

Given the large number of papers found using our approach we randomly sub-sampled 3000 abstracts out of the 88125 total papers identified in our search, and subsequently categorized them in accordance with C13 (See table 1 ).

Table 1.  Categorization of climate papers, as per C13.

As per C13 we rated the abstracts of papers, assigning them numbers according to their level of implicit or explicit endorsement or rejection of ACC (table 2 ). Abstracts were rated with only the title and abstract visible; information about authors, date and journal were hidden at this stage.

Table 2.  Rating of climate papers, as per C13.

To further extend our approach for identifying as many sceptical papers as possible within the full dataset, we created an algorithm to identify keywords within the papers rated by C13 as sceptical that had appeared more often in sceptical papers than consensus papers. The software counted the appearance of every word in the title, author list and abstract of every sceptical paper. For each word that appeared in at least two papers, the algorithm counted the number of sceptical and consensus papers it appeared in to calculate its predictive power. We took the 150 most predictive words, then manually reviewed them to remove words that appeared to be there by chance (e.g. 'walk' and 'nearest') leaving those we believed could be predictively useful (e.g. 'cosmic' and 'rays'). A second algorithm then scored all 88125 papers (including the 3000 sampled separately earlier) based on the appearance of the predictive words. (See supplementary info for precise details of this exercise (available online at stacks.iop.org/ERL/16/114005/mmedia )). We then rated and categorized the 1000 papers with the highest score using the same approaches from C13 as detailed in tables 1 and 2 . As stated earlier, this approach was taken in order to increase the chances of us finding sceptical papers in the full dataset, allowing for a robust assessment and inclusion of any dissent.

In contrast to C13, we did not perform an author elicitation survey asking authors to carry out a self-rating of their papers.

3.1. Results of random sampling

Our random sample of 3000 papers revealed a total of 282 papers that were categorized as 'not climate-related'. These false-positives occurred because, even though the climate keywords occurred in their title/abstracts, the published articles dealt with social science, education or research about people's views on climate change rather than original scientific work. Hence, we excluded these papers in accordance with C13's approach. We then assessed the remaining total of 2718 papers in the data set and found four that argued against the scientific consensus of ACC.

The ratings and categorizations for the 3000 randomly sampled papers are shown in table 3 . Note that 'not climate-related' papers are displayed in table 3 for completeness. Figure 1 shows the same data, but with 'not climate-related' papers excluded.

Figure 1.   Ratings and categorizations given to 2718 randomly-sampled climate abstracts.

Download figure:

Table 3.  Results of rating and categorization of 3000 abstracts.

Our estimate of the proportion of consensus papers was 1 − (4/2718) = 99.85%. The 95% confidence limits for this proportion are 99.62%–99.96% (see R code in supplementary info), therefore it is likely that the proportion of climate papers that favour the consensus is at least 99.62%.

Recalculating at the 99.999% confidence level gives us the interval 99.212%–99.996%, therefore it is virtually certain that the proportion of climate papers that do not dispute that the consensus is above 99.212%.

If we repeat the methods of C13 and further exclude papers that take no position on AGW (i.e. those rated 4a), we estimate the proportion of consensus papers to be 99.53% with the 95% confidence interval being 98.80%–99.87%.

3.2. Keywords indicating scepticism

We reviewed the 1000 studies that our keyword matching software identified as most likely to be sceptical out of the entire 88125 dataset. After manual review, 28 sceptical papers within the most likely 1000 papers were identified, with the majority being in the top rows of the dataset. The first paper was sceptical, as were 12 out of the first 50, and 16 out of the first 100. (See supplementary data for the full list.) Table 4 shows how the 1000 studies that the keywords found to be most likely to be sceptical were rated.

Table 4.  Ratings and categorizations for the 1000 abstracts most likely to be sceptical.

In other words, the predictive keywords successfully allowed us to identify a total of 28 papers from the full dataset of 88125 which appeared implicitly or explicitly sceptical of ACC. Only one of these papers had already appeared in the first 3000 randomized sample. While we are aware that this approach does not reveal all sceptical papers that exist in the full dataset, it provides an absolute upper bound to the percentage of papers that agree with the consensus. Knowing that at least 31 (including the 3 additional papers found in the random sample) out of the full 88125 dataset are sceptical, we can say the consensus on ACC is at most 99.966%.

4. Discussion

Our analysis demonstrates >99% agreement in the peer-reviewed scientific literature on the principal role of greenhouse gas (GHG) emissions from human activities in driving modern climate change (i.e. since the Industrial Revolution). This result further advances our understanding of the scientific consensus view on climate change as evidenced by the peer reviewed scientific literature, and provides additional evidence that the statements made by the Intergovernmental Panel on Climate Change (see below) accurately reflect the overwhelming view of the international scientific community. We conclude that alternative explanations for the dominant cause of modern (i.e., post-industrial) climate change beyond the role of rising GHG emissions from human activities are exceedingly rare in the peer-reviewed scientific literature.

Previous researchers have debated how to define and therefore quantify 'consensus' in the scientific literature on an array of issues. While C13 define consensus rather narrowly as explicit or implicit agreement, a broader definition can be employed which defines consensus as lack of objection to a prevailing position or worldview. In 2015 James Powell argued for this broader definition, pointing out that the C13 methodology, if applied to other scientific research areas such as plate tectonics or evolution, would fail to find consensus because few authors of papers in the expert literature feel the need to re-state their adherence in both cases to what has long been universally-accepted theory [ 10 ].

In a rejoinder to this critique, several C13 authors argued that their narrower definition of consensus was still relevant in other well-established fields if both implicit and explicit agreement was included [ 11 ]. Therefore in plate tectonics, for example, sea-floor spreading, mountain-building by means of continental collision, subduction etc, could be implicitly supportive of a consensus on the reality of the theory of plate tectonics without this necessarily being explicitly stated.

In our paper, for the sake of clarity and comparability with previous literature, we present results using both approaches. However, having reviewed, rated and categorized several thousand papers we believe that there is now a stronger case for the broader approach given how widely accepted ACC has become in the peer-reviewed literature. For example, a majority of the papers we categorized as being about 'impacts' of climate change did not state a position on whether the phenomenon they were studying—the changing climate—was human-caused. It seems highly unlikely that if researchers felt sceptical about the reality of ACC they would publish numerous studies of its impacts without ever raising the question of attribution.

In other words, given that most 4a ('no position') ratings do not either explicitly or implicitly differ from the consensus view of GHG emissions as the principal driver of climate change it does not follow in our view that these analyses should be a priori excluded from the consensus. In another example, we gave rating '2' ('explicit endorsement without quantification') to all papers referencing future emissions scenarios in their abstracts, because emissions scenarios by definition imply an evaluation of humanity's role in GHG emissions and their subsequent impact on climate. Thus the authors choice of wording on emissions scenarios or other issues implying human causation to climate change in the abstracts of their climate impact studies might lead to arbitrariness if these were taken as the sole indicators of the authors' adherence to the consensus on ACC.

In addition, decisions about whether to give rating '3' ('implicit endorsement') are subjective in that the rating of a position on ACC is considered to be implied by the authors without this being explicitly stated in the abstract of a paper. Thus subjective judgements by those doing the ratings about the implicit meanings communicated by abstract wording choices of paper authors are critical to the numerical consensus result obtained using C13's method, potentially introducing a source of bias. It is unclear to us why this is preferable to defining consensus in a clearer and more objectively transparent way as simply the absence of clearly-stated rejection or disagreement.

We also note that our keyword choices, in particular not requiring the word 'global' in front of 'climate change' led to our discovery of many sceptical papers that would not have been identified by searches only of 'global warming' and 'global climate change'. This suggests—but does not prove—that a number of sceptical papers may have been missed in the original C13 study. However, these minor disagreements aside, we are indebted to C13 for the rigor of their methodology, much of which we re-employ directly here.

4.1. Review of sceptical papers

In supplementary table 1 we present the full list of all 31 sceptical papers we found in our dataset. An in-depth evaluation of their merits is outside the scope of this paper, and could be an interesting area for further work. We note some recurring themes however, such as the hypothesis that changes in cosmic rays are significantly influencing the Earth's changes in climate, that the Sun is driving modern climate change, or that natural fluctuations are somehow involved. An additional area of research might investigate how far these themes in the published literature are reflected in popular discourse outside of the scientific community.

5. Conclusion

Our results confirm, as has been found in numerous other previous studies of this question, that there is no significant scientific debate among experts about whether or not climate change is human-caused. This issue has been comprehensively settled, and the reality of ACC is no more in contention among scientists than is plate tectonics or evolution. The tiny number of papers that have been published during our time period which disagree with this overwhelming scientific consensus have had no discernible impact, presumably because they do not provide any convincing evidence to refute the hypothesis that—in the words of IPCC AR5—'it is extremely likely that human influence has been the dominant cause of the observed warming since the mid-20th century' [ 12 ], and, most recently in IPCC AR6—'it is unequivocal that human influence has warmed the atmosphere, ocean and land' [ 13 ].

Our finding is that the broadly-defined scientific consensus likely far exceeds 99% regarding the role of anthropogenic GHG emissions in modern climate change, and may even be as high as 99.9%. Of course, the prevalence of mis/disinformation about the role of GHG emissions in modern climate change is unlikely to be driven purely by genuine scientific illiteracy or lack of understanding [ 14 ]. Even so, in our view it remains important to continue to inform society on the state of the evidence. According to the IPCC AR6 summary and many other previous studies, mitigating future warming requires urgent efforts to eliminate fossil fuels combustion and other major sources of anthropogenic greenhouse gas emissions. Our study helps confirm that there is no remaining scientific uncertainty about the urgency and gravity of this task.

Acknowledgments

The authors would like to thank David Colquhoun for helpful discussions about the statistical methodology. We would also like to thank John Cook for useful comments on an early draft, and Sarah Evanega at the Alliance for Science. Support for the Alliance for Science is provided by the Bill & Melinda Gates Foundation.

Data availability statement

All data that support the findings of this study are included within the article (and any supplementary files).

Author contributions

M L conceived the paper. S P wrote software and algorithms for data extraction and performed data analysis. M L performed ratings and categorizations. B H, S P and M L wrote the paper.

Supplementary information

Random sample of 3000 papers

Initial data with random order added.xlsx

wordAnalysis-flagged

Studies predicted to be sceptical by manually selected predictive keywords

Climate Change: Evidence and Causes: Update 2020 (2020)

Chapter: conclusion, c onclusion.

This document explains that there are well-understood physical mechanisms by which changes in the amounts of greenhouse gases cause climate changes. It discusses the evidence that the concentrations of these gases in the atmosphere have increased and are still increasing rapidly, that climate change is occurring, and that most of the recent change is almost certainly due to emissions of greenhouse gases caused by human activities. Further climate change is inevitable; if emissions of greenhouse gases continue unabated, future changes will substantially exceed those that have occurred so far. There remains a range of estimates of the magnitude and regional expression of future change, but increases in the extremes of climate that can adversely affect natural ecosystems and human activities and infrastructure are expected.

Citizens and governments can choose among several options (or a mixture of those options) in response to this information: they can change their pattern of energy production and usage in order to limit emissions of greenhouse gases and hence the magnitude of climate changes; they can wait for changes to occur and accept the losses, damage, and suffering that arise; they can adapt to actual and expected changes as much as possible; or they can seek as yet unproven “geoengineering” solutions to counteract some of the climate changes that would otherwise occur. Each of these options has risks, attractions and costs, and what is actually done may be a mixture of these different options. Different nations and communities will vary in their vulnerability and their capacity to adapt. There is an important debate to be had about choices among these options, to decide what is best for each group or nation, and most importantly for the global population as a whole. The options have to be discussed at a global scale because in many cases those communities that are most vulnerable control few of the emissions, either past or future. Our description of the science of climate change, with both its facts and its uncertainties, is offered as a basis to inform that policy debate.

A CKNOWLEDGEMENTS

The following individuals served as the primary writing team for the 2014 and 2020 editions of this document:

• Eric Wolff FRS, (UK lead), University of Cambridge
• Inez Fung (NAS, US lead), University of California, Berkeley
• Brian Hoskins FRS, Grantham Institute for Climate Change
• John F.B. Mitchell FRS, UK Met Office
• Tim Palmer FRS, University of Oxford
• Benjamin Santer (NAS), Lawrence Livermore National Laboratory
• John Shepherd FRS, University of Southampton
• Keith Shine FRS, University of Reading.
• Susan Solomon (NAS), Massachusetts Institute of Technology
• Kevin Trenberth, National Center for Atmospheric Research
• John Walsh, University of Alaska, Fairbanks
• Don Wuebbles, University of Illinois

Staff support for the 2020 revision was provided by Richard Walker, Amanda Purcell, Nancy Huddleston, and Michael Hudson. We offer special thanks to Rebecca Lindsey and NOAA Climate.gov for providing data and figure updates.

The following individuals served as reviewers of the 2014 document in accordance with procedures approved by the Royal Society and the National Academy of Sciences:

• Richard Alley (NAS), Department of Geosciences, Pennsylvania State University
• Alec Broers FRS, Former President of the Royal Academy of Engineering
• Harry Elderfield FRS, Department of Earth Sciences, University of Cambridge
• Joanna Haigh FRS, Professor of Atmospheric Physics, Imperial College London
• Isaac Held (NAS), NOAA Geophysical Fluid Dynamics Laboratory
• John Kutzbach (NAS), Center for Climatic Research, University of Wisconsin
• Jerry Meehl, Senior Scientist, National Center for Atmospheric Research
• John Pendry FRS, Imperial College London
• John Pyle FRS, Department of Chemistry, University of Cambridge
• Gavin Schmidt, NASA Goddard Space Flight Center
• Emily Shuckburgh, British Antarctic Survey
• Gabrielle Walker, Journalist
• Andrew Watson FRS, University of East Anglia

The Support for the 2014 Edition was provided by NAS Endowment Funds. We offer sincere thanks to the Ralph J. and Carol M. Cicerone Endowment for NAS Missions for supporting the production of this 2020 Edition.

F OR FURTHER READING

For more detailed discussion of the topics addressed in this document (including references to the underlying original research), see:

• Intergovernmental Panel on Climate Change (IPCC), 2019: Special Report on the Ocean and Cryosphere in a Changing Climate [ https://www.ipcc.ch/srocc ]
• National Academies of Sciences, Engineering, and Medicine (NASEM), 2019: Negative Emissions Technologies and Reliable Sequestration: A Research Agenda [ https://www.nap.edu/catalog/25259 ]
• Royal Society, 2018: Greenhouse gas removal [ https://raeng.org.uk/greenhousegasremoval ]
• U.S. Global Change Research Program (USGCRP), 2018: Fourth National Climate Assessment Volume II: Impacts, Risks, and Adaptation in the United States [ https://nca2018.globalchange.gov ]
• IPCC, 2018: Global Warming of 1.5°C [ https://www.ipcc.ch/sr15 ]
• USGCRP, 2017: Fourth National Climate Assessment Volume I: Climate Science Special Reports [ https://science2017.globalchange.gov ]
• NASEM, 2016: Attribution of Extreme Weather Events in the Context of Climate Change [ https://www.nap.edu/catalog/21852 ]
• IPCC, 2013: Fifth Assessment Report (AR5) Working Group 1. Climate Change 2013: The Physical Science Basis [ https://www.ipcc.ch/report/ar5/wg1 ]
• NRC, 2013: Abrupt Impacts of Climate Change: Anticipating Surprises [ https://www.nap.edu/catalog/18373 ]
• NRC, 2011: Climate Stabilization Targets: Emissions, Concentrations, and Impacts Over Decades to Millennia [ https://www.nap.edu/catalog/12877 ]
• Royal Society 2010: Climate Change: A Summary of the Science [ https://royalsociety.org/topics-policy/publications/2010/climate-change-summary-science ]
• NRC, 2010: America’s Climate Choices: Advancing the Science of Climate Change [ https://www.nap.edu/catalog/12782 ]

Much of the original data underlying the scientific findings discussed here are available at:

• https://data.ucar.edu/
• https://climatedataguide.ucar.edu
• https://iridl.ldeo.columbia.edu
• https://ess-dive.lbl.gov/
• https://www.ncdc.noaa.gov/
• https://www.esrl.noaa.gov/gmd/ccgg/trends/
• http://scrippsco2.ucsd.edu
• http://hahana.soest.hawaii.edu/hot/

Climate change is one of the defining issues of our time. It is now more certain than ever, based on many lines of evidence, that humans are changing Earth's climate. The Royal Society and the US National Academy of Sciences, with their similar missions to promote the use of science to benefit society and to inform critical policy debates, produced the original Climate Change: Evidence and Causes in 2014. It was written and reviewed by a UK-US team of leading climate scientists. This new edition, prepared by the same author team, has been updated with the most recent climate data and scientific analyses, all of which reinforce our understanding of human-caused climate change.

Scientific information is a vital component for society to make informed decisions about how to reduce the magnitude of climate change and how to adapt to its impacts. This booklet serves as a key reference document for decision makers, policy makers, educators, and others seeking authoritative answers about the current state of climate-change science.

READ FREE ONLINE

Welcome to OpenBook!

You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

Do you want to take a quick tour of the OpenBook's features?

Show this book's table of contents , where you can jump to any chapter by name.

...or use these buttons to go back to the previous chapter or skip to the next one.

Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

Switch between the Original Pages , where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

To search the entire text of this book, type in your search term here and press Enter .

Share a link to this book page on your preferred social network or via email.

View our suggested citation for this chapter.

Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

Get Email Updates

Do you enjoy reading reports from the Academies online for free ? Sign up for email notifications and we'll let you know about new publications in your areas of interest when they're released.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.33(2); Fall 2010

Climate Change: The Evidence and Our Options

Glaciers serve as early indicators of climate change. Over the last 35 years, our research team has recovered ice-core records of climatic and environmental variations from the polar regions and from low-latitude high-elevation ice fields from 16 countries. The ongoing widespread melting of high-elevation glaciers and ice caps, particularly in low to middle latitudes, provides some of the strongest evidence to date that a large-scale, pervasive, and, in some cases, rapid change in Earth's climate system is underway. This paper highlights observations of 20th and 21st century glacier shrinkage in the Andes, the Himalayas, and on Mount Kilimanjaro. Ice cores retrieved from shrinking glaciers around the world confirm their continuous existence for periods ranging from hundreds of years to multiple millennia, suggesting that climatological conditions that dominate those regions today are different from those under which these ice fields originally accumulated and have been sustained. The current warming is therefore unusual when viewed from the millennial perspective provided by multiple lines of proxy evidence and the 160-year record of direct temperature measurements. Despite all this evidence, plus the well-documented continual increase in atmospheric greenhouse gas concentrations, societies have taken little action to address this global-scale problem. Hence, the rate of global carbon dioxide emissions continues to accelerate. As a result of our inaction, we have three options: mitigation, adaptation, and suffering.

Climatologists, like other scientists, tend to be a stolid group. We are not given to theatrical rantings about falling skies. Most of us are far more comfortable in our laboratories or gathering data in the field than we are giving interviews to journalists or speaking before Congressional committees. Why then are climatologists speaking out about the dangers of global warming? The answer is that virtually all of us are now convinced that global warming poses a clear and present danger to civilization ( “Climate Change,” 2010 ).

That bold statement may seem like hyperbole, but there is now a very clear pattern in the scientific evidence documenting that the earth is warming, that warming is due largely to human activity, that warming is causing important changes in climate, and that rapid and potentially catastrophic changes in the near future are very possible. This pattern emerges not, as is so often suggested, simply from computer simulations, but from the weight and balance of the empirical evidence as well.

THE EVIDENCE

Figure 1 shows northern hemisphere temperature profiles for the last 1,000 years from a variety of high-resolution climate recorders such as glacier lengths ( Oerlemans, 2005 ), tree rings ( Briffa, Jones, Schwerngruber, Shiyatov, & Cook, 2002 ; Esper, Cook, & Schweingruber, 2002 ), and combined sources that include some or all of the following: tree rings, sediment cores, ice cores, corals, and historical records ( Crowley & Lowery, 2000 ; Jones, Briffa, Barnett, & Tett, 1998 ; Mann, Bradley, & Hughes, 1999 ; Moberg, Sonechkin, Holmgrem, Datsenko, & Karlen, 2005 ). The heavy gray line is a composite of all these temperatures ( Mann & Jones, 2003 ), and the heavy black line depicts actual thermometer readings back to 1850 (see National Research Council, 2006 , for a review of surface temperature reconstructions). Although the various curves differ from one another, their general shapes are similar. Each data source shows that average northern hemisphere temperatures remained relatively stable until the late 20th century. It is the agreement of these diverse data sets and the pattern that make climatologists confident that the warming trend is real.

A variety of temperature records over the last 1,000 years, based on a variety of proxy recorders such as tree rings, ice cores, historical records, instrumental data, etc., shows the extent of the recent warming. The range of temperature projected by Meehl et al. (2007) to 2100 AD is shown by the shaded region, and the average of the range is depicted by the filled circle.

Because these temperature numbers are based on northern hemisphere averages, they do not reflect regional, seasonal, and altitudinal variations. For example, the average temperature in the western United States is rising more rapidly than in the eastern part of the country, and on average winters are warming faster than summers ( Meehl, Arblaster, & Tebaldi, 2007 ). The most severe temperature increases appear to be concentrated in the Arctic and over the Antarctic Peninsula as well as within the interior of the large continents. This variability complicates matters, and adds to the difficulty of convincing the public, and even scientists in other fields, that global warming is occurring. Because of this, it may be useful to examine another kind of evidence: melting ice.

Retreat of Mountain Glaciers

The world's mountain glaciers and ice caps contain less than 4% of the world's ice cover, but they provide invaluable information about changes in climate. Because glaciers are smaller and thinner than the polar ice sheets, their ratio of surface area to volume is much greater; thus, they respond more quickly to temperature changes. In addition, warming trends are amplified at higher altitudes where most glaciers are located ( Bradley, Keimig, Diaz, & Hardy, 2009 ; Bradley, Vuille, Diaz, & Vergara, 2006 ). Thus, glaciers provide an early warning system of climate change; they are our “canaries in the coal mine.”

Consider the glaciers of Africa's Mount Kilimanjaro ( Figure 2 ). Using a combination of terrestrial photogrammetric maps, satellite images, and aerial photographs, we have determined that the ice fields on Kibo, the highest crater on Kilimanjaro, have lost 85% of their coverage since 1912 ( Thompson, Brecher, Mosley-Thompson, Hardy, & Mark, 2009 ).

The retreat of glaciers on Mount Kilimanjaro can be seen in the photographs from 1912, 1970, 2000, and 2006; from 1912 to 2006, 85% of the ice has disappeared.

Figure 3 shows a series of aerial photographs of Furtwängler glacier, in the center of Kibo crater, taken between 2000 and 2007, when the glacier split into two sections. As Furtwängler recedes, it is also thinning rapidly, from 9.5 m in 2000 to 4.7 m in 2009 (for more images of Furtwängler's retreat, see http://www.examiner.com/examiner/x-10722-Orlando-Science-Policy-Examiner∼y2009m11d2-Mt-Kilimanjaros-Furtwängler-Glacier-in-retreat ). If you connect the dots on the changes seen to date and assume the same rate of loss in the future, within the next decade many of the glaciers of Kilimanjaro, a Swahili word meaning “shining mountain,” will have disappeared.

Deterioration of the Furtwängler glacier in the center of Kibo crater on Mount Kilimanjaro. Since 2000 the ice field has decreased in size and thickness and has divided in two.

The Quelccaya ice cap, which is located in southern Peru adjacent to the Amazon Basin, is the largest tropical ice field on Earth. Quelccaya has several outlet glaciers, glaciers that extend from the edges of an ice cap like fingers from a hand. The retreat of one of these, Qori Kalis, has been studied and photographed since 1963. At the beginning of this study, Qori Kalis extended 1,200 m out from the ice cap, and there was no melt water at the end ( Figure 4 , map top left). By the summer of 2008, Qori Kalis had retreated to the very edge of Quelccaya, leaving behind an 84-acre lake, 60 m deep. Over the years, a boulder near the base camp has served as a benchmark against which to record the changes in the position of the edge of the ice. In 1977 the ice was actually pushing against the boulder ( Figure 5 , top), but by 2006 a substantial gap had appeared and been filled by a lake ( Figure 5 , bottom). Thus, the loss of Quelccaya's ice is not only on the Qori Kalis glacier but also on the margin of the ice cap itself. Since 1978, about 25% of this tropical ice cap has disappeared.

Retreat of the Qori Kalis outlet glacier on the Quelccaya ice cap. Each line shows the extent of the ice. The photos along the bottom provide a pictorial history of the melting of the Qori Kalis outlet glacier and the formation of a lake. The retreat of Qori Kalis is similar to the loss of several Peruvian glaciers, as shown in the graph insert.

Top: photo taken in 1978 shows a margin of the Quelccaya ice cap pushing against a boulder. Bottom: the same margin is shown in a 2005 photo. The ice has receded and has been replaced by a small lake. The boulder shown in the top photo is located in the center of the white circle to the right.

The Himalayan Mountains are home to more than 15,000 glaciers. Unfortunately, only a few of these glaciers have been monitored over an extended period, so reliable ground observations that are crucial for determining regional retreat rates do not yet exist. However, a recent study of an ice core from the Naimona'nyi glacier in the southwestern Himalayas ( Kehrwald et al., 2008 ) shows that ice is disappearing from the top of the glacier, as shown by the lack of the radioactive bomb layers from the 1950s and early 1960s that appear in all Tibetan and Himalayan ice core records ( Thompson, 2000 ; Thompson et al., 1990 , 1997 , 2006 ).

Glaciologists at the Institute of Tibetan Plateau Research in Beijing have been monitoring 612 glaciers across the High Asian region since 1980. These scientists found that from 1980 to 1990, 90% of these glaciers were retreating; from 1990 to 2005, the proportion of retreating glaciers increased to 95% ( Yao, Pu, Lu, Wang, & Yu, 2007 ).

A study of 67 glaciers in Alaska from the mid-1950s to the mid-1990s shows that all are thinning ( Arendt, Echelmeyer, Harrison, Lingle, & Valentine, 2002 ). In northern Alaska's Brooks Range, 100% of the glaciers are in retreat, and in southeastern Alaska 98% are shrinking ( Molnia, 2007 ). Glacier National Park in Montana contained more than 100 glaciers when it was established in 1910. Today, just 26 remain, and at the current rate of decrease it is estimated that by 2030 there will be no glaciers in Glacier National Park ( Hall & Fagre, 2003 ). The oldest glacier photos come from the Alps. Ninety-nine percent of the glaciers in the Alps are retreating, and 92% of Chile's Andean glaciers are retreating ( Vince, 2010 ).

The pattern described here is repeated around the world. Mountain glaciers nearly everywhere are retreating.

Loss of Polar Ice

Satellite documentation of the area covered by sea ice in the Arctic Ocean extends back three decades. This area, measured each September, decreased at a rate of about 8.6% per decade from 1979 to 2007. In 2007 alone, 24% of the ice disappeared. In 2006 the Northwest Passage was ice free for the first time in recorded history.

As noted earlier, polar ice sheets are slower to respond to temperature rise than the smaller mountain glaciers, but they, too, are melting. The Greenland ice sheet has also experienced dramatic ice melt in recent years. There has been an increase in both the number and the size of lakes in the southern part of the ice sheet, and crevices can serve as conduits (called moulins) that transport meltwater rapidly into the glacier. Water has been observed flowing through these moulins down to the bottom of the ice sheet where it acts as a lubricant that speeds the flow of ice to the sea ( Das et al., 2008 ; Zwally et al., 2002 ).

The ice in Antarctica is also melting. The late John Mercer, a glacial geologist at The Ohio State University, long ago concluded that the first evidence of global warming due to increasing carbon dioxide (CO 2 ) would be the breakup of the Antarctic ice shelves ( Mercer, 1978 ). Mean temperatures on the Antarctic Peninsula have risen 2.5° C (4.5° F) in the last 50 years, resulting in the breakup of the ice shelves in just the way Mercer predicted. One of the most rapid of these shelf deteriorations occurred in 2002, when the Larsen B, a body of ice over 200 m deep that covered an area the size of Rhode Island, collapsed in just 31 days (see images http://earthobservatory.nasa.gov/IOTD/view.php?id = 2351). An ice shelf is essentially an iceberg attached to land ice. Just as an ice cube does not raise the water level in a glass when it melts, so a melting ice shelf leaves sea levels unchanged. But ice shelves serve as buttresses to glaciers on land, and when those ice shelves collapse it speeds the flow of the glaciers they were holding back into the ocean, which causes sea level to rise rapidly.

Just days before this paper went to press, a giant ice island four times the size of Manhattan broke off the Petermann glacier in Greenland. This event alone does not prove global climate change, because half of the ice loss from Greenland each year comes from icebergs calving from the margins. It is the fact that this event is part of a long-term trend of increasing rates of ice loss, coupled with the fact that temperature is increasing in this region at the rate of 2° C (3.6° F) per decade, that indicates that larger scale global climate change is underway.

The loss of ice in the Arctic and Antarctic regions is especially troubling because these are the locations of the largest ice sheets in the world. Of the land ice on the planet, 96% is found on Greenland and Antarctica. Should all this ice melt, sea level would rise over 64 m ( Church et al., 2001 ; Lemke et al., 2007 ), and of course the actual sea level would be much higher due to thermal expansion of the world's oceans as they warm.

Although research shows some variability in the rate of ice loss, it is clear that mountain glaciers and polar ice sheets are melting, and there is no plausible explanation for this but global warming. Add to this the laboratory evidence and the meteorological measurements, and the case for global warming cannot be denied. So what causes global temperatures to rise?

CAUSES OF GLOBAL WARMING

Climatologists strive to reconstruct past climate variations on regional and global scales, but they also try to determine the mechanisms, called forcers , that drive climate change. Climatologists recognize two basic categories of forcers. Natural forcers are recurring processes that have been around for millions of years; anthropogenic forcers are more recent processes caused by human activity.

One familiar natural forcer is the earth's orbit around the sun, which gives us our seasons. In the northern hemisphere, June is warm because the sun's rays fall more directly on it, and the sun appears high in the sky; in the southern hemisphere, June is cool because the sun's rays hit the earth at a deep angle, and the sun appears low in the sky.

Less obvious natural forcers include short- and long-term changes in the atmosphere and ocean. For example, when Mount Pinatubo erupted in the Philippines in 1991, it spewed millions of tons of sulfuric gases and ash particles high into the atmosphere, blocking the sun's rays. This lowered global temperatures for the next few years. Another natural forcer is the linked oceanic and atmospheric system in the equatorial Pacific Ocean known as the El Niño-Southern Oscillation (ENSO). ENSO occurs every 3 to 7 years in the tropical Pacific and brings warm, wet weather to some regions and cool, dry weather to other areas.

Other natural forcers include periodic changes in energy from the sun. These include the 11- to 12-year sunspot cycle and the 70- to 90-year Wolf-Gleissberg cycle, a modulation of the amplitude of the 11-year solar cycle. These changes in solar energy can affect atmospheric temperature across large regions for hundreds of years and may have caused the “medieval climate anomaly” in the northern hemisphere that lasted from about 1100 AD to 1300 AD. Solar cycles may also have played a role in the cause of the “little ice age” in North America and Europe during the 16th to 19th centuries. These changes in climate, which are often cited by those who dismiss global warming as a normal, cyclical event, affected large areas, but not the Earth as a whole. The medieval climate anomaly showed warmth that matches or exceeds that of the past decade in some regions, but it fell well below recent levels globally ( Mann et al., 2009 ).

The most powerful natural forcers are variations in the orbit of the Earth around the Sun, which last from 22,000 to 100,000 years. These “orbital forcings” are partly responsible for both the ice ages (the glacial periods during which large regions at high and midddle latitudes are covered by thick ice sheets), and for the warm interglacial periods such as the present Holocene epoch which began about 10,000 years ago.

There is consensus among climatologists that the warming trend we have been experiencing for the past 100 years or so cannot be accounted for by any of the known natural forcers. Sunspot cycles, for example, can increase the sun's output, raising temperatures in our atmosphere. We are seeing a temperature increase in the troposphere, the lower level of our atmosphere, and a temperature decrease in the stratosphere, the upper level. But this is the exact opposite of what we would get if increased solar energy were responsible. Similarly, global temperatures have increased more at night than during the day, again the opposite of what would occur if the sun were driving global warming. In addition, temperatures have risen more in winter than in summer. This, too, is the opposite of what would be expected if the sun were responsible for the planet's warming. High latitudes have warmed more than low latitudes, and because we get more radiation from the sun at low latitudes, we again would expect the opposite if the sun were driving these changes. Thus, changes in solar output cannot account for the current period of global warming ( Meehl et al., 2007 ). ENSO and other natural forcers also fail to explain the steady, rapid rise in the earth's temperature. The inescapable conclusion is that the rise in temperature is due to anthropogenic forces, that is, human behavior.

The relatively mild temperatures of the past 10,000 years have been maintained by the greenhouse effect, a natural phenomenon. As orbital forcing brought the last ice age to an end, the oceans warmed, releasing CO 2 into the atmosphere, where it trapped infrared energy reflected from the earth's surface. This warmed the planet. The greenhouse effect is a natural, self-regulating process that is absolutely essential to sustain life on the planet. However, it is not immutable. Change the level of greenhouse gases in the atmosphere, and the planet heats up or cools down.

Greenhouse gases are captured in ice, so ice cores allow us to see the levels of greenhouse gases in ages past. The longest ice core ever recovered (from the European Project for Ice Coring in Antarctica) takes us 800,000 years back in time, and includes a history of CO 2 and methane levels preserved in bubbles in the ice ( Loulergue et al., 2008 ; Lüthi et al., 2008 ). The CO 2 and methane curves illustrated in Figure 6 show that the modern levels of these gases are unprecedented in the last 800 millennia.

Concentrations of carbon dioxide (CO 2 ) and methane (CH 4 ) over the last 800,000 years (eight glacial cycles) from East Antarctic ice cores. Data from Loulergue et al. (2008) and Lüthi et al. (2008) . The current concentrations of CO 2 and CH 4 are also shown ( Forster et al., 2007 ).

Globally, CO 2 concentrations have varied between 180 and 190 parts per million per volume (ppmv) during glacial (cold) periods and between 270 and 290 ppmv during interglacial (warm) periods. However, since the onset of the Industrial Revolution, when fossil fuel use (chiefly coal and oil) began to burgeon, CO 2 concentration has increased about 38% over the natural interglacial levels ( Forster et al., 2007 ). Between 1975 and 2005, CO 2 emissions increased 70%, and between 1999 and 2005 global emissions increased 3% per year ( Marland, Boden, & Andres, 2006 ). As of this writing, the CO 2 concentration in the atmosphere is 391 ppmv (Mauna Loa CO 2 annual mean data from the National Oceanic and Atmospheric Administration, 2010 ), a level not seen at any time in 800,000 years. Climatologists have identified no natural forcers that could account for this rapid and previously unseen rise in CO 2 .

Methane raises temperature even more than CO 2 , and the amount of methane in the atmosphere, like that of CO 2 , is also at a level not seen in 800 millennia. Two thirds of current emissions of methane are by-products of human activity, things like the production of oil and natural gas, deforestation, decomposition of garbage and sewage, and raising farm animals.

Many people find it difficult to believe that human activity can affect a system as large as Earth's climate. After all, we are so tiny compared to the planet. But every day we tiny human beings drive cars; watch television; turn on lamps; heat or cool our houses and offices; eat food transported to us by planes, ships, and trucks; clear or burn forests; and behave in countless other ways that directly or indirectly release greenhouse gases into the air. Together, we humans emitted eight billion metric tons of carbon to our planet's atmosphere in 2007 alone ( Boden, Marland, & Andres, 2009 ). (CO 2 weighs 3.66 times more than carbon; that means we released 29.3 billion metric tons of CO 2 .) The evidence is overwhelming that human activity is responsible for the rise in CO 2 , methane, and other greenhouse gas levels, and that the increase in these gases is fueling the rise in mean global temperature.

A global temperature rise of a few degrees may not seem such a bad thing, especially to people living in harsh, cold climates. But global warming does not mean merely that we will trade parkas for T-shirts or turn up the air conditioning. A warming planet is a changing planet, and the changes will have profound consequences for all species that live on it, including humans. Those changes are not just something our children and grandchildren will have to deal with in the future; they are taking place now, and are affecting millions of people.

EFFECTS OF GLOBAL WARMING

One effect of global warming that everyone has heard about is a rise in sea levels. About half of this rise is due to thermal expansion: Ocean temperatures are rising, and as water warms it expands. Put a nearly full cup of water in a microwave and heat it, and the water will spill over the cup.

In addition to thermal expansion, the oceans are rising because ice is melting, and most of that water inevitably finds its way to the sea. So far, most of that water has come from mountain glaciers and ice caps ( Meier et al., 2007 ). As global temperatures increase, sea level rise will mainly reflect polar ice melt. So far, ocean rise has been measured in millimeters, but there is enough water in the Greenland ice sheet alone to raise sea levels by about 7 m, West Antarctica over 5 m, and East Antarctica about 50 m ( Lemke et al., 2007 ). If the Earth were to lose just 8% of its ice, the consequences for some coastal regions would be dramatic. The lower part of the Florida peninsula and much of Louisiana, including New Orleans, would be submerged, and low-lying cities, including London, New York, and Shanghai, would be endangered (to see the effects of various magnitudes of sea level rise in the San Francisco Bay area, go to http://cascade.wr.usgs.gov/data/Task2b-SFBay/data.shtm ).

Low-lying continental countries such as the Netherlands and much of Bangladesh already find themselves battling flooding more than ever before. Many small island nations in the western Pacific (e.g., Vanuatu) are facing imminent destruction as they are gradually overrun by the rising ocean. Indonesia is an island nation, and many of its 17,000 islands are just above sea level. At the 2007 United Nations Climate Change Conference in Bali, Indonesian environmental minister Rachmat Witoelar stated that 2,000 of his country's islands could be lost to sea level rise by 2030. At current rates of sea level rise, another island nation, the Republic of Maldives, will become uninhabitable by the end of the century ( http://unfcc.int/resource/docs/napa/mdv01.pdf ). In 2008, the president of that country, Mohamed Nasheed, announced that he was contemplating moving his people to India, Sri Lanka, and Australia ( Schmidle, 2009 ). One of the major effects of continued sea level rise will be the displacement of millions of people. Where millions of climate refugees will find welcome is unclear. The migration of large numbers of people to new territories with different languages and cultures will be disruptive, to say the least.

In addition to the danger of inundation, rising sea levels bring salt water into rivers, spoil drinking wells, and turn fertile farmland into useless fields of salty soil. These effects of global warming are occurring now in places like the lowlands of Bangladesh ( Church et al., 2001 ).

People on dry land need the fresh water that is running into the sea. In the spring, melting ice from mountain glaciers, ice caps, and snowfields furnish wells and rivers that provide fresh water for drinking, agriculture, and hydroelectric power. For example, in the dry season, people in large areas of India, Nepal, and southern China depend on rivers fed by Himalayan glaciers. The retreat of these glaciers threatens the water supply of millions of people in this part of the world. Peru relies on hydroelectric power for 80% of its energy ( Vergara et al., 2007 ), a significant portion of which comes from mountain streams that are fed by mountain glaciers and ice fields. In Tanzania, the loss of Mount Kilimanjaro's fabled ice cover would likely have a negative impact on tourism, which is the country's primary source of foreign currency. The glaciers and snow packs in the Rocky Mountains are essential for farming in California, one of the world's most productive agricultural areas.

Global warming is expanding arid areas of the Earth. Warming at the equator drives a climate system called the Hadley Cell. Warm, moist air rises from the equator, loses its moisture through rainfall, moves north and south, and then falls to the Earth at 30° north and south latitude, creating deserts and arid regions. There is evidence that over the last 20 years the Hadley Cell has expanded north and south by about 2° latitude, which may broaden the desert zones ( Seidel, Fu, Randel, & Reichler, 2008 ; Seidel & Randel, 2007 ). If so, droughts may become more persistent in the American Southwest, the Mediterranean, Australia, South America, and Africa.

Global warming can also have effects that seem paradoxical. Continued warming may change ocean currents that now bring warm water to the North Atlantic region, giving it a temperate climate. If this happens, Europe could experience a cooling even as other areas of the world become warmer.

Accelerating Change

It is difficult to assess the full effects of global warming, and harder still to predict future effects. Climate predictions are made with computer models, but these models have assumed a slow, steady rate of change. Our best models predict a temperature rise in this century of between 2.4° and 4.5° C (4.3° and 8.1° F), with an average of about 3° C (5.4° F; Meehl et al., 2007 ; Figure 1 ). But these models assume a linear rise in temperature. Increasingly, computer models have underestimated the trends because, in fact, the rate of global temperature rise is accelerating. The average rise in global temperature was 0.11° F per decade over the last century ( National Oceanic and Atmospheric Administration, 2009 ). Since the late 1970s, however, this rate has increased to 0.29° F per decade, and 11 of the warmest years on record have occurred in the last 12 years. May, 2010, was the 303rd consecutive month with a global temperature warmer than its 20th-century average ( National Oceanic and Atmospheric Administration, 2010 ).

The acceleration of global temperature is reflected in increases in the rate of ice melt. From 1963 to 1978, the rate of ice loss on Quelccaya was about 6 m per year. From 1991 to 2006, it averaged 60 m per year, 10 times faster than the initial rate ( Thompson et al., 2006 ). A recent paper by Matsuo and Heki (2010) reports uneven ice loss from the high Asian ice fields, as measured by the Gravity Recovery and Climate Experiment satellite observations between 2003 and 2009. Ice retreat in the Himalayas slowed slightly during this period, and loss in the mountains to the northwest increased markedly over the last few years. Nevertheless, the average rate of ice melt in the region was twice the rate of four decades before. In the last decade, many of the glaciers that drain Greenland and Antarctica have accelerated their discharge into the world's oceans from 20% to 100% ( Lemke et al., 2007 ).

Increasing rates of ice melt should mean an increasing rate of sea level rise, and this is in fact the case. Over most of the 20th century, sea level rose about 2 mm per year. Since 1990, the rate has been about 3 mm per year.

So, not only is Earth's temperature rising, but the rate of this change is accelerating. This means that our future may not be a steady, gradual change in the world's climate, but an abrupt and devastating deterioration from which we cannot recover.

Abrupt Climate Change Possible

We know that very rapid change in climate is possible because it has occurred in the past. One of the most remarkable examples was a sudden cold, wet event that occurred about 5,200 years ago, and left its mark in many paleoclimate records around the world.

The most famous evidence of this abrupt weather change comes from Otzi, the “Tyrolean ice man” whose remarkably preserved body was discovered in the Eastern Alps in 1991 after it was exposed by a melting glacier. Forensic evidence suggests that Otzi was shot in the back with an arrow, escaped his enemies, then sat down behind a boulder and bled to death. We know that within days of Otzi's dying there must have been a climate event large enough to entomb him in snow; otherwise, his body would have decayed or been eaten by scavengers. Radiocarbon dating of Otzi's remains revealed that he died around 5,200 years ago ( Baroni & Orombelli, 1996 ).

The event that preserved Otzi could have been local, but other evidence points to a global event of abrupt cooling. Around the world organic material is being exposed for the first time in 5,200 years as glaciers recede. In 2002, when we studied the Quelccaya ice cap in southern Peru, we found a perfectly preserved wetland plant. It was identified as Distichia muscoides , which today grows in the valleys below the ice cap. Our specimen was radiocarbon dated at 5,200 years before present ( Thompson et al., 2006 ). As the glacier continues to retreat, more plants have been collected and radiocarbon dated, almost all of which confirm the original findings ( Buffen, Thompson, Mosley-Thompson, & Huh, 2009 ).

Another record of this event comes from the ice fields on Mount Kilimanjaro. The ice dating back 5,200 years shows a very intense, very sudden decrease in the concentration of heavy oxygen atoms, or isotopes, in the water molecules that compose the ice ( Thompson et al., 2002 ). Such a decrease is indicative of colder temperatures, more intense snowfall, or both.

The Soreq Cave in Israel contains speleothems that have produced continuous climate records spanning tens of thousands of years. The record shows that an abrupt cooling also occurred in the Middle East about 5,200 years ago, and that it was the most extreme climatic event in the last 13,000 years ( Bar-Matthews, Ayalon, Kaufman, & Wasserburg, 1999 ).

One way that rapid climate change can occur is through positive feedback. In the physical sciences, positive feedback means that an event has an effect which, in turn, produces more of the initial event. The best way to understand this phenomenon as it relates to climate change is through some very plausible examples:

Higher global temperatures mean dryer forests in some areas, which means more forest fires, which means more CO 2 and ash in the air, which raises global temperature, which means more forest fires, which means …

Higher global temperatures mean melting ice, which exposes darker areas (dirt, rock, water) that reflect less solar energy than ice, which means higher global temperatures, which means more melting ice, which means …

Higher global temperatures mean tundra permafrost melts, releasing CO 2 and methane from rotted organic material, which means higher global temperature, which means more permafrost melting, which means …

Positive feedback increases the rate of change. Eventually a tipping point may be reached, after which it could be impossible to restore normal conditions. Think of a very large boulder rolling down a hill: When it first starts to move, we might stop it by pushing against it or wedging chocks under it or building a barrier, but once it has reached a certain velocity, there is no stopping it. We do not know if there is a tipping point for global warming, but the possibility cannot be dismissed, and it has ominous implications. Global warming is a very, very large boulder.

Even if there is no tipping point (or we manage to avoid it), the acceleration of warming means serious trouble. In fact, if we stopped emitting greenhouse gases into the atmosphere tomorrow, temperatures would continue to rise for 20 to 30 years because of what is already in the atmosphere. Once methane is injected into the troposphere, it remains for about 8 to 12 years ( Prinn et al., 1987 ). Carbon dioxide has a much longer residence: 70 to 120 years. Twenty percent of the CO 2 being emitted today will still affect the earth's climate 1,000 years from now ( Archer & Brovkin, 2008 ).

If, as predicted, global temperature rises another 3° C (5.4° F) by the end of the century, the earth will be warmer than it has been in about 3 million years ( Dowsett et al., 1994 ; Rahmstorf, 2007 ). Oceans were then about 25 m higher than they are today. We are already seeing important effects from global warming; the effects of another 3° C (5.4° F) increase are hard to predict. However, such a drastic change would, at the very least, put severe pressure on civilization as we know it.

OUR OPTIONS

Global warming is here and is already affecting our climate, so prevention is no longer an option. Three options remain for dealing with the crisis: mitigate, adapt, and suffer.

Mitigation is proactive, and in the case of anthropogenic climate change it involves doing things to reduce the pace and magnitude of the changes by altering the underlying causes. The obvious, and most hotly debated, remedies include those that reduce the volume of greenhouse gas emissions, especially CO 2 and methane. Examples include not only using compact fluorescent lightbulbs, adding insulation to our homes, and driving less, but societal changes such as shutting down coal-fired power plants, establishing a federal carbon tax (as was recently recommended by the National Academy of Sciences), and substantially raising minimum mileage standards on cars ( National Research Council, 2010 ). Another approach to mitigation that has received widespread attention recently is to enhance the natural carbon sinks (storage systems) through expansion of forests. Some have suggested various geo-engineering procedures (e.g., Govindasamy & Caldeira, 2000 ; Wigley, 2006 ). One example is burying carbon in the ocean or under land surfaces ( Brewer, Friederich, Peltzer, & Orr, 1999 ). Geo-engineering ideas are intriguing, but some are considered radical and may lead to unintended negative consequences ( Parkinson, 2010 ).

Adaptation is reactive. It involves reducing the potential adverse impacts resulting from the by-products of climate change. This might include constructing sea barriers such as dikes and tidal barriers (similar to those on the Thames River in London and in New Orleans), relocating coastal towns and cities inland, changing agricultural practices to counteract shifting weather patterns, and strengthening human and animal immunity to climate-related diseases.

Our third option, suffering, means enduring the adverse impacts that cannot be staved off by mitigation or adaptation. Everyone will be affected by global warming, but those with the fewest resources for adapting will suffer most. It is a cruel irony that so many of these people live in or near ecologically sensitive areas, such as grasslands (Outer Mongolia), dry lands (Sudan and Ethiopia), mountain glaciers (the Quechua of the Peruvian Andes), and coastal lowlands (Bangledesh and the South Sea island region). Humans will not be the only species to suffer.

Clearly mitigation is our best option, but so far most societies around the world, including the United States and the other largest emitters of greenhouse gases, have done little more than talk about the importance of mitigation. Many Americans do not even accept the reality of global warming. The fossil fuel industry has spent millions of dollars on a disinformation campaign to delude the public about the threat, and the campaign has been amazingly successful. (This effort is reminiscent of the tobacco industry's effort to convince Americans that smoking poses no serious health hazards.) As the evidence for human-caused climate change has increased, the number of Americans who believe it has decreased. The latest Pew Research Center (2010) poll in October, 2009, shows that only 57% of Americans believe global warming is real, down from 71% in April, 2008.

There are currently no technological quick fixes for global warming. Our only hope is to change our behavior in ways that significantly slow the rate of global warming, thereby giving the engineers time to devise, develop, and deploy technological solutions where possible. Unless large numbers of people take appropriate steps, including supporting governmental regulations aimed at reducing greenhouse gas emissions, our only options will be adaptation and suffering. And the longer we delay, the more unpleasant the adaptations and the greater the suffering will be.

Sooner or later, we will all deal with global warming. The only question is how much we will mitigate, adapt, and suffer.

Acknowledgments

This paper is based on the Presidential Scholar's Address given at the 35th annual meeting of the Association for Behavior Analysis International, Phoenix, Arizona. I am grateful to Bill Heward for inviting me to give the address. I thank Mary Davis for her help editing the text and figures. I wish to thank all the field and laboratory team members from the Byrd Polar Research Center who have worked so diligently over the years. I am especially indebted to the hard work of our current research team: Ellen Mosley-Thompson, Henry Brecher, Mary Davis, Paolo Gabrielli, Ping-Nan Lin, Matt Makou, Victor Zagorodnov, and all of our graduate students. Funding for our research over the years has been provided by the National Science Foundation's Paleoclimate Program, the National Oceanic and Atmospheric Administration's Paleoclimatology and Polar Programs, the National Aeronautic and Space Administration, Gary Comer Foundation, and The Ohio State University's Climate, Water and Carbon Program. This is Byrd Polar Research Center Publication 1402.

• Archer D, Brovkin V. Millennial atmospheric lifetime of anthropogenic CO 2 . Climatic Change. 2008; 90 ((3)):283–297. [ Google Scholar ]
• Arendt A.A, Echelmeyer K.A, Harrison W.D, Lingle C.S, Valentine V.B. Rapid wastage of Alaska glaciers and their contribution to rising sea level. Science. 2002; 297 :382–386. [ PubMed ] [ Google Scholar ]
• Bar-Matthews M, Ayalon A, Kaufman A, Wasserburg G.J. The eastern Mediterranean paleoclimate as a reflection of regional events: Soreq Cave, Israel. Earth and Planetary Science Letters. 1999; 166 :85–95. [ Google Scholar ]
• Baroni C, Orombelli G. The Alpine “Iceman” and Holocene climatic change. Quaternary Research. 1996; 46 :78–83. [ Google Scholar ]
• Boden T.A, Marland G, Andres R.J. Global, regional, and national fossil-fuel CO 2 emissions. Oak Ridge, TN: Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy; 2009. Retrieved from http://cdiac.ornl.gov/trends/emis/tre_glob.html . [ Google Scholar ]
• Bradley R.S, Keimig F.T, Diaz H.F, Hardy D.R. Recent changes in the freezing level heights in the tropics with implications for the deglacierization of high mountain regions. Geophysical Research Letters. 2009; 36 :L17701. [ Google Scholar ]
• Bradley R.S, Vuille M, Diaz H.F, Vergara W. Threats to water supplies in the tropical Andes. Science. 2006; 312 :1755–1756. [ PubMed ] [ Google Scholar ]
• Brewer P.G, Friederich G, Peltzer E.T, Orr F.M., Jr Direct experiments on the ocean disposal of fossil fuel CO 2 . Science. 1999; 284 :943–945. [ PubMed ] [ Google Scholar ]
• Briffa K.R, Jones P.D, Schweingruber F.H, Shiyatov S.G, Cook E.R. Unusual twentieth-century summer warmth in a 1,000-year temperature record from Siberia. Nature. 2002; 376 :156–159. [ Google Scholar ]
• Buffen A.M, Thompson L.G, Mosley-Thompson E, Huh K.-I. Recently exposed vegetation reveals Holocene changes in the extent of the Quelccaya ice cap, Peru. Quaternary Research. 2009; 72 :157–163. [ Google Scholar ]
• Chappellaz J, Blunier T, Kints S, Dällenbach A, Barnola J-M, Schwander J, et al. Changes in the atmospheric CH 4 gradient between Greenland and Antarctica during the Holocene. Journal of Geophysical Research. 1997; 102 :15,987–15,997. [ Google Scholar ]
• Church J.A, Gregory J.M, Huybrechts P, Kuhn M, Lambeck K, Nhuan M.T, et al. Climate change 2001. The scientific basis. Contributions of Working Group I to the 3rd assessment of the IPCC. Cambridge, UK: Cambridge University Press; 2001. Changes in sea level. [ Google Scholar ]
• Climate change and the integrity of science. 2010. Retreived from http://www.pacinst.org/climate/climate_statement.pdf . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Crowley T.J, Lowery T.S. How warm was the medieval warm period? AMBIO: A Journal of the Human Environment. 2000; 29 :51–54. [ Google Scholar ]
• Das S.B, Joughin I, Behn M.D, Howat I.M, King M.A, Lizarralde D, et al. Fracture propagation to the base of the Greenland ice sheet during supraglacial lake drainage. Science. 2008; 320 :778–781. [ PubMed ] [ Google Scholar ]
• Dowsett H.J, Thompson R, Barron J, Cronin T, Fleming F, Ishman S, et al. Joint investigations of the middle Pleistocene climate 1: PRISM paleoenvironmental reconstructions. Global and Planetary Change. 1994; 9 :169–195. [ Google Scholar ]
• Esper J, Cook E.R, Schweingruber F.H. Low-frequency signals in long tree-ring chronologies for reconstructing past temperature variability. Science. 2002; 295 :2250–2253. [ PubMed ] [ Google Scholar ]
• Forster P, Ramaswamy V, Arttaxo P, Berntsen T, Betts R, Fahey D.W, et al. Climate change 2007: The physical science basis. Contributions of Working Group I to the 4th assessment of the IPCC. Cambridge, UK: Cambridge University Press; 2007. Changes in atmospheric constituents and in radiative forcing. [ Google Scholar ]
• Govindasamy B, Caldeira K. Geoengineering Earth's radiation balance to mitigate CO 2 -induced climate change. Geophysical Research Letters. 2000; 27 :2141–2144. [ Google Scholar ]
• Hall M.H.P, Fagre D.B. Modeled climate-induced glacier change in Glacier National Park, 1850–2100. BioScience. 2003; 53 :131–140. [ Google Scholar ]
• Jones P.D, Briffa K.R, Barnett T.P, Tett S.F.B. High-resolution paleoclimate records for the last millennium: Interpretation, integration and comparison with general circulation model control-run temperatures. The Holocene. 1998; 8 :455–471. [ Google Scholar ]
• Kehrwald N.M, Thompson L.G, Yao T, Mosley-Thompson E, Schotterer U, Alfimov V, et al. Mass loss on Himalayan glacier endangers water resources. Geophysical Research Letters. 2008; 35 :L22503. [ Google Scholar ]
• Lemke P, Ren J, Alley R.B, Carrasco J, Flato G, Fujii Y, et al. Observations: Changes in snow, ice and frozen ground in climate change 2007: The physical science basis. Contributions of Working Group I to the 4th assessment of the IPCC. Cambridge, UK: Cambridge University Press; 2007. [ Google Scholar ]
• Loulergue L, Schilt A, Spahni R, Masson-Delmotte V, Blunier T, Lemieux B, et al. Orbital and millennial-scale features of atmospheric CH 4 over the past 800,000 years. Nature. 2008; 453 :383–386. [ PubMed ] [ Google Scholar ]
• Lüthi D, Le Floch M, Bereiter B, Blunier T, Barnola J.-M, Siegenthaler U, et al. High-resolution carbon dioxide concentration record 650,000–800,000 years before present. Nature. 2008; 453 :379–382. [ PubMed ] [ Google Scholar ]
• Mann M.E, Bradley R.S, Hughes M.K. Northern hemisphere temperatures during the past millennium: Inferences, uncertainties, and limitations. Geophysical Research Letters. 1999; 26 :759–762. [ Google Scholar ]
• Mann M.E, Jones P.D. Global surface temperatures over the past two millennia. Geophysical Research Letters. 2003; 30 :1820. [ Google Scholar ]
• Mann M.E, Zhang Z, Rutherford S, Bradley R.S, Hughes M.K, Shindell D, et al. Global signatures and dynamical origins of the little ice age and medieval climate anomaly. Science. 2009; 326 :1256–1260. [ PubMed ] [ Google Scholar ]
• Marland G, Boden T.A, Andres R. Trends: A compendium of data on global change. Oak Ridge, TN: Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy; 2006. Global, regional, and national CO 2 emissions. Retrieved from http://cdiac.esd.ornl.gov/trends/emis/tre_glob.htm . [ Google Scholar ]
• Matsuo K, Heki K. Time-variable ice loss in Asian high mountains from satellite gravimetry. Earth and Planetary Science Letters. 2010; 290 :30–36. [ Google Scholar ]
• Meehl G.A, Arblaster J.M, Tebaldi C. Contributions of natural and anthropogenic forcing to changes in temperature extremes over the United States. Geophysical Research Letters. 2007; 34 :L19709. [ Google Scholar ]
• Meehl G.A, Stocker T.F, Collins W.D, Friedlingstein P, Gaye A.T, Gregory J.M, et al. Climate change 2007: The physical science basis. Contributions of Working Group I to the 4th assessment of the IPCC. Cambridge, UK: Cambridge University Press; 2007. Global climate projections. [ Google Scholar ]
• Meier M.F, Dyurgerov M.B, Rick U.K, O'Neel S, Pfeffer W.T, et al. Glaciers dominate eustatic sea-level rise in the 21st century. Science. 2007; 317 :1064–1067. [ PubMed ] [ Google Scholar ]
• Mercer J.H. West Antarctic ice sheet and CO 2 greenhouse effect: A threat of disaster. Nature. 1978; 271 :321–325. [ Google Scholar ]
• Moberg A, Sonechkin D.M, Holmgren K, Datsenko N.M, Karlen W. Highly variable northern hemisphere temperatures reconstructed from low- and high-resolution proxy data. Nature. 2005; 433 :613–617. [ PubMed ] [ Google Scholar ]
• Molnia B.F. Late nineteenth to early twenty-first century behavior of Alaskan glaciers as indicators of changing regional climate. Global and Planetary Change. 2007; 56 :23–56. [ Google Scholar ]
• National Oceanic and Atmospheric Administration. State of the climate global analysis. 2009. Retrieved from http://www.ncdc.noaa.gov/sotc/?report=global&year=2009&month=13&submitted=Get+Report#trends .
• National Oceanic and Atmospheric Administration. May 2010 global state of the climate—Supplemental figures and information. 2010. Jun 15, Retrieved from http://www.noaanews.noaa.gov/stories2010/20100615_globalstats_sup.html .
• National Research Council. Surface temperature reconstructions for the last 2,000 years. Washington DC: National Academy of Sciences; 2006. [ Google Scholar ]
• National Research Council. Limiting the magnitude of future climate change: Report in brief. Washington, DC: National Academies Press; 2010. Retrieved from http://dels.nas.edu/resources/static-assets/materials-based-on-reports/reports-in-brief/Limiting_Report_Brief_final.pdf . [ Google Scholar ]
• Oerlemans J. Extracting a climate signal from 169 glacier records. Science. 2005; 308 :675–677. [ PubMed ] [ Google Scholar ]
• Parkinson C.L. Coming climate crisis? Consider the past, beware the big fix. Lanham, MD: Rowland & Littlefield; 2010. [ Google Scholar ]
• Perovich D.K, Richter-Menge J.A. Loss of sea ice in the Arctic. Annual Review of Marine Science. 2009; 1 :417–441. [ PubMed ] [ Google Scholar ]
• Pew Research Center. Fewer Americans see solid evidence of global warming. 2010. Retrieved from http://pewresearch.org/pubs/1386/cap-and-trade-global-warming-opinion .
• Prinn R, Cunnold D, Rasmussen R, Simmonds P, Alyea F, Crawford A, et al. Atmospheric trends in methylchloroform and the global average for the hydroxyl radical. Science. 1987; 238 :945–950. [ PubMed ] [ Google Scholar ]
• Rahmstorf S. A semi-empirical approach to projecting future sea-level rise. Science. 2007; 315 :368–370. [ PubMed ] [ Google Scholar ]
• Schmidle N. Wanted: A new home for my country. 2009. Retrieved from http://www.nytimes.com/2009/05/10/magazine/10MALDIVES-t.html?_r=3&partner=rss&emc=rss&pagewanted=all .
• Seidel D.J, Fu Q, Randel W.J, Reichler T.J. Widening of the tropical belt in a changing climate. Nature Geoscience. 2008; 1 :21–24. [ Google Scholar ]
• Seidel D.J, Randel W.J. Recent widening of the tropical belt: Evidence from tropopause observations. Journal of Geophysical Research. 2007; 112 :D20113. [ Google Scholar ]
• Thompson L.G. Ice core evidence for climate change in the tropics: Implications for our future. Quaternary Science Reviews. 2000; 19 :19–35. [ Google Scholar ]
• Thompson L.G, Brecher H.H, Mosley-Thompson E, Hardy D.R, Mark B.G. Glacier loss on Kilimanjaro continues unabated. Proceedings of the National Academy of Sciences. 2009; 106 :19,770–19,775. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Thompson L.G, Davis M.E, Mosley-Thompson E. Glacial records of global climate: A 1500-year tropical ice core record of climate. Human Ecology. 1994; 22 :83–95. [ Google Scholar ]
• Thompson L.G, Mosley-Thompson E, Brecher H.H, Davis M.E, Leon B, Les D, et al. Evidence of abrupt tropical climate change: Past and present. Proceedings of the National Academy of Sciences. 2006; 103 :10,536–10,543. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Thompson L.G, Mosley-Thompson E, Davis M.E, Bolzan J.F, Dai J, Klein L, et al. Glacial stage ice-core records from the subtropical Dunde ice cap, China. Annals of Glaciology. 1990; 14 :288–297. [ Google Scholar ]
• Thompson L.G, Mosley-Thompson E, Davis M.E, Henderson K.A, Brecher H.H, Zagorodnov V.S, et al. Kilimanjaro ice core records: Evidence of Holocene climate change in tropical Africa. Science. 2002; 289 :589–593. [ PubMed ] [ Google Scholar ]
• Thompson L.G, Yao T, Davis M.E, Henderson K.A, Mosley-Thompson E, Lin P.-N, et al. Tropical climate instability: The last glacial cycle from a Qinghai-Tibetan ice core. Science. 1997; 276 :1821–1825. [ Google Scholar ]
• Thompson L.G, Yao T, Mosley-Thompson E, Davis M.E, Henderson K.A, Lin P.-N. A high-resolution millennial record of the South Asian monsoon from Himalayan ice cores. Science. 2000; 289 :1916–1919. [ PubMed ] [ Google Scholar ]
• Vergara W, Deeb A.M, Valencia A.M, Bradley R.S, Francou B, Zarzar A, et al. Economic impacts of rapid glacier retreat in the Andes. EOS. 2007; 88 :261–268. [ Google Scholar ]
• Vince G. Dams for Patagonia. Newsfocus. Science. 2010; 329 :382–385. [ PubMed ] [ Google Scholar ]
• Wigley T.M.L. A combined mitigation/geoengineering approach to climate stabilization. Science. 2006; 314 :452–454. [ PubMed ] [ Google Scholar ]
• Yao T, Pu J, Lu A, Wang Y, Yu W. Recent glacial retreat and its impact on hydrological processes on the Tibetan Plateau, China and surrounding regions. Arctic and Alpine Research. 2007; 39 :642–650. [ Google Scholar ]
• Zwally H.J, Abdalati W, Herring T, Larson K, Saba J, Steffen K. Surface melt-induced acceleration of Greenland ice-sheet flow. Science. 2002; 297 :218–222. [ PubMed ] [ Google Scholar ]

A theoretical model of climate anxiety and coping

• Perspective
• Open access
• Published: 11 August 2024
• Volume 4 , article number  94 , ( 2024 )

Cite this article

You have full access to this open access article

• Tara J. Crandon   ORCID: orcid.org/0000-0002-5915-7040 1 , 2 ,
• James G. Scott 1 , 3 , 4 , 5 ,
• Fiona J. Charlson 2 , 5 , 6 &
• Hannah J. Thomas 1 , 2 , 5  

238 Accesses

2 Altmetric

Explore all metrics

Research on climate anxiety is rapidly growing, with ongoing exploration of population prevalence, contributing factors, and mitigation strategies that transform anxiety into helpful action. What remains unclear is whether and how to delineate climate anxiety from mental ill health. A limited conceptualization of climate anxiety restricts efforts to identify and support those adversely affected. This paper draws on psychological and existential theories to propose a theoretical model of climate anxiety and coping, extending previous conceptualizations. The model theorizes that climate change evokes an existential conflict that manifests affectively as climate anxiety (and other emotional experiences), wherein cognitive and behavioral coping processes are activated. These processes fall on a continuum of adaptivity, depending on functional impact. Responses might range from meaningful engagement with activities that address climate change to maladaptive strategies that negatively impact personal, social, and occupational functioning. Applications of this model in research and practice are proposed.

Similar content being viewed by others

Anxiety and climate change: a validation of the Climate Anxiety Scale in a German-speaking quota sample and an investigation of psychological correlates

Validation of the Hogg Climate Anxiety Scale

Informing future directions for climate anxiety interventions: a mixed-method study of professional perspectives.

Avoid common mistakes on your manuscript.

1 Introduction

Climate change has negative implications for population mental health and well-being [ 1 ]. Direct exposure to extreme weather events is associated with trauma-related disorders, as well as elevated anxiety, depression, substance use, grief, and suicidal ideation [ 2 , 3 ]. In addition, temperature fluctuation is associated with anger, stress, increased fatigue, major depressive disorder, self-harm, and suicide attempts [ 3 , 4 ]. Climate change may also indirectly cause significant social, economic, and environmental risk factors for mental illness, including forced migration, food scarcity, and conflict between and within nations [ 2 , 5 , 6 , 7 ]. Irrespective of degree of exposure to such stressors, psychological and emotional responses to anthropogenic climate change are well documented [ 7 , 8 ]. Simple awareness of the climate crisis can evoke ‘climate anxiety’ [ 6 ].

The definition of climate anxiety varies across different fields of research; however, it is commonly understood as a fear of environmental change and its impacts [ 9 ], as well as a cognizance that the ecological foundations of the planet are in the process of collapse [ 9 ]. Reports of those with lived experience of climate anxiety has brought attention to this phenomenon in the media, across research and in healthcare [ 10 , 11 ]. Despite growing efforts to expand scientific knowledge on climate anxiety, a conceptual psychological understanding of how one copes is yet to be developed [ 12 , 13 ]. This restricts the ability to accurately identify and provide meaningful support to those affected [ 12 , 14 ]. This review aims to:

Provide an overview of the historical background, theories and measures of climate anxiety, with consideration of broader theoretical perspectives on anxiety and coping,

Respond to recommendations made by systematic reviews to refine the definition of climate anxiety and clarify its adaptivity or helpfulness,

Propose an integrated theoretical model to conceptualize the experience of climate anxiety and how people may cope with these feelings; and,

Draw on the model to discuss considerations and approaches that may be relevant for researchers and practitioners working with or exploring climate anxiety (see Fig.  1 for a summary).

A summary of considerations and approaches that could be relevant for research and practice, as proposed by the theoretical model of climate anxiety and coping

1.1 A history of research on climate anxiety

1.1.1 eco-anxiety.

While climate anxiety is a subject that has grown in interest more recently, research investigating attitudes towards global environmental issues (e.g., ozone depletion, global warming) dates back to the 1980’s. Between 1982 and 1986, 12 European studies showed that on average, 34–38% of the public were concerned or worried about changes to the climate as a result of carbon dioxide emissions [ 15 ]. In a 1994 global survey of environmental attitudes, the proportion of those viewing global warming as a “serious threat” varied by country, from 20–40% in countries such as Nigeria, India, and the United States to more than 70% in Brazil, Germany and Portugal [ 16 ].

Research on mental health and the environment proliferated from the 1990’s to the early 2000’s, with evidence suggesting that contact with the natural world is beneficial to well-being and can enhance feelings of relaxation, peacefulness and tranquility [ 17 ]. The term ‘Eco-anxiety’ subsequently emerged in print media feature articles, predominantly in political discourse or in interviews with ‘eco-psychologists’. Initially, there was a focus on nature-based therapeutic treatment approaches (e.g., wilderness therapy, green exercise) to support individuals reporting distress regarding environmental issues and disconnection from nature [ 18 , 19 , 20 , 21 ]. With increasing reports of distress over climate change in the news and in healthcare, research turned its attention to climate change and mental health symptoms, reporting associations between concern or distress about climate change with anxiety, stress, and impairments to daily living (e.g., interfering with work, school or relationships) [ 22 , 23 , 24 ]. The term eco-anxiety was soon adopted into the literature and has been characterized as a “chronic fear of environmental doom” [ 25 , 26 ].

1.1.2 Climate anxiety and its impacts on wellbeing

In recent years, the introduction of the term ‘climate anxiety’ has provided increased specificity in terminology. While climate anxiety continues to be used interchangeably with eco-anxiety in the literature, eco-anxiety is more broadly understood as encompassing emotional distress towards environmental issues and is not specific to climate change [ 9 ]. In contrast, climate anxiety may involve anxiety, dread, and despair specifically in response to climate change-related issues, including loss of natural places, ecological collapse and anticipated future harm due to climate change-related stressors (e.g., extreme weather disasters, forced migration). Awareness of climate change is sufficient to elicit climate anxiety [ 21 , 24 , 27 ], where worries may be about the implications for oneself, others, future generations, animals, the environment, or the future state of the world [ 28 ]. Awareness may occur through direct exposure (e.g., experiencing extreme weather-related events or gradual environmental changes over time) as well as indirect exposure (e.g., news, media or education around climate change-related issues) [ 29 ]. Empirical research on eco anxiety and climate anxiety continues to grow rapidly, and several systematic reviews have been conducted on the topic (see Table  1 for a summary). Collectively, these reviews highlight the need to strengthen the conceptualization of climate anxiety and to explore avenues of support for those who are affected by it [ 12 , 13 , 30 , 31 , 32 ]. In responding to these recommendations, the adaptivity (i.e., helpfulness) of climate anxiety must be clarified, which we attempt to address within this paper.

A growing consensus is that for some, climate anxiety can become maladaptive and interfere with functioning [ 24 , 33 , 34 , 35 , 36 ]. Based on experiences with clients in clinical practice, Hickman [ 10 ] proposes different levels of severity from mild (some distress that responds to distraction, reassurance, and individual action, e.g., altering diet and recycling used materials) to severe (changes in cognition such as intrusive thinking, terror, no trust in solutions or experts, and an inability to regulate emotional responses). This severe level of climate anxiety appears to overlap in some respects with clinical anxiety disorders, particularly in relation to a significant level of distress leading to impairments in personal, social, and occupational functioning [ 37 , 38 ]. For some, climate anxiety may even exacerbate pre-existing mental health problems such as generalized anxiety [ 39 ]. However, fundamental to clinical anxiety disorders is that the response is often out of proportion to the threat or a maladaptive response to uncertainty, whereas climate change carries complex, real and devastating impacts on human life.

1.1.3 Considering the function of climate anxiety

Given the negative implications of climate change, climate anxiety could serve a constructive purpose. Emotions are psychological states accompanied by cognitive processes, behavioral responses, and patterns of neurophysiological changes [ 40 , 41 ]. According to Basic Emotion Theory, emotions enable one to respond to threats and opportunities in the environment [ 42 ]. Anxiety particularly functions as a warning sign when survival or well-being is threatened, triggering adaptive responses to eliminate the threat [ 43 ]. Autonomic, neurobiological, cognitive, and behavioral patterns may be activated as a means of escaping or managing the danger (e.g., fight, flight, freeze responses) [ 24 , 34 ]. Common catalysts for anxiety are situations i) that are open to different interpretations (ambiguous situations), ii) in which there is no prior experience to draw from (novel situations) or iii) where it is unclear what may transpire (unpredictable situations). These conditions, which lead to high uncertainty and a level of uncontrollability, may exacerbate perceptions of threat [ 44 , 45 ]. In turn, anxiety and the accompanying cognitive and behavioral processes are enacted to try to manage the ongoing threat and the underlying uncertainty, when unable to eliminate the threat completely [ 45 ].

Climate anxiety could therefore be, to an extent, necessary to motivate the transition to a sustainable future. As such, it is important to avoid the use of terminology that unhelpfully pathologizes the experience. It is also worth noting that many individuals may respond to feelings of climate anxiety in ways that are constructive (e.g., activism, research, helping others) [ 29 , 46 ], with some studies showing associations between climate anxiety and pro-environmental behavior, environmental activism and increased engagement with politics [ 47 , 48 ]. Emerging evidence also suggests that climate anxiety can potentially lead to a gain in functioning (such as being more motivated or more socially connected) when accompanied by more balanced (i.e., more hopeful or positive) re-appraisal styles of thinking [ 49 ]. Climate anxiety may therefore be the alarm bell signaled when one appraises climate change as catastrophic and uncertain. This response prompts an individual to act in order to reduce or manage the threat of climate change. It can thus be argued that it is how someone copes with climate change and anxiety, rather than the emotion itself, that varies in adaptivity.

1.1.4 Considering coping with anxiety and other emotions

Coping is a cognitive and behavioral process activated to manage, tolerate, or reduce adversity or stress [ 50 , 51 ]. Several styles of coping are evidenced in the climate change literature, especially in work with young people [ 52 , 53 , 54 ]. Firstly, emotion-focused coping involves attempts to soothe, regulate, or remove emotions engendered by the climate crisis. Problem-focused coping relates to thinking about, talking about, and acting on climate mitigation. Meaning-focused coping involves a positive or more balanced re-appraisal of climate change through hope and trust. Distancing refers to the strategies one uses to move away from or distract oneself from climate change or their negative emotions toward it. Lastly, de-emphasizing the seriousness of climate change can range from apathy to skepticism, and even denial. Pihkala [ 55 ] synthesizes some of these ideas using a process model, which posits that eco anxiety and ecological grief activate a process of coping. Three dimensions of coping are described: action, grieving (and the processing of other emotions) and distancing. As individuals traverse these coping tasks over time, it is theorized that they may then enter a phase of living with the climate crisis. During this phase, coping becomes focused on action, emotional engagement (including grieving), and self-care (which can include distancing). Pihkala [ 55 ] suggests that difficulties in adjustment or coping could lead to anxiety, depression, and problematic avoidance. Indeed, we argue that coping strategies can become adaptive or maladaptive depending on the effect to one’s well-being and functioning.

While attempting to cope with climate anxiety, other emotional responses may understandably arise, which may require further use of coping resources. For example, learning about ineffectual government responses or how corporations contribute to emissions, may provoke anger, frustration, and feelings of betrayal [ 33 ]. Alternatively, ecological grief may be felt in response to physical ecological losses (e.g., species, environmental landscapes), loss of environmental and cultural knowledge, sense of place and home, and to security, stability, and even life [ 34 , 56 ]. This grief can become disenfranchised (prolonged and exacerbated) when invalidated by others through dismissal or minimization [ 26 , 27 , 29 ]. Failure to prevent these ecological losses due to collective inaction may prompt feelings of hopelessness or helplessness [ 39 ]. On the other hand, connecting with nature, others, or one’s values within the context of climate change, could lead to hope, connection, optimism, gratitude, or soliphilia (a feeling of love and sense of responsibility to protect the planet) [ 52 , 57 , 58 , 59 ]. These emotional states could overlap and intersect with climate anxiety, influencing the way one cognitively and/or behaviorally responds [ 60 ]. Evidence suggests anxiety combined with anger or hope, for example, may prompt pro-environmental behavior [ 35 , 60 ]. While it can be hypothesized that climate anxiety is an initial emotional response typically evoked by the threats of climate change, it is important to acknowledge that emotions are sensitive to situational factors such as context, cultural norms, personality, and cognitive attributions [ 41 , 60 , 61 , 62 ]. Therefore, as the context of the climate crisis continues to change, emotional states including climate anxiety may oscillate in salience, valence (pleasant to unpleasant), arousal (passive to activated), duration and intensity or severity [ 10 ]. This too will have flow-on effects for coping. Other emotional responses to climate change, which are difficult to disentangle from climate anxiety, should therefore be incorporated in measurement and research.

1.2 Considering existing measures of climate anxiety

Existing measures of climate anxiety provide compelling evidence of climate anxiety as a multi-dimensional construct. In particular, Clayton and Karazsia [ 24 ] developed the Climate Change Anxiety Scale, identifying two unique dimensions: cognitive-emotional impairment and functional impairment [ 24 , 63 , 64 ]. Hogg and others [ 65 ] also developed and validated a scale and identified four underlying constructs of climate anxiety: affective symptoms, rumination, behavioral symptoms and anxiety regarding one’s negative impact on the planet. Although both scales helpfully point to the presence of affective, cognitive, and behavioral experiences, each has some conceptual limitations. Firstly, there is a tendency in both scales to focus on impairment. While capturing the presence of potentially unhelpful responses, the scales do not capture the range and presence of adaptive responses. By overlooking the potential for an individual to respond or cope constructively, research and intervention efforts are somewhat limited. Furthermore, individuals who experience climate anxiety whose coping may fall on the adaptive end of the continuum may require different types of support, which could be detected if measures focus primarily on coping.

Another limitation to existing measures is that they reduce the cognitive dimension of climate anxiety to excessive preoccupation with climate change or one’s own distress. This potentially overemphasizes repetitive thinking /excessive worry and limits the role that specific cognitive appraisals may have in developing and maintaining climate anxiety. While some climate anxious individuals may excessively worry, others may de-emphasize the seriousness of the threats, filter scientific versus catastrophic information, or reappraise the climate crisis by drawing on hope and trust in mitigation efforts [ 18 , 66 ]. These cognitive patterns may in turn influence both emotions (e.g., anxiety, dread, fear) and associated behaviors (e.g., pro-environmental behavior, avoidance).

Despite limitations, existing measures of climate anxiety are a useful platform for which to develop and extend the forthcoming theoretical conceptualization. Based on the current state of the literature, we argue that climate anxiety is best considered as an expected and understandable emotional response that encompasses emotions (e.g., fear, anxiety, dread) and physiological symptoms typical of anxiety (e.g., feeling tense). Yet how an individual processes or copes with these emotions (cognitively or behaviorally) likely falls on a continuum of adaptivity depending on functional impact of the coping response [ 67 ]. When coping becomes highly maladaptive, this may be attributable to pre-existing or underlying symptoms of a diagnosable mental health disorder.

This conceptualization aligns with both suggestions that climate anxiety emotions should not be considered a mental health disorder, but also that coping can become clinically maladaptive and lead to functional impairment. Individuals should be supported to express their experiences of climate anxiety, which may help them to develop more helpful and adaptive coping strategies [ 18 , 46 ]. To further illustrate and expand on this conceptualization, broader psychological theories of anxiety will be drawn on to propose a theoretical model of climate anxiety and coping.

1.3 Considering broader theoretical perspectives

1.3.1 cognitive psychology.

Of the many cognitive theories that explain why or how an individual experiences anxiety, two will be applied to climate anxiety. Beck and Clark’s [ 68 ] information processing model proposes that the experience of anxiety is characterized by (1) an automatic detection of a threat (rapid, memory-based), (2) primal threat mode activation (e.g., escape/avoidance behavior or physiological arousal), and (3) conscious appraisal and reflective thinking (emotionally and/or logically reasoned) used to determine the best course of action to eliminate the threat [ 29 , 68 ]. Further, this conscious appraisal comprises thought content ( what the individual thinks) and thought process (the way the individual thinks). Applied to climate anxiety, an individual’s thoughts (content and process) may arise as a way to make sense of climate change and their feelings towards it, with the function of determining an appropriate course of action. For example, there is evidence that hopeful efficacy beliefs [ 69 ], meaning-focused thoughts and problem-focused thinking [ 70 ], are associated with environmental action and engagement. This, however, may not always be adaptive. Beck and Clark [ 68 ] suggest that dominance of the primal threat mode can lead to thinking that is unconstructive, excessive, or pathological. An individual experiencing climate anxiety may have less capacity for constructive thought when consistently under threat by adversity in life as well as climate-induced stressors (e.g., migration). For some, these thought processes may become unhelpfully habitual [ 71 ]. Furthermore, knowledge and experience stored in the long-term memory can influence how information is processed [ 72 ]. Adverse experiences can stimulate negative biases, which can lead to thinking biases and disordered cognition [ 73 ]. Experiencing a severe season of bushfires, for example, may lead to cognitive styles characterized by catastrophizing (e.g., “humanity is doomed”) or excessive preoccupation with past or anticipated losses. Conversely, beliefs and experiences held in long-term memory could motivate constructive thinking. One study, for example, found associations between habitual worrying about climate change (repetitive, automatic thinking) and pro-ecological worldviews, pro-environmental values, a ‘green’ identity, and pro-environmental behavior, suggesting a more constructive cognitive style [ 71 ].

A person’s beliefs about their own thoughts may also impact and maintain their anxiety [ 74 ]. According to metacognitive theory, one’s awareness, knowledge and perception of their thoughts can guide which coping strategies they select [ 74 ]. For example, an individual who believes they struggle to control their thoughts about climate change may unsuccessfully try to suppress their emotions or thoughts, leading to greater anxiety. For further consideration is whether the experience of climate anxiety is in general a direct and proportionate response to climate change related threats (e.g., ‘state anxiety’), or an extension of an individual’s tendency to experience anxiety or worry-based thinking styles (e.g., ‘trait anxiety’) [ 75 , 76 ]. While some research has found that climate anxiety is associated with generalized anxiety and trait pathological worry [ 71 ], another study found that approximately 60% of participants scoring highly on state climate anxiety were absent in high trait anxiety [ 76 ]. While further research is needed to explore the role of cognitive styles and processes in managing and processing climate anxiety (e.g., metacognitive beliefs, knowledge, tendency to use specific thinking styles), existing literature points to a range of adaptive to maladaptive cognitive responses [ 29 ]. It is important to note that while unhelpful cognitions may play a role in climate anxiety, concerns about climate change reflect a natural response to an existential threat. There is an emerging body of work that seeks to understand the existential nature of climate anxiety.

1.3.2 The existential perspective

Existentialists regard humans as reflective and meaning-making beings, who are expected to naturally confront and reconcile shared existential concerns considered core to human life [ 77 ]. Awareness of existential concerns can elicit apprehension, angst, anxiety, grief, and dread [ 11 , 78 ], threatening the core self and tasking one to make the change necessary to move forward in life with greater meaning and authenticity [ 11 ]. Death, meaning, guilt and isolation are among the common existential concerns that arise during the lifetime [ 77 , 78 , 79 , 80 ].

While limited empirical evidence has linked climate anxiety with existential concerns, several authors propose that climate change raises the salience of existential concerns such as mortality, thereby evoking climate anxiety [ 11 , 29 , 79 ]. One study conducted a thematic analysis to ascertain whether existential themes were present in interviews with psychotherapy patients describing their experiences of climate anxiety [ 11 ]. Within these interviews, patients described the salience of mortality, fearing the “death of humanity,” and the meaninglessness of existence, among other existential themes. It has also been suggested that feelings of vulnerability that arise when confronted with existential concerns may prompt the use of defense strategies (e.g., denial and apathy) as a means of restoring psychological equilibrium [ 68 ]. The proposed theoretical model thus considers climate anxiety as a manifestation of the existential conflict evoked by climate change, whereby one’s accompanying cognitive and behavioral strategies attempt to mitigate this conflict.

1.3.3 Systemic influences

How climate anxiety is experienced and coped with can be influenced by a range of factors outside of an individual’s cognitive, emotional, and behavioral characteristics. There is accumulating evidence that suggests specific populations are at greater risk of climate anxiety. Research has linked an increased risk of climate anxiety or climate change-related distress to females [ 12 , 23 , 71 , 81 ], young people [ 24 , 33 , 82 ], and individuals with pro-environmental identity [ 23 , 71 ], left-leaning political values [ 83 ], and those with pre-existing anxiety or stress [ 9 , 35 , 76 ]. First responders, health care providers, activists, and scientists, who are more exposed to or have greater knowledge of the direct impacts of climate change, may be more likely to experience higher levels of climate anxiety [ 18 , 29 , 84 ]. Unique vulnerability to specific climate-related events may also foster heightened anxiety. For example, an individual may experience greater climate anxiety if damage brought by natural disasters (e.g., flooding, fires) presents a risk to their homes, livelihoods, and environment, particularly for those with strong cultural or spiritual connections to land [ 12 , 85 ]. This may be especially the case for populations residing in the Global South, where climate change exacerbates pre-existing vulnerabilities such as extreme weather, scarcity of resources, forced migration, and poverty [ 86 ]. What is clear is that as climate change stressors (e.g., extreme weather events) fluctuate in salience for different groups across time, psychological and emotional responses are likely to be similarly dynamic [ 87 ].

Psychological responses to climate change should not be viewed in isolation, but rather as existing within layers of social, community, cultural, and political systems [ 86 , 88 , 89 ]. This has been conceptualized within a social-ecological framework, which illustrates how various nested contextual ‘systems’ within which an individual is embedded (e.g., family, peers, work or school, community, culture, the government) may shape whether and how that person experiences climate anxiety [ 86 ]. At the individual level, potential influences include temperament, biology/neurology, coping styles, any pre-existing mental health conditions, knowledge about climate change, as well as unique experience and vulnerability to climate change in their immediate context. At the microsystemic level, influences include the attitudes and experiences of family and friends, as well as how climate change is discussed or communicated in close relationships. Factors in the mesosystem may include community resources, community responses to climate change, as well as shared communal stress resulting from these changes. Exosystemic factors include governmental attitudes, collaboration or conflict between nations, the selection and implementation (or lack) of policies aimed to mitigate climate change, as well as how climate change is communicated in media. Additionally, overarching cultural influences in the macrosystem include cultural knowledge and attitudes, spiritual or religious connections to land, and potential loss of culture or cultural places due to climate change. While the conceptual model proposed in the current paper highlights affective, cognitive, behavioral, and existential dimensions of climate anxiety and coping at the individual level, it is important for researchers and practitioners to consider how these domains interact with an individual’s wider social-ecological contexts (for further detail, refer to Crandon, Scott [ 86 ]).

1.4 Model development

In the current paper, the authors constructed a model of climate anxiety and coping by drawing together the theoretical concepts previously described using a ‘top-down’ approach. Based on social-ecological theory and its application to climate change (see Crandon et al. [ 86 ]), environmental factors were considered as shaping all climate-related triggers and therefore how someone responds to those triggers. For this reason, systemic and contextual factors are depicted at the top of the model. From here, the model shows that acute and chronic triggers can then evoke climate anxiety. Existential theories were considered and ultimately included in the model to highlight that climate anxiety is an existential conflict that can be minimized or strengthened depending on the subsequent process of coping. The existential conflict (i.e., climate anxiety) then establishes an affective experience which has emotional and physiological components. The authors then applied the literature on coping to illustrate how climate anxiety can prompt a process of coping, which encompasses ongoing cognitive and behavioral efforts/strategies to reduce or mitigate the existential conflict of climate change and its associated experienced affect. Well-established evidence-based cognitive and behavioral theories were acknowledged in the model by showing how affect, cognitive appraisals and behavioral responses are bidirectionally influenced, and impact on overall functioning. Finally, the ongoing coping process informs the degree of adaptivity to the existential conflict caused by climate change.

2 Introduction of a theoretical model of climate anxiety and coping

The current conceptual model proposes that even after acknowledging systemic influences [ 86 ], climate anxiety is an existential conflict that leads to affective/emotional experiences. Individuals manage and respond to these experiences using cognitive and behavioral coping processes (see Fig.  2 ), which can be adaptive (when reinforcing functioning) or maladaptive (when interfering with functioning). Cognitive and behavioral coping may separately or simultaneously focus on solving or mitigating the problems of climate change, as well as soothing or removing the emotions evoked by it. We propose that individuals are not fixed to one process but can engage in more than one at the same time and over time, thereby oscillating on a continuum between low and high degrees of adaptive coping. Importantly, the affective (felt emotion/s), cognitive (thought content and process), and behavioral responses (actions and impacts) will be unique for an individual, as influenced by intrapersonal psychological factors, as well as the external social-ecological systems surrounding them.

2.1 Adaptive coping

When faced with triggers of the climate crisis, processing of climate change as a significant, overwhelming challenge may lead to existential conflict. As part of this conflict, autonomic arousal (e.g., discomfort, unease, shakiness, impaired concentration, increased heart rate, muscular tension, tight chest) and emotional experiences (e.g., angst, anxiety, fear, dread, anger, helplessness) act as a cue, whereby cognitive appraisal determines a course of action to reduce the threat [ 90 ]. This appraisal involves specific thoughts (e.g., what an individual thinks about climate change) and thought process (e.g., the mechanics of thinking). Adaptive thought content may be meaning-focused or hope inducing, be influenced by strongly held beliefs or values, involve interpretations about the greater historical context of climate change, or considerations of different climate solutions being developed and implemented [ 66 ]. For adaptive thought processes, an individual may engage in problem solving/solutions-focused thinking or cultivate acceptance of uncertainty and negative emotions [ 49 ].

These adaptive cognitive modes interact to identify behaviors that can help connect an individual to respond in ways that are consistent with their values. Responses might include practical strategies such as pro-environmental behavior, activism, or emotion-focused strategies such as engaging with social support and activities in nature [ 49 , 66 , 91 ]. While this does not remove the threat, the individual integrates the conflict in a way that allows them to function and move forward with life in a meaningful way [ 91 ] so as to manage the threat as adaptively as possible. However, with the likely continued exposure of climate change related triggers in the future, the individual may continue to undergo an ongoing cycle of direct and indirect exposure to climate change triggers, making subsequent attempts to resolve or alleviate distress an ongoing process.

2.2 Maladaptive coping

Humans have the innate capacity to manage and resolve existential conflicts in personally meaningful ways [ 92 ]. However, there is potential for this process to be challenged when complicated by internal (e.g., pre-existing mental illness, a tendency for ruminative thought patterns or unhelpful coping strategies) and external (e.g., unexpected loss, migration, death) factors. When this occurs, an individual’s thoughts about climate change may be negatively biased (e.g., catastrophizing) or they may use unhelpful thought processes (e.g., ruminative, or excessive preoccupation styles of thinking) [ 49 , 66 ]. Individuals may struggle to identify or implement strategies that could help mitigate their existential conflict, or they may succumb to habitual maladaptive behavior to avoid or alleviate negative feelings in the short term (e.g., excessive suppression, substance use, risk-taking behaviors) [ 18 , 93 , 94 ].

Alternatively, those who engage strategies to address climate change may do so to an unhelpful degree, making significant lifestyle changes that negatively impact their well-being in the long term by devoting much of their personal resources (e.g., time, energy) to climate action and experiencing ‘burnout’ or disillusionment as a result [ 18 ]. The loss of control ensued from unhelpful cognitive and behavioral patterns may then impair personal, social, and occupational domains. In turn, the feelings of climate anxiety and existential conflict are perpetuated and the individual may be less equipped to cope with future climate change triggers [ 95 ].

3 Discussion

3.1 using the theoretical model of climate anxiety and coping in measurement.

Developing a measure of psychological phenomena that is reliable, valid, and operational, is difficult without a consistent and comprehensive theoretical underpinning [ 96 ]. The theoretical model of climate anxiety and coping presented in the current paper draws together existing theories and measures to highlight climate anxiety as an affective construct, with potential correlates or sub-domains of (1) cognitive coping; (2) behavioral coping; and (3) functional impact. Existing measures may help to inform the development and selection of items that may assess these constructs, as well as any potential lower order constructs (see Table  2 for an example).

It is important for a climate anxiety measure to include both adaptive and maladaptive dimensions of cognitive and behavioral coping. The Measure of Affect Regulation (MARS) provides a useful example of a scale that measures frequency of 13 coping strategies, which differ based on whether the strategy is cognitive or behavioral, and whether the strategy focuses on resolving the situation or one’s mood [ 97 ]. Some examples of coping strategies include cognitive reappraisal (finding meaning or considering alternative perspectives), suppression (not allowing the expression of emotion), problem solving action (planning or acting to solve the problem), socializing or seeking help, and withdrawal or self-isolation. One study extended the MARS by including environmental strategies (e.g., going to favorite natural places). A climate anxiety measure might similarly assess coping across three dimensions: (1) cognitive vs. behavioral, (2) adaptive vs. maladaptive, and (3) focusing on resolving climate change, or the existential conflict/anxious feelings.

3.2 Identifying potential areas to intervene

Addressing the drivers of climate change will most significantly alleviate the negative implications for mental health and well-being. Yet, there is also increasing demand for interventions to support those who experience climate anxiety and the unmitigated impacts that climate change is having on mental health. Using a multiple needs framework, Bingley, Tran [ 36 ] identified that climate anxiety interventions can and do focus on meeting one or a combination of individual (e.g., improving individual well-being), social (e.g., fostering social connection), and environmental needs (e.g., pro-environmental attitude or behavior). As a starting platform, the theoretical model of climate anxiety and coping could be used to identify where to intervene in order to meet these needs or outcomes. More specifically, targeting cognitive or behavioral coping may be the mechanism for shifting levels of functioning across personal, social, and occupational domains. Interventions addressing cognitive coping may, for example, aim to support one to manage dysfunctional/unhelpful styles of thinking (individual needs), identify their shared values (social needs), or explore their role as it relates to climate change (environmental needs). Behaviorally focused interventions may, for example, focus on developing relaxation strategies (individual needs), fostering peer interaction or community building (social needs) or target climate mitigation through project engagement or decreasing one’s carbon footprint (environmental needs). These are few of the many ideas that can be developed based on the premise that coping will impact functioning, rather than emotion.

3.2.1 Ideas for interventions for maladaptive climate anxiety

Given the potential for maladaptive coping responses to climate anxiety, some individuals may need more comprehensive individual support. It is important, however, to note that little evidence exists for the efficacy of existing therapeutic interventions in supporting those who specifically experience climate anxiety. Potential interventions that could be drawn on to target cognitive or behavioral coping could include cognitive behavior therapy [ 66 ], acceptance and commitment therapy (accepting ecological emotions such as ecological grief or climate anxiety, connecting one to their values and committing to action) [ 7 , 95 ], and interventions which may be existentially-focused, self-care focused [ 7 , 31 ], emotion-focused [ 21 ], nature-based [ 6 ], as well as peer or group-based [ 21 , 100 ]. As mentioned, research is needed to evaluate the effectiveness of these approaches in helping to strengthen adaptive coping and minimize impairments to functioning. Furthermore, such support should also consider the social and ecological influences that may shape adaptivity (e.g., geography, political landscape, culture) [ 29 , 86 ].

As climate change impacts the world, there is an urgent need for intervention to be delivered beyond an individual receiving one-to-one support. Systemic interventions that aim to support public mental health can aim to promote adaptive cognitive and behavioral coping. For example, public health campaigns might consider reducing guilt, fear, and shame messaging, and instead communicating that climate anxiety is a shared experience that encompasses a range of different emotions that are understandable and can be responded to adaptively. Alternatively, media outlets that disseminate information on organizations, policies and initiatives that are working to address climate change could help to promote hope-based thinking and positive re-appraisals. Tangible strategies for coping could be given alongside these messages, particularly as they relate to meeting individual, social and environmental needs (e.g., online self-help tools for managing emotions, promoting community projects, funding climate mitigation projects) [ 7 ]. As with individual interventions, the focus of public-health level strategies is not to eliminate climate anxiety feelings, but to support planetary and human health and well-being.

3.3 Limitations and ideas for future research

The proposed theoretical model draws on the current state of evidence on climate anxiety along with well-established psychological theories of emotion and coping. It was developed based on a wide-ranging narrative review of existing research which included assessment of previous systematic reviews, and psychological theory brought together through clinical expertise of the contributing authors. Given the study was not a systematic review, this may have introduced a source of bias as to the way psychological concepts reviewed in the paper were used to construct the model presented. It must also be acknowledged that the theoretical model is untested, and empirical validation studies are greatly needed. Importantly, the model should be further informed and refined through future research and practice. Pathway analyses could help to evaluate potential associations between emotions about climate change, how individuals cognitively and behaviorally cope, and their relationship with functioning. Investigating how systemic factors influence climate anxiety and the capacity for coping will also help to determine how best to strengthen adaptive coping and over time, psychological resilience. Despite these limitations and the need for future research, this paper importantly responded to recommendations made by previous systematic reviews [ 12 , 13 , 30 ]. Specifically, the theoretical model of climate anxiety and coping helps to refine the conceptual understanding of climate anxiety and coping. From here, its application provides a way to assess psychological responses to climate change using measurement tools, which may in turn help in developing ways to support well-being as the climate crisis continues.

4 Conclusions

Within the theoretical model of climate anxiety and coping, competing viewpoints on the nature of climate anxiety converge. Specifically, climate anxiety may vary in adaptivity, intensity and severity. While existing measures capture maladaptive dimensions of climate anxiety, there is need for continued focus on developing a measure that can fully capture the spectrum of coping responses. Development of such a measure would (1) contribute to a more comprehensive understanding of how climate anxiety is experienced, (2) reduce potential inappropriate pathologizing of individuals who experience significant anxiety, yet who are responding to that anxiety in helpful ways; (3) identify those individuals who may be most in need of targeted mental health intervention; and (4) be used in future research to explore which factors may be associated with more or less adaptive responses. For interventions to be most effective, the range of experiences across cognitive, emotional, behavioral, existential, and systemic domains must be considered. Crucially, whether an intervention is delivered with the individual or at the public health level, focus should not be on eliminating climate anxiety, but in supporting the ability to channel that anxiety using cognitive and behavioral strategies to best meet the needs of the individual. With a more theoretically driven and robust conceptualization of climate anxiety, further research on climate anxiety may be able to better identify ways in which to support those adversely affected.

Data availability

Not applicable.

Manning C, Clayton S. Threats to mental health and wellbeing associated with climate change. In: Clayton S, Manning C, editors. Psychology and climate change. Academic Press; 2018. p. 217–44.

Chapter   Google Scholar  

Hayes K, Blashki G, Wiseman J, Burke S, Reifels L. Climate change and mental health: risks, impacts and priority actions. Int J Ment Health Syst. 2018. https://doi.org/10.1186/s13033-018-0210-6 .

Article   PubMed   PubMed Central   Google Scholar  

Charlson F, Ali S, Benmarhnia T, Pearl M, Massazza A, Augustinavicius J, et al. Climate change and mental health: a scoping review. Int J Environ Res Public Health. 2021. https://doi.org/10.3390/ijerph18094486 .

Chen NT, Lin PH, Guo YL. Long-term exposure to high temperature associated with the incidence of major depressive disorder. Sci Total Environ. 2019. https://doi.org/10.1016/j.scitotenv.2018.12.434 .

Ayeb-Karlsson S. ‘When we were children we had dreams, then we came to Dhaka to survive’: urban stories connecting loss of wellbeing, displacement and (im)mobility. Clim Dev. 2021. https://doi.org/10.1080/17565529.2020.1777078 .

Article   Google Scholar  

Clayton S. Climate anxiety: psychological responses to climate change. J Anxiety Disord. 2020. https://doi.org/10.1016/j.janxdis.2020.102263 .

Article   PubMed   Google Scholar  

Doherty TJ. 10—individual impacts and resilience. In: Clayton S, Manning C, editors. Psychology and climate change. Academic Press; 2018. p. 245–66.

Thompson HE. Climate “psychopathology.” Eur Psychol. 2021. https://doi.org/10.1027/1016-9040/a000433 .

Panu P. Anxiety and the ecological crisis: an analysis of eco-anxiety and climate anxiety. Sustainability. 2020. https://doi.org/10.3390/su12197836 .

Hickman C. We need to (find a way to) talk about … Eco-anxiety. J Soc Work Pract. 2020. https://doi.org/10.1080/02650533.2020.1844166 .

Budziszewska M, Jonsson SE. From climate anxiety to climate action: an existential perspective on climate change concerns within psychotherapy. J Humanist Psychol. 2021. https://doi.org/10.1177/0022167821993243 .

Coffey Y, Bhullar N, Durkin J, Islam MS, Usher K. Understanding eco-anxiety: a systematic scoping review of current literature and identified knowledge gaps. J Clim Change Health. 2021. https://doi.org/10.1016/j.joclim.2021.100047 .

Soutar C, Wand APF. Understanding the spectrum of anxiety responses to climate change: a systematic review of the qualitative literature. Int J Environ Res Public Health. 2022. https://doi.org/10.3390/ijerph19020990 .

DeVellis RF, Thorpe CT. Scale development: theory and applications. SAGE Publications; 2021.

Google Scholar  

Richard JB, Ann F, Robert EOC. Public perceptions of global warming: United States and international perspectives. Clim Res. 1998;11(1):75–84.

Dunlap R. International attitudes towards environment and development. In: Bergesen H, Parmann G, editors. Yearbook of international co-operation on environment and development. Oxford: Oxford University Press; 1994. p. 115–26.

Frumkin H. Beyond toxicity: human health and the natural environment. Am J Prev Med. 2001. https://doi.org/10.1016/S0749-3797(00)00317-2 .

Dodds J. The psychology of climate anxiety. BJPsych Bulletin. 2021. https://doi.org/10.1192/bjb.2021.18[Opensinanewwindow] .

Ellison K. Conservation on the couch. Front Ecol Environ. 2008. https://doi.org/10.1890/1540-9295(2008)6[168:COTC]2.0.CO;2 .

Nobel J. Eco-anxiety: something else to worry about. The Philadelphia Inquirer; 2007.

Dailianis A. Eco-anxiety: a scoping review towards a clinical conceptualisation and therapeutic approach [thesis]: Auckland Univ. Techno.; 2020.

Fritze JG, Blashki GA, Burke S, Wiseman J. Hope, despair and transformation: climate change and the promotion of mental health and wellbeing. Int J Ment Health Syst. 2008. https://doi.org/10.1186/1752-4458-2-13 .

Searle K, Gow K. Do concerns about climate change lead to distress? Int J Clim Chang Strateg Manag. 2010. https://doi.org/10.1108/17568691011089891 .

Clayton S, Karazsia BT. Development and validation of a measure of climate change anxiety. J Environ Psychol. 2020. https://doi.org/10.1016/j.jenvp.2020.101434 .

Albrecht G. Chronic environmental change: emerging ‘psychoterratic’ syndromes. In: Weissbecker I, editor. Climate change and human well-being: global challenges and opportunities. New York: Springer; 2011. p. 43–56.

Pihkala P. Eco-anxiety and environmental education. Sustainability. 2020. https://doi.org/10.3390/su122310149 .

Pihkala P. Eco-anxiety, tragedy and hope: psychological and spiritual dimensions of climate change. Zygon. 2018. https://doi.org/10.1111/zygo.12407 .

Helm SV, Pollitt A, Barnett MA, Curran MA, Craig ZR. Differentiating environmental concern in the context of psychological adaption to climate change. Glob Environ Change. 2018. https://doi.org/10.1016/j.gloenvcha.2017.11.012 .

Ojala M, Cunsolo A, Ogunbode CA, Middleton J. Anxiety, worry, and grief in a time of environmental and climate crisis: a narrative review. Annu Rev Environ Resour. 2021. https://doi.org/10.1146/annurev-environ-012220-022716 .

Boluda-Verdú I, Senent-Valero M, Casas-Escolano M, Matijasevich A, Pastor-Valero M. Fear for the future: eco-anxiety and health implications, a systematic review. J Environ Psychol. 2022. https://doi.org/10.1016/j.jenvp.2022.101904 .

Baudon P, Jachens L. A scoping review of interventions for the treatment of eco-anxiety. Int J Environ Res Public Health. 2021. https://doi.org/10.3390/ijerph18189636 .

Koder J, Dunk J, Rhodes P. Climate distress: a review of current psychological research and practice. Sustainability. 2023;15:8115.

Hickman C, Marks E, Pihkala P, Clayton S, Lewandowski RE, Mayall EE, et al. Climate anxiety in children and young people and their beliefs about government responses to climate change: a global survey. Lancet Planet Health. 2021. https://doi.org/10.1016/S2542-5196(21)00278-3 .

Comtesse H, Ertl V, Hengst SMC, Rosner R, Smid GE. Ecological grief as a response to environmental change: a mental health risk or functional response? Int J Environ Res Public Health. 2021. https://doi.org/10.3390/ijerph18020734 .

Stanley SK, Hogg TL, Leviston Z, Walker I. From anger to action: differential impacts of eco-anxiety, eco-depression, and eco-anger on climate action and wellbeing. J Clim Change Health. 2021. https://doi.org/10.1016/j.joclim.2021.100003 .

Bingley WJ, Tran A, Boyd CP, Gibson K, Kalokerinos EK, Koval P, et al. A multiple needs framework for climate change anxiety interventions. Am Psychol. 2022. https://doi.org/10.1037/amp0001012 .

American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-5. 5th ed. Washington: American Psychiatric Association; 2013.

Book   Google Scholar  

Wang H, Safer DL, Cosentino M, Cooper R, Van Susteren L, Coren E, et al. Coping with eco-anxiety: an interdisciplinary perspective for collective learning and strategic communication. J Clim Change Health. 2023. https://doi.org/10.1016/j.joclim.2023.100211 .

Schwartz SEO, Benoit L, Clayton S, Parnes MF, Swenson L, Lowe SR. Climate change anxiety and mental health: environmental activism as buffer. Curr Psychol. 2022. https://doi.org/10.1007/s12144-022-02735-6 .

Daum I, Markowitsch HJ, Vandekerckhove M. Neurobiological basis of emotions. In: Markowitsch HJ, Röttger-Rössler B, editors. Emotions as bio-cultural processes. New York: Springer; 2009. p. 111–38.

Harth NS. Affect, (group-based) emotions, and climate change action. Curr Opin in Psychol. 2021. https://doi.org/10.1016/j.copsyc.2021.07.018 .

Keltner D, Sauter D, Tracy J, Cowen A. Emotional expression: advances in basic emotion theory. J Nonverbal Behav. 2019. https://doi.org/10.1007/s10919-019-00293-3 .

Steimer T. The biology of fear- and anxiety-related behaviors. Dialogues Clin Neurosci. 2002. https://doi.org/10.31887/DCNS.2002.4.3/tsteimer .

Hebert EA, Dugas MJ. Behavioral experiments for intolerance of uncertainty: challenging the unknown in the treatment of generalized anxiety disorder. Cogn Behav Pract. 2019. https://doi.org/10.1016/j.cbpra.2018.07.007 .

Grupe DW, Nitschke JB. Uncertainty and anticipation in anxiety: an integrated neurobiological and psychological perspective. Nat Rev Neurosci. 2013. https://doi.org/10.1038/nrn3524 .

Lewis J. In the room with climate anxiety. Psychiatr Times. 2018; 35(11). https://www.psychiatrictimes.com/view/room-climate-anxiety .

Ogunbode CA, Doran R, Hanss D, Ojala M, Salmela-Aro K, van den Broek KL, et al. Climate anxiety, wellbeing and pro-environmental action: correlates of negative emotional responses to climate change in 32 countries. J Environ Psychol. 2022. https://doi.org/10.1016/j.jenvp.2022.101887 .

Sciberras E, Fernando JW. Climate change-related worry among Australian adolescents: an eight-year longitudinal study. Child Adolesc Mental Health. 2021. https://doi.org/10.1111/camh.12521 .

Crandon TJ, Scott JG, Charlson FJ, Thomas HJ. Coping with climate anxiety: impacts on functioning in Australian adolescents. Submitted. 2023.

Lazarus RS, Folkman S. Stress, appraisal, and coping. New York: Springer; 1984.

Stanisławski K. The coping circumplex model: an integrative model of the structure of coping with stress. Front Psychol. 2019. https://doi.org/10.3389/fpsyg.2019.00694 .

Ojala M. Regulating worry, promoting hope: how do children, adolescents, and young adults cope with climate change? Int J Environ Sci Educ. 2012;7(4):537–61.

Ojala M. Coping with climate change among adolescents: implications for subjective well-being and environmental engagement. Sustainability. 2013. https://doi.org/10.3390/su5052191 .

Ojala M, Bengtsson H. Young people’s coping strategies concerning climate change: relations to perceived communication with parents and friends and proenvironmental behavior. Environ Behav. 2018. https://doi.org/10.1177/001391651876389 .

Pihkala P. The process of eco-anxiety and ecological grief: a narrative review and a new proposal. Sustainability. 2022;14:16628.

Cunsolo A, Ellis NR. Ecological grief as a mental health response to climate change-related loss. Nat Clim Change. 2018. https://doi.org/10.1038/s41558-018-0092-2 .

Schneider CR, Zaval L, Markowitz EM. Positive emotions and climate change. Curr Opin Behav Sci. 2021. https://doi.org/10.1016/j.cobeha.2021.04.009 .

Ojala M. Adolescents’ worries about environmental risks: subjective well-being, values, and existential dimensions. J Youth Stud. 2005. https://doi.org/10.1080/13676260500261934 .

Albrecht GA. Negating solastalgia: an emotional revolution from the Anthropocene to the Symbiocene. Am Imago. 2020. https://doi.org/10.1353/aim.2020.0001 .

Kleres J, Wettergren Å. Fear, hope, anger, and guilt in climate activism. Soc Mov Stud. 2017. https://doi.org/10.1080/14742837.2017.1344546 .

Okon-Singer H, Hendler T, Pessoa L, Shackman AJ. The neurobiology of emotion–cognition interactions: fundamental questions and strategies for future research. Front Hum Neurosci. 2015. https://doi.org/10.3389/fnhum.2015.00058 .

Niedenthal PM, Ric F. Psychology of emotion. London: Taylor & Francis Group; 2017.

Innocenti M, Santarelli G, Faggi V, Castellini G, Manelli I, Magrini G, et al. Psychometric properties of the Italian version of the Climate Change Anxiety Scale. J Clim Change Health. 2021. https://doi.org/10.1016/j.joclim.2021.100080 .

Wullenkord M, Toger J, Hamann KR, Loy L, Reese G. Anxiety and climate change: a validation of the climate anxiety scale in a German-speaking quota sample and an investigation of psychological correlates. Clim Change. 2021. https://doi.org/10.31234/osf.io/76ez2 .

Hogg T, Stanley S, O’Brien L, Wilson M, Watsford C. The Hogg eco-anxiety scale: development and validation of a multidimensional scale. Glob Environ Change. 2021. https://doi.org/10.31219/osf.io/rxudb .

Burke S, Blashki G. Climate change anxiety and our mental health. In: Climate health and courage. Melbourne: Future Leaders; 2020. https://www.futureleaders.com.au/book_chapters/pdf/Climate-Health-and-Courage/Susie-Burke-and-Grant-Blashki.pdf .

Lutz PK, Passmore H-A, Howell AJ, Zelenski JM, Yang Y, Richardson M. The continuum of eco-anxiety responses: a preliminary investigation of its nomological network. Collabra: Psychology. 2023. https://doi.org/10.1525/collabra.67838 .

Beck AT, Clark DA. An information processing model of anxiety: automatic and strategic processes. Behav Res Ther. 1997. https://doi.org/10.1016/S0005-7967(96)00069-1 .

Sangervo J, Jylhä KM, Pihkala P. Climate anxiety: conceptual considerations, and connections with climate hope and action. Glob Environ Chang. 2022. https://doi.org/10.1016/j.gloenvcha.2022.102569 .

Ojala M. How do children, adolescents, and young adults relate to climate change? Implications for developmental psychology. Eur J Dev Psychol. 2022. https://doi.org/10.1016/j.jenvp.2012.02.004 .

Verplanken B, Marks E, Dobromir AI. On the nature of eco-anxiety: How constructive or unconstructive is habitual worry about global warming? J Environ Psychol. 2020. https://doi.org/10.1016/j.jenvp.2020.101528 .

Wells A. Metacognitive therapy for anxiety and depression. New York: Guilford Publications; 2011.

Beck AT, Haigh EAP. Advances in cognitive theory and therapy: the generic cognitive model. Annu Rev Clin Psychol. 2014. https://doi.org/10.1146/annurev-clinpsy-032813-153734 .

Ramos-Cejudo J, Salguero JM. Negative metacognitive beliefs moderate the influence of perceived stress and anxiety in long-term anxiety. Psychiatry Res. 2017. https://doi.org/10.1016/j.psychres.2017.01.056 .

Pollyana Caldeira L, Tiago Costa G, da Luiz Carlos Ferreira S, Teixeira-Silva F. Trait vs. state anxiety in different threatening situations. Trends in Psychiatry and Psychother. 2017. https://doi.org/10.1590/2237-6089-2016-0044 .

Materia CJ. Climate state anxiety and connectedness to nature in rural Tasmania [thesis]: Univ. Tasmania; 2016.

Yalom ID, Josselson R. Existential psychotherapy. In: Wedding D, Corsini RJ, editors. Current psychotherapies. 10th ed. California: Cengage Learning; 2013.

Weems CF, Costa NM, Dehon C, Berman SL. Paul Tillich’s theory of existential anxiety: a preliminary conceptual and empirical examination. Anxiety Stress Coping. 2004. https://doi.org/10.1080/10615800412331318616 .

Pihkala P. Death, the environment, and theology. Dialog. 2018. https://doi.org/10.1111/dial.12437 .

Yalom ID. Existential psychotherapy. Michigan: Basic Books; 1980.

Ballman C. Emotions and actions: eco-anxiety and pro-environmental behaviours [thesis]: Univ. Regina; 2020.

Chhokar K, Dua S, Taylor N, Boyes E, Stanisstreet M. Senior secondary Indian students’ views about global warming, and their implications for education. Sci Educ. 2012;23(2):133–49.

Gregersen T, Doran R, Böhm G, Tvinnereim E, Poortinga W. Political orientation moderates the relationship between climate change beliefs and worry about climate change. Front Psychol. 2020. https://doi.org/10.3389/fpsyg.2020.01573 .

Clayton S. Mental health risk and resilience among climate scientists. Nat Clim Change. 2018. https://doi.org/10.1038/s41558-018-0123-z .

Kubo T, Tsuge T, Abe H, Yamano H. Understanding island residents’ anxiety about impacts caused by climate change using Best-Worst Scaling: a case study of Amami islands. Japan Sustain. 2019. https://doi.org/10.1007/s11625-018-0640-8 .

Crandon TJ, Scott JG, Charlson FJ, Thomas HJ. A social-ecological perspective of climate anxiety in children and adolescents. Nat Clim Change. 2022. https://doi.org/10.1038/s41558-021-01251-y .

Crandon TJ, Dey C, Scott JG, Thomas HJ, Ali S, Charlson FJ. The clinical implications of climate change for mental health. Nat Hum Behav. 2022. https://doi.org/10.1038/s41562-022-01477-6 .

Adams M. Critical psychologies and climate change. Curr Opin in Psychol. 2021. https://doi.org/10.1016/j.copsyc.2021.01.007 .

Gislason MK, Kennedy AM, Witham SM. The interplay between social and ecological determinants of mental health for children and youth in the climate crisis. Int J Environ Res Public Health. 2021. https://doi.org/10.3390/ijerph18094573 .

Ingle HE, Mikulewicz M. Mental health and climate change: tackling invisible injustice. The Lancet Planet Health. 2020. https://doi.org/10.1016/S2542-5196(20)30081-4 .

Stollberg J, Jonas E. Existential threat as a challenge for individual and collective engagement: climate change and the motivation to act. Curr Opin in Psychol. 2021. https://doi.org/10.1016/j.copsyc.2021.10.004 .

Wong PTP. Existential positive psychology and integrative meaning therapy. Int Rev Psychiatry. 2020. https://doi.org/10.1080/09540261.2020.1814703 .

Swerdlow BA, Pearlstein JG, Sandel DB, Mauss IB, Johnson SL. Maladaptive behavior and affect regulation: a functionalist perspective. Emotion. 2020. https://doi.org/10.1037/emo0000660 .

Boehme S, Biehl SC, Mühlberger A. Effects of differential strategies of emotion regulation. Brain Sci. 2019. https://doi.org/10.3390/brainsci9090225 .

Koger SM. A burgeoning ecopsychological recovery movement. Ecopsychology. 2015. https://doi.org/10.1089/eco.2015.0021 .

DeVellis RF, Thorpe CT. Scale development: theory and applications. Bickman L, Rog DJ, editors. California: SAGE Publications; 2021.

Korpela KM, Pasanen T, Repo V, Hartig T, Staats H, Mason M, et al. Environmental strategies of effect regulation and their associations with subjective well-being. Front Psychol. 2018. https://doi.org/10.3389/fpsyg.2018.00562 .

Lovibond PF, Lovibond SH. The structure of negative emotional states: comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories. Behav Res Ther. 1995. https://doi.org/10.1016/0005-7967(94)00075-U .

Knowles SR, Apputhurai P, Bates G. Development and validation of the brief unhelpful thoughts scale (BUTs). J Psychol Psychother Res. 2017. https://doi.org/10.12974/2313-1047.2017.04.02.1 .

Gillespie S. Climate change and psyche: conversations with and through dreams. Int J Mult Res. 2013. https://doi.org/10.5172/mra.2013.7.3.343 .

Download references

T.C. is supported by the Child and Youth Mental Health Research Group PhD Scholarship. H.T. and F.C. are supported by the Queensland Centre for Mental Health Research which is funded by the Queensland Department of Health.

Author information

Authors and affiliations.

QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, QLD, 4006, Australia

Tara J. Crandon, James G. Scott & Hannah J. Thomas

School of Public Health, The University of Queensland, St Lucia, QLD, Australia

Tara J. Crandon, Fiona J. Charlson & Hannah J. Thomas

Child Health Research Centre, The University of Queensland, Herston, QLD, Australia

James G. Scott

Child and Youth Mental Health Service, Child Health Queensland, South Brisbane, QLD, Australia

Queensland Centre for Mental Health Research, Wacol, QLD, Australia

James G. Scott, Fiona J. Charlson & Hannah J. Thomas

Institute for Health Metrics and Evaluation, University of Washington, Seattle, USA

Fiona J. Charlson

You can also search for this author in PubMed   Google Scholar

Contributions

T.C. led the writing of the article, with J.S., F.C., and H.T., contributing to writing at all stages. All authors contributed to conceptualization and the editing process.

Corresponding author

Correspondence to Tara J. Crandon .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

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

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, 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 you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. 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-nc-nd/4.0/ .

Reprints and permissions

About this article

Crandon, T.J., Scott, J.G., Charlson, F.J. et al. A theoretical model of climate anxiety and coping. Discov Psychol 4 , 94 (2024). https://doi.org/10.1007/s44202-024-00212-8

Download citation

Received : 03 August 2023

Accepted : 01 August 2024

Published : 11 August 2024

DOI : https://doi.org/10.1007/s44202-024-00212-8

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

• Climate anxiety
• Eco anxiety
• Climate change
• Mental health

Advertisement

• Find a journal
• Publish with us
• Track your research

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

A novel urban heat vulnerability analysis: integrating machine learning and remote sensing for enhanced insights.

1. Introduction and Background

1.1. introduction, 1.2. the current state of urban heat vulnerability assessment, 1.3. the emerging trend of machine learning and remote sensing integration, 1.4. the focus of this study, 2. research design, the urban heat vulnerability analysis framework.

• Historical Mapping: Historical mapping in U-HEAT utilizes advanced ML and RS techniques to reconstruct detailed heat maps from diverse data sources. This phase is pivotal as it provides a comprehensive understanding of past and present heat vulnerabilities, allowing for the identification of long-term trends and spatial patterns in urban heat exposure. By fusing various socio-economic, environmental, and health-related data with high-resolution RS imagery, U-HEAT generates a nuanced portrayal of how urban heat vulnerability has evolved over time. This retrospective analysis is essential for establishing a baseline, understanding the historical context of current vulnerabilities, and identifying persistent hotspots that require targeted interventions. The detailed historical maps produced in this phase serve as a foundational reference for subsequent predictive mapping efforts, ensuring that future projections are grounded in a robust empirical understanding of past conditions.
• Predictive Mapping: Building on the insights gained from historical mapping, predictive mapping in U-HEAT integrates urban planning data to forecast future trends and distributions of urban heat vulnerability. This phase leverages the predictive power of ML models to simulate how urban heat patterns might evolve under various scenarios, such as climate change, population growth, and urban development. By incorporating forward-looking data, such as planned infrastructure projects and anticipated demographic shifts, U-HEAT can generate projections that inform proactive urban planning and policymaking. The predictive mapping capability is crucial for identifying emerging areas of concern and guiding the implementation of preventative measures. This forward-thinking approach ensures that cities can anticipate and mitigate future heat risks, enhancing their resilience and adaptability to climate change. Predictive mapping transforms U-HEAT from a reactive tool into a proactive planning resource, enabling urban planners to design cities that are better equipped to handle the challenges of rising temperatures.
• Relevance: Rooted in established frameworks and empirical research, U-HEAT’s approach to selecting indicators and gathering data is both relevant and representative.
• Precision: By transitioning from broad statistical areas to a more detailed grid-scale, U-HEAT provides a finer-grained and accurate depiction of urban heat vulnerability, benefitting from the integration of ML and RS.
• Comprehensiveness: U-HEAT not only maps historical data, but also predicts future urban heat trends, resulting in spatially detailed and temporally extensive outcomes.
• Sustainability: The U-HEAT framework’s ability to recommend mitigation strategies, adapt to new data, and provide ongoing monitoring highlights its sustainability.
• Criteria Development: To formulate a universal set of criteria for the selection and categorization of indicators, establishing a reference framework.
• Feasibility Demonstration: To showcase the practicality of conducting long-term, grid-scale, and precise assessments by integrating ML and RS technologies.
• Predictive Methodology: To bridge the existing gap in predictive methods by introducing an innovative approach for forecasting urban heat vulnerability trends in future decades.
• Framework Proposal: To offer a robust, enduring, and sustainable framework for the continuous, accurate, and focused monitoring and management of urban heat vulnerability challenges.

3. Integrated Urban Heat Vulnerability Analysis with Machine Learning and Remote Sensing

3.1. indicators and data selection, 3.1.1. popular reference frameworks, 3.1.2. indicator collection and categorization, 3.1.3. data collection and pre-processing, 3.2. historical mapping of urban heat vulnerability, 3.2.1. two scenarios of historical mapping, 3.2.2. challenges and algorithm selection for historical mapping, 3.2.3. model development, validation and effective communication of results, 3.3. future prediction of urban heat vulnerability, 3.3.1. lack of future prediction, 3.3.2. challenges and algorithm selection for future prediction, 3.3.3. model development, validation and presentation of results, 3.4. strategy recommendation, 3.5. continuous monitoring and updating, 4. findings and discussion, 4.1. key challenges and limitations in existing approaches, 4.2. prospective applications, 4.3. contributions to sustainable development, 4.4. implications in policy and public engagement, 4.5. assumptions and limitations, 5. conclusions, author contributions, data availability statement, acknowledgments, conflicts of interest.

• Kamruzzaman, M.; Deilami, K.; Yigitcanlar, T. Investigating the Urban Heat Island Effect of Transit Oriented Development in Brisbane. J. Transp. Geogr. 2018 , 66 , 116–124. [ Google Scholar ] [ CrossRef ]
• Degirmenci, K.; Desouza, K.; Fieuw, W.; Watson, R.; Yigitcanlar, T. Understanding Policy and Technology Responses in Mitigating Urban Heat Islands: A Literature Review and Directions for Future Research. Sustain. Cities Soc. 2021 , 70 , 102873. [ Google Scholar ] [ CrossRef ]
• Australian Institute of Health and Welfare. Let’s Talk about the Weather: Injuries Related to Extreme Weather. 2023. Available online: https://www.aihw.gov.au/reports/injury/extreme-weather-injuries/contents/about (accessed on 19 June 2024).
• Vescovi, L.; Rebetez, M.; Rong, F. Assessing Public Health Risk Due to Extremely High Temperature Events: Climate and Social Parameters. Clim. Res. 2005 , 30 , 71–78. [ Google Scholar ] [ CrossRef ]
• Harlan, S.; Brazel, A.J.; Prashad, L.; Stefanov, W.; Larsen, L. Neighborhood Microclimates and Vulnerability to Heat Stress. Soc. Sci. Med. 2006 , 63 , 2847–2863. [ Google Scholar ] [ CrossRef ]
• Reid, C.E.; O’Neill, M.S.; Gronlund, C.J.; Brines, S.J.; Brown, D.G.; Diez-Roux, A.V.; Schwartz, J. Mapping Community Determinants of Heat Vulnerability. Environ. Health Perspect. 2009 , 117 , 1730–1736. [ Google Scholar ] [ CrossRef ]
• Boumans, R.; Phillips, D.; Victery, W.; Fontaine, T. Developing a Model for Effects of Climate Change on Human Health and Health–Environment Interactions: Heat Stress in Austin, Texas. Urban Clim. 2014 , 8 , 78–99. [ Google Scholar ] [ CrossRef ]
• Zhang, W.; McManus, P.; Duncan, E. A Raster-Based Subdividing Indicator to Map Urban Heat Vulnerability: A Case Study in Sydney, Australia. Int. J. Environ. Res. Public Health 2018 , 15 , 2516. [ Google Scholar ] [ CrossRef ]
• Cheng, W.; Li, D.; Liu, Z.; Brown, R. Approaches for Identifying Heat-Vulnerable Populations and Locations: A Systematic Review. Sci. Total Environ. 2021 , 799 , 149417. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Adger, W.N. Vulnerability. Glob. Environ. Change 2006 , 16 , 268–281. [ Google Scholar ] [ CrossRef ]
• Tate, E. Social Vulnerability Indices: A Comparative Assessment Using Uncertainty and Sensitivity Analysis. Nat. Hazards 2012 , 63 , 325–347. [ Google Scholar ] [ CrossRef ]
• Chen, Q.; Ding, M.; Yang, X.; Hu, K.; Qi, J. Spatially Explicit Assessment of Heat Health Risk by Using Multi-Sensor Remote Sensing Images and Socioeconomic Data in Yangtze River Delta, China. Int. J. Health Geogr. 2018 , 17 , 13. [ Google Scholar ] [ CrossRef ]
• Niu, Y.; Li, Z.; Gao, Y.; Liu, X.; Xu, L.; Vardoulakis, S.; Liu, Q. A Systematic Review of the Development and Validation of the Heat Vulnerability Index: Major Factors, Methods, and Spatial Units. Curr. Clim. Chang. Rep. 2021 , 7 , 87–97. [ Google Scholar ] [ CrossRef ]
• Li, F.; Yigitcanlar, T.; Nepal, M.; Nguyen, K.; Dur, F. Understanding Urban Heat Vulnerability Assessment Methods: A PRISMA Review. Energies 2022 , 15 , 6998. [ Google Scholar ] [ CrossRef ]
• Kazak, J. The Use of a Decision Support System for Sustainable Urbanization and Thermal Comfort in Adaptation to Climate Change Actions: The Case of the Wrocław Larger Urban Zone (Poland). Sustainability 2018 , 10 , 1083. [ Google Scholar ] [ CrossRef ]
• He, C.; Ma, L.; Zhou, L.; Kan, H.; Zhang, Y.; Ma, W.; Chen, B. Exploring the Mechanisms of Heat Wave Vulnerability at the Urban Scale Based on the Application of Big Data and Artificial Societies. Environ. Int. 2019 , 127 , 573–583. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Holec, J.; Šveda, M.; Szatmári, D.; Feranec, J.; Bobáľová, H.; Kopecká, M.; Šťastný, P. Heat Risk Assessment Based on Mobile Phone Data: Case Study of Bratislava, Slovakia. Nat. Hazards 2021 , 108 , 3099–3120. [ Google Scholar ] [ CrossRef ]
• Weber, S.; Sadoff, N.; Zell, E.; de Sherbinin, A. Policy-Relevant Indicators for Mapping the Vulnerability of Urban Populations to Extreme Heat Events: A Case Study of Philadelphia. Appl. Geogr. 2015 , 63 , 231–243. [ Google Scholar ] [ CrossRef ]
• Wilson, B.; Chakraborty, A. Mapping Vulnerability to Extreme Heat Events: Lessons from Metropolitan Chicago. J. Environ. Plan. Manag. 2019 , 62 , 1065–1088. [ Google Scholar ] [ CrossRef ]
• Heaton, M.; Sain, S.; Greasby, T.; Uejio, C.; Hayden, M.; Monaghan, A.; Boehnert, J.; Sampson, K.; Banerjee, D.; Nepal, V.; et al. Characterizing Urban Vulnerability to Heat Stress Using a Spatially Varying Coefficient Model. Spat. Spatio-Temporal Epidemiol. 2014 , 8 , 23–33. [ Google Scholar ] [ CrossRef ]
• Yin, Y.; Grundstein, A.; Mishra, D.R.; Ramaswamy, L.; Tonekaboni, N.; Dowd, J. DTEx: A Dynamic Urban Thermal Exposure Index Based on Human Mobility Patterns. Environ. Int. 2021 , 155 , 106573. [ Google Scholar ] [ CrossRef ]
• Bu, S.; Smith, K.L.; Masoud, F.; Sheinbaum, A. Spatial distribution of heat vulnerability in Toronto, Canada. Urban Clim. 2024 , 54 , 101838. [ Google Scholar ] [ CrossRef ]
• Barron, L.; Ruggieri, D.; Branas, C. Assessing Vulnerability to Heat: A Geospatial Analysis for the City of Philadelphia. Urban Sci. 2018 , 2 , 38. [ Google Scholar ] [ CrossRef ]
• Prosdocimi, D.; Klima, K. Health Effects of Heat Vulnerability in Rio de Janeiro: A Validation Model for Policy Applications. SN Appl. Sci. 2020 , 2 , 1948. [ Google Scholar ] [ CrossRef ]
• Oh, K.; Lee, M.; Jeon, S. Development of the Korean Climate Change Vulnerability Assessment Tool (VESTAP): Centered on Health Vulnerability to Heat Waves. Sustainability 2017 , 9 , 1103. [ Google Scholar ] [ CrossRef ]
• Dong, J.; Peng, J.; He, X.; Corcoran, J.; Qiu, S.; Wang, X. Heatwave-Induced Human Health Risk Assessment in Megacities Based on Heat Stress-Social Vulnerability-Human Exposure Framework. Landsc. Urban Plan. 2020 , 203 , 103907. [ Google Scholar ] [ CrossRef ]
• Ho, H.; Knudby, A.; Huang, W. A Spatial Framework to Map Heat Health Risks at Multiple Scales. Int. J. Environ. Res. Public Health 2015 , 12 , 16110–16123. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Krstic, N.; Yuchi, W.; Ho, H.; Walker, B.; Knudby, A.; Henderson, S. The Heat Exposure Integrated Deprivation Index (HEIDI): A Data-Driven Approach to Quantifying Neighborhood Risk During Extreme Hot Weather. Environ. Int. 2017 , 109 , 42–52. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Tomlinson, C.; Chapman, L.; Thornes, J.; Baker, C. Including the Urban Heat Island in Spatial Heat Health Risk Assessment Strategies: A Case Study for Birmingham, UK. Int. J. Health Geogr. 2011 , 10 , 42. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Macintyre, H.; Heaviside, C.; Taylor, J.; Picetti, R.; Symonds, P.; Cai, X.; Vardoulakis, S. Assessing Urban Population Vulnerability and Environmental Risks Across an Urban Area During Heatwaves: Implications for Health Protection. Sci. Total Environ. 2017 , 610–611 , 678–690. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Johnson, D.; Stanforth, A.; Lulla, V.; Luber, G. Developing an Applied Extreme Heat Vulnerability Index Utilizing Socioeconomic and Environmental Data. Appl. Geogr. 2012 , 35 , 23–31. [ Google Scholar ] [ CrossRef ]
• Kamal, N.; Ash’aari, Z.; Abdullah, A.; Kusin, F.; Mohamat Yusuff, F.; Sharaai, A.; Mohd Ariffin, N. Extreme Heat Vulnerability Assessment in Tropical Region: A Case Study in Malaysia. Clim. Dev. 2021 , 4 , 472–486. [ Google Scholar ] [ CrossRef ]
• Conlon, K.; Mallen, E.; Gronlund, C.; Berrocal, V.; Larsen, L.; O’Neill, M. Mapping Human Vulnerability to Extreme Heat: A Critical Assessment of Heat Vulnerability Indices Created Using Principal Components Analysis. Environ. Health Perspect. 2020 , 128 , 97001. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Estoque, R.; Ooba, M.; Seposo, X.; Togawa, T.; Hijioka, Y.; Takahashi, K.; Nakamura, S. Heat Health Risk Assessment in Philippine Cities Using Remotely Sensed Data and Social-Ecological Indicators. Nat. Commun. 2020 , 11 , 1581. [ Google Scholar ] [ CrossRef ]
• Chen, N.; Lin, P.; Guo, Y. Long-Term Exposure to High Temperature Associated with the Incidence of Major Depressive Disorder. Sci. Total Environ. 2019 , 659 , 1016–1020. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Yoo, E.; Eum, Y.; Roberts, J.; Gao, Q.; Chen, K. Association Between Extreme Temperatures and Emergency Room Visits Related to Mental Disorders: A Multi-Region Time-Series Study in New York, USA. Sci. Total Environ. 2021 , 792 , 148246. [ Google Scholar ] [ CrossRef ]
• Huang, F.; Yu, Y.; Feng, T. Automatic Extraction of Impervious Surfaces from High Resolution Remote Sensing Images Based on Deep Learning. J. Vis. Commun. Image Represent. 2019 , 58 , 453–461. [ Google Scholar ] [ CrossRef ]
• Wang, P.; Bayram, B.; Sertel, E. A Comprehensive Review on Deep Learning Based Remote Sensing Image Super-Resolution Methods. Earth-Sci. Rev. 2022 , 234 , 104110. [ Google Scholar ] [ CrossRef ]
• Osaku, D.; Nakamura, R.Y.M.; Pereira, L.A.M.; Pisani, R.J.; Levada, A.L.M.; Cappabianco, F.A.M.; Falcão, A.X.; Papa, J.P. Improving Land Cover Classification through Contextual-Based Optimum-Path Forest. Inf. Sci. 2015 , 324 , 60–87. [ Google Scholar ] [ CrossRef ]
• Kafy, A.A.; Saha, M.; Faisal, A.-A.; Rahaman, Z.A.; Rahman, M.T.; Liu, D.; Fattah, M.A.; Al Rakib, A.; AlDousari, A.E.; Rahaman, S.N.; et al. Predicting the Impacts of Land Use/Land Cover Changes on Seasonal Urban Thermal Characteristics Using Machine Learning Algorithms. Build. Environ. 2022 , 217 , 109066. [ Google Scholar ] [ CrossRef ]
• Uniyal, S.; Purohit, S.; Chaurasia, K.; Rao, S.S.; Amminedu, E. Quantification of Carbon Sequestration by Urban Forest Using Landsat 8 OLI and Machine Learning Algorithms in Jodhpur, India. Urban For. Urban Green. 2022 , 67 , 127445. [ Google Scholar ] [ CrossRef ]
• Wu, X.; Ou, J.; Wen, Y.; Liu, X.; He, J.; Zhang, J. Developing a Data-Fusing Method for Mapping Fine-Scale Urban Three-Dimensional Building Structure. Sustain. Cities Soc. 2022 , 80 , 103716. [ Google Scholar ] [ CrossRef ]
• Kim, Y.; Kim, Y. Deep Learning-Based Analytic Framework Using Comprehensive Urbanization Index for Heat Vulnerability Assessment in Urban Areas. Expert Syst. Appl. 2024 , 235 , 121140. [ Google Scholar ] [ CrossRef ]
• Li, F.; Yigitcanlar, T.; Nepal, M.; Nguyen, K.; Dur, F. Machine Learning and Remote Sensing Integration for Leveraging Urban Sustainability: A Review and Framework. Sustain. Cities Soc. 2023 , 96 , 104653. [ Google Scholar ] [ CrossRef ]
• Zhu, Q.; Sun, Y.; Guan, Q.; Wang, L.; Lin, W. A Weakly Pseudo-Supervised Decorrelated Subdomain Adaptation Framework for Cross-Domain Land-Use Classification. IEEE Trans. Geosci. Remote Sens. 2022 , 60 , 1–13. [ Google Scholar ] [ CrossRef ]
• Jamali, A. Evaluation and Comparison of Eight Machine Learning Models in Land Use/Land Cover Mapping Using Landsat 8 OLI: A Case Study of the Northern Region of Iran. SN Appl. Sci. 2019 , 1 , 1527. [ Google Scholar ] [ CrossRef ]
• Frantz, D.; Schug, F.; Okujeni, A.; Navacchi, C.; Wagner, W.; van der Linden, S.; Hostert, P. National-Scale Mapping of Building Height Using Sentinel-1 and Sentinel-2 Time Series. Remote Sens. Environ. 2021 , 252 , 112128. [ Google Scholar ] [ CrossRef ]
• Guo, H.; Huang, J.J.; Chen, B.; Guo, X.; Singh, V.P. A Machine Learning-Based Strategy for Estimating Non-Optically Active Water Quality Parameters Using Sentinel-2 Imagery. Int. J. Remote Sens. 2021 , 42 , 1841–1866. [ Google Scholar ] [ CrossRef ]
• Jato-Espino, D.; Manchado, C.; Roldán-Valcarce, A.; Moscardó, V. ArcUHI: A GIS Add-In for Automated Modelling of the Urban Heat Island Effect through Machine Learning. Urban Clim. 2022 , 44 , 101203. [ Google Scholar ] [ CrossRef ]
• Abunnasr, Y.; Mhawej, M. Towards a Combined Landsat-8 and Sentinel-2 for 10-m Land Surface Temperature Products: The Google Earth Engine Monthly Ten-ST-GEE System. Environ. Model. Softw. 2022 , 155 , 105456. [ Google Scholar ] [ CrossRef ]
• Casali, Y.; Aydin, N.Y.; Comes, T. Machine Learning for Spatial Analyses in Urban Areas: A Scoping Review. Sustain. Cities Soc. 2022 , 85 , 104050. [ Google Scholar ] [ CrossRef ]
• Demarchi, L.; Canters, F.; Cariou, C.; Licciardi, G.; Chan, J.C.-W. Assessing the Performance of Two Unsupervised Dimensionality Reduction Techniques on Hyperspectral APEX Data for High Resolution Urban Land-Cover Mapping. ISPRS J. Photogramm. Remote Sens. 2014 , 87 , 166–179. [ Google Scholar ] [ CrossRef ]
• Deng, C.; Wu, C. The Use of Single-Date MODIS Imagery for Estimating Large-Scale Urban Impervious Surface Fraction with Spectral Mixture Analysis and Machine Learning Techniques. ISPRS J. Photogramm. Remote Sens. 2013 , 86 , 100–110. [ Google Scholar ] [ CrossRef ]
• Priem, F.; Okujeni, A.; van der Linden, S.; Canters, F. Comparing Map-Based and Library-Based Training Approaches for Urban Land-Cover Fraction Mapping from Sentinel-2 Imagery. Int. J. Appl. Earth Obs. Geoinf. 2019 , 78 , 295–305. [ Google Scholar ] [ CrossRef ]
• Yigitcanlar, T. Australian Local Governments’ Practice and Prospects with Online Planning. URISA J. 2006 , 18 , 7–17. [ Google Scholar ]
• Patino, J.E.; Duque, J.C. A Review of Regional Science Applications of Satellite Remote Sensing in Urban Settings. Comput. Environ. Urban Syst. 2013 , 37 , 1–17. [ Google Scholar ] [ CrossRef ]
• Galaz, V.; Centeno, M.A.; Callahan, P.W.; Causevic, A.; Patterson, T.; Brass, I.; Baum, S.; Farber, D.; Fischer, J.; Garcia, D.; et al. Artificial Intelligence, Systemic Risks, and Sustainability. Technol. Soc. 2021 , 67 , 101741. [ Google Scholar ] [ CrossRef ]
• United Nations. Global Indicator Framework for the Sustainable Development Goals and Targets of the 2030 Agenda for Sustainable Development ; United Nations: San Francisco, CA, USA, 2023. [ Google Scholar ]
• Touzani, S.; Granderson, J. Open Data and Deep Semantic Segmentation for Automated Extraction of Building Footprints. Remote Sens. 2021 , 13 , 2578. [ Google Scholar ] [ CrossRef ]
• Chi, Y.; Fan, M.; Zhao, C.; Yang, Y.; Fan, H.; Yang, X.; Yang, J.; Tao, J. Machine Learning-Based Estimation of Ground-Level NO 2 Concentrations over China. Sci. Total Environ. 2022 , 807 , 150721. [ Google Scholar ] [ CrossRef ]
• Peng, B.; Meng, Z.; Huang, Q.; Wang, C. Patch Similarity Convolutional Neural Network for Urban Flood Extent Mapping Using Bi-Temporal Satellite Multispectral Imagery. Remote Sens. 2019 , 11 , 2492. [ Google Scholar ] [ CrossRef ]
• Ghaffarian, S.; Roy, D.; Filatova, T.; Kerle, N. Agent-Based Modelling of Post-Disaster Recovery with Remote Sensing Data. Int. J. Disaster Risk Reduct. 2021 , 60 , 102285. [ Google Scholar ] [ CrossRef ]
• Zhang, P.; Liu, L.; Yang, L.; Zhao, J.; Li, Y.; Qi, Y.; Ma, X.; Cao, L. Exploring the Response of Ecosystem Service Value to Land Use Changes under Multiple Scenarios Coupling a Mixed-Cell Cellular Automata Model and System Dynamics Model in Xi’an, China. Ecol. Indic. 2023 , 147 , 110009. [ Google Scholar ] [ CrossRef ]
• Li, F.; Yigitcanlar, T.; Li, W.; Nepal, M.; Nguyen, K.; Dur, F. Understanding urban heat vulnerability: Scientometric analysis of five decades of research. Urban Clim. 2024 , 56 , 102035. [ Google Scholar ] [ CrossRef ]
• Yigitcanlar, T. Sustainable Urban and Regional Infrastructure Development: Technologies, Applications and Management ; IGI Global: Hersey, PA, USA, 2010. [ Google Scholar ]
• Koumetio Tekouabou, S.C.; Diop, E.B.; Azmi, R.; Jaligot, R.; Chenal, J. Reviewing the Application of Machine Learning Methods to Model Urban Form Indicators in Planning Decision Support Systems: Potential, Issues and Challenges. J. King Saud Univ.-Comput. Inf. Sci. 2022 , 34 , 5943–5967. [ Google Scholar ] [ CrossRef ]
• Karanja, J.; Kiage, L. Perspectives on Spatial Representation of Urban Heat Vulnerability. Sci. Total Environ. 2021 , 774 , 145634. [ Google Scholar ] [ CrossRef ]
• Intergovernmental Panel On Climate Change. Climate Change 2007: Impacts, Adaptation and Vulnerability ; Intergovernmental Panel On Climate Change: Genebra, Switzerland, 2001. [ Google Scholar ]
• Rinner, C.; Patychuk, D.; Bassil, K.; Nasr, S.; Gower, S.; Campbell, M. The Role of Maps in Neighborhood-Level Heat Vulnerability Assessment for the City of Toronto. Cartogr. Geogr. Inf. Sci. 2010 , 37 , 31–44. [ Google Scholar ] [ CrossRef ]
• Depietri, Y.; Welle, T.; Renaud, F. Social Vulnerability Assessment of the Cologne Urban Area (Germany) to Heat Waves: Links to Ecosystem Services. Int. J. Disaster Risk Reduct. 2013 , 6 , 98–117. [ Google Scholar ] [ CrossRef ]
• Hulley, G.; Shivers, S.; Wetherley, E.; Cudd, R. New ECOSTRESS and MODIS Land Surface Temperature Data Reveal Fine-Scale Heat Vulnerability in Cities: A Case Study for Los Angeles County, California. Remote Sens. 2019 , 11 , 2136. [ Google Scholar ] [ CrossRef ]
• Crichton, D. The Risk Triangle. In Natural Disaster Management ; Ingleton, J., Ed.; Tudor Rose: London, UK, 1999; pp. 102–103. ISBN 0-9536140-1-8. [ Google Scholar ]
• Buscail, C.; Upegui, E.; Viel, J. Mapping Heatwave Health Risk at the Community Level for Public Health Action. Int. J. Health Geogr. 2012 , 11 , 38. [ Google Scholar ] [ CrossRef ]
• Dugord, P.; Lauf, S.; Schuster, C.; Kleinschmit, B. Land Use Patterns, Temperature Distribution, and Potential Heat Stress Risk: The Case Study Berlin, Germany. Comput. Environ. Urban Syst. 2014 , 48 , 86–98. [ Google Scholar ] [ CrossRef ]
• Paranunzio, R.; Dwyer, E.; Fitton, J.; Alexander, P.; O’Dwyer, B. Assessing Current and Future Heat Risk in Dublin City, Ireland. Urban Clim. 2021 , 40 , 100983. [ Google Scholar ] [ CrossRef ]
• Luber, G.; McGeehin, M. Climate Change and Extreme Heat Events. Am. J. Prev. Med. 2008 , 35 , 429–435. [ Google Scholar ] [ CrossRef ]
• Cheng, W.; Brown, R. An Energy Budget Model for Estimating the Thermal Comfort of Children. Int. J. Biometeorol. 2020 , 64 , 1355–1366. [ Google Scholar ] [ CrossRef ]
• Alonso, L.; Renard, F. A Comparative Study of the Physiological and Socio-Economic Vulnerabilities to Heat Waves of the Population of the Metropolis of Lyon (France) in a Climate Change Context. Int. J. Environ. Res. Public Health 2020 , 17 , 1004. [ Google Scholar ] [ CrossRef ]
• Hu, K.; Yang, X.; Zhong, J.; Fei, F.; Qi, J. Spatially Explicit Mapping of Heat Health Risk Utilizing Environmental and Socioeconomic Data. Environ. Sci. Technol. 2017 , 51 , 1498–1507. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Kwon, Y.; Lee, D.; Kwon, Y. Is Sensible Heat Flux Useful for the Assessment of Thermal Vulnerability in Seoul (Korea)? Int. J. Environ. Res. Public Health 2020 , 17 , 963. [ Google Scholar ] [ CrossRef ]
• Reid, C.; Mann, J.; Alfasso, R.; English, P.; King, G.; Lincoln, R.; Balmes, J. Evaluation of a Heat Vulnerability Index on Abnormally Hot Days: An Environmental Public Health Tracking Study. Environ. Health Perspect. 2012 , 120 , 715–720. [ Google Scholar ] [ CrossRef ]
• Xiao, Y.; Wang, Y.; Yuan, Q.; He, J.; Zhang, L. Generating a Long-Term (2003−2020) Hourly 0.25° Global PM2.5 Dataset via Spatiotemporal Downscaling of CAMS with Deep Learning (DeepCAMS). Sci. Total Environ. 2022 , 848 , 157747. [ Google Scholar ] [ CrossRef ]
• Loughnan, M.; Nicholls, N.; Tapper, N. Mapping Heat Health Risks in Urban Areas. Int. J. Popul. Res. 2012 , 2012 , 518687. [ Google Scholar ] [ CrossRef ]
• Hammer, J.; Ruggieri, D.; Thomas, C.; Caum, J. Local Extreme Heat Planning: An Interactive Tool to Examine a Heat Vulnerability Index for Philadelphia, Pennsylvania. J. Urban Health 2020 , 97 , 519–528. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Lammers, R.; Li, A.; Nag, S.; Ravindra, V. Prediction Models for Urban Flood Evolution for Satellite Remote Sensing. J. Hydrol. 2021 , 603 , 127175. [ Google Scholar ] [ CrossRef ]
• Wei, G.; Zhang, Z.; Ouyang, X.; Shen, Y.; Jiang, S.; Liu, B.; He, B.-J. Delineating the Spatial-Temporal Variation of Air Pollution with Urbanization in the Belt and Road Initiative Area. Environ. Impact Assess. Rev. 2021 , 91 , 106646. [ Google Scholar ] [ CrossRef ]
• Ronneberger, O.; Fischer, P.; Brox, T. U-Net: Convolutional Networks for Biomedical Image Segmentation. Lect. Notes Comput. Sci. 2015 , 9351 , 234–241. [ Google Scholar ] [ CrossRef ]
• Yan, C.; Fan, X.; Fan, J.; Wang, N. Improved U-Net Remote Sensing Classification Algorithm Based on Multi-Feature Fusion Perception. Remote Sens. 2022 , 14 , 1118. [ Google Scholar ] [ CrossRef ]
• Li, Z.; Demir, I. U-Net-Based Semantic Classification for Flood Extent Extraction Using SAR Imagery and GEE Platform: A Case Study for 2019 Central US Flooding. Sci. Total Environ. 2023 , 869 , 161757. [ Google Scholar ] [ CrossRef ]
• Yigitcanlar, T. Rethinking Sustainable Development: Urban Management, Engineering, and Design ; IGI Global: Hersey, PA, USA, 2010. [ Google Scholar ]
• Stehman, S.V.; Hansen, M.C.; Broich, M.; Potapov, P.V. Adapting a Global Stratified Random Sample for Regional Estimation of Forest Cover Change Derived from Satellite Imagery. Remote Sens. Environ. 2011 , 115 , 650–658. [ Google Scholar ] [ CrossRef ]
• Amani, S.; Shafizadeh-Moghadam, H. A Review of Machine Learning Models and Influential Factors for Estimating Evapotranspiration Using Remote Sensing and Ground-Based Data. Agric. Water Manag. 2023 , 284 , 108324. [ Google Scholar ] [ CrossRef ]
• Wang, S.; Sun, Q.C.; Huang, X.; Tao, Y.; Dong, C.; Das, S.; Liu, Y. Health-Integrated Heat Risk Assessment in Australian Cities. Environ. Impact Assess. Rev. 2023 , 102 , 107176. [ Google Scholar ] [ CrossRef ]
• Vu, H.L.; Ng, K.T.W.; Richter, A.; An, C. Analysis of Input Set Characteristics and Variances on k-Fold Cross Validation for a Recurrent Neural Network Model on Waste Disposal Rate Estimation. J. Environ. Manag. 2022 , 311 , 114869. [ Google Scholar ] [ CrossRef ]
• Azhar, G.; Saha, S.; Ganguly, P.; Mavalankar, D.; Madrigano, J. Heat Wave Vulnerability Mapping for India. Int. J. Environ. Res. Public Health 2017 , 14 , 357. [ Google Scholar ] [ CrossRef ]
• Guo, X.; Huang, G.; Jia, P.; Wu, J. Estimating Fine-Scale Heat Vulnerability in Beijing through Two Approaches: Spatial Patterns, Similarities, and Divergence. Remote Sens. 2019 , 11 , 2358. [ Google Scholar ] [ CrossRef ]
• Shang, W.L.; Chen, Y.; Yu, Q.; Song, X.; Chen, Y.; Ma, X.; Chen, X.; Tan, Z.; Huang, J.; Ochieng, W. Spatio-Temporal Analysis of Carbon Footprints for Urban Public Transport Systems Based on Smart Card Data. Appl. Energy 2023 , 352 , 121859. [ Google Scholar ] [ CrossRef ]
• Matthews, T.; Lo, A.Y.; Byrne, J.A. Reconceptualizing Green Infrastructure for Climate Change Adaptation: Barriers to Adoption and Drivers for Uptake by Spatial Planners. Landsc. Urban Plan. 2015 , 138 , 155–163. [ Google Scholar ] [ CrossRef ]
• Mitchell, B.; Chakraborty, J. Landscapes of Thermal Inequity: Disproportionate Exposure to Urban Heat in the Three Largest US Cities. Environ. Res. Lett. 2015 , 10 , 115005. [ Google Scholar ] [ CrossRef ]
• Ferreira, F.; Pereira, E.; Gonçalves, A.; Costa, R.; Bezerra, F. An Explicitly Spatial Approach to Identify Heat Vulnerable Urban Areas and Landscape Patterns. Urban Clim. 2021 , 40 , 101021. [ Google Scholar ] [ CrossRef ]
• Wieland, M.; Pittore, M. Large-Area Settlement Pattern Recognition from Landsat-8 Data. ISPRS J. Photogramm. Remote Sens. 2016 , 119 , 294–308. [ Google Scholar ] [ CrossRef ]
• Das, T.; Jana, A.; Mandal, B.; Sutradhar, A. Spatio-Temporal Pattern of Land Use and Land Cover and Its Effects on Land Surface Temperature Using Remote Sensing and GIS Techniques: A Case Study of Bhubaneswar City, Eastern India (1991–2021). GeoJournal 2021 , 87 , 765–795. [ Google Scholar ] [ CrossRef ]
• Tao, C.; Jia, M.; Wang, G.; Zhang, Y.; Zhang, Q.; Wang, X.; Wang, Q.; Wang, W. Time-Sensitive Prediction of NO 2 Concentration in China Using an Ensemble Machine Learning Model from Multi-Source Data. J. Environ. Sci. 2023 , 137 , 30–40. [ Google Scholar ] [ CrossRef ]
• Tella, A.; Balogun, A.-L.; Adebisi, N.; Abdullah, S. Spatial Assessment of PM10 Hotspots Using Random Forest, K-Nearest Neighbour and Naïve Bayes. Atmos. Pollut. Res. 2021 , 12 , 101202. [ Google Scholar ] [ CrossRef ]
• Al-Aizari, A.R.; Al-Masnay, Y.A.; Aydda, A.; Zhang, J.; Ullah, K.; Islam, A.R.M.T.; Habib, T.; Kaku, D.U.; Nizeyimana, J.C.; Al-Shaibah, B.; et al. Assessment Analysis of Flood Susceptibility in Tropical Desert Area: A Case Study of Yemen. Remote Sens. 2022 , 14 , 4050. [ Google Scholar ] [ CrossRef ]
• Singleton, A.; Arribas-Bel, D.; Murray, J.; Fleischmann, M. Estimating Generalized Measures of Local Neighbourhood Context from Multispectral Satellite Images Using a Convolutional Neural Network. Comput. Environ. Urban Syst. 2022 , 95 , 101802. [ Google Scholar ] [ CrossRef ]
• Papadomanolaki, M.; Vakalopoulou, M.; Karantzalos, K. A Deep Multitask Learning Framework Coupling Semantic Segmentation and Fully Convolutional LSTM Networks for Urban Change Detection. IEEE Trans. Geosci. Remote Sens. 2021 , 59 , 7651–7668. [ Google Scholar ] [ CrossRef ]
• Zhu, Y.; Geiß, C.; So, E.; Jin, Y. Multitemporal Relearning with Convolutional LSTM Models for Land Use Classification. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021 , 14 , 3251–3265. [ Google Scholar ] [ CrossRef ]
• He, Z.; Chow, C.Y.; Zhang, J.D. STCNN: A Spatio-Temporal Convolutional Neural Network for Long-Term Traffic Prediction. In Proceedings of the 2019 20th IEEE International Conference on Mobile Data Management (MDM), Hong Kong, China, 10–13 June 2019; pp. 226–233. [ Google Scholar ] [ CrossRef ]
• Xiao, Z.; Li, H.; Jiang, H.; Li, Y.; Alazab, M.; Zhu, Y.; Dustdar, S. Predicting Urban Region Heat via Learning Arrive-Stay-Leave Behaviors of Private Cars. IEEE Trans. Intell. Transp. Syst. 2023 , 24 , 10843–10856. [ Google Scholar ] [ CrossRef ]
• Yu, B.; Yin, H.; Zhu, Z. Spatio-temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting. arXiv 2017 , arXiv:1709.04875. [ Google Scholar ] [ CrossRef ]
• Xu, Z.; Li, J.; Lv, Z.; Wang, Y.; Li, H. A Prediction Method for Population Density in Key Areas. In Proceedings of the International Symposium on Artificial Intelligence and Robotics 2021, Fukuoka, Japan, 28 October 2021; SPIE: Bellingham, WA, USA; Volume 11884, pp. 13–28. [ Google Scholar ] [ CrossRef ]
• Kim, Y.; Newman, G. Advancing Scenario Planning through Integrating Urban Growth Prediction with Future Flood Risk Models. Comput. Environ. Urban Syst. 2020 , 82 , 101498. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ma, Y.; Hu, Y.; Moncrieff, G.R.; Slingsby, J.A.; Wilson, A.M.; Maitner, B.; Zhou, R.Z. Forecasting Vegetation Dynamics in an Open Ecosystem by Integrating Deep Learning and Environmental Variables. Int. J. Appl. Earth Obs. Geoinf. 2022 , 114 , 103060. [ Google Scholar ] [ CrossRef ]
• Dehghani, A.; Moazam, H.M.Z.H.; Mortazavizadeh, F.; Ranjbar, V.; Mirzaei, M.; Mortezavi, S.; Ng, J.L.; Dehghani, A. Comparative Evaluation of LSTM, CNN, and ConvLSTM for Hourly Short-Term Streamflow Forecasting Using Deep Learning Approaches. Ecol. Inform. 2023 , 75 , 102119. [ Google Scholar ] [ CrossRef ]
• Maier, G.; Grundstein, A.; Jang, W.; Li, C.; Naeher, L.; Shepherd, M. Assessing the Performance of a Vulnerability Index during Oppressive Heat across Georgia, United States. Weather. Clim. Soc. 2014 , 6 , 253–263. [ Google Scholar ] [ CrossRef ]
• Mushore, T.; Mutanga, O.; Odindi, J.; Dube, T. Determining Extreme Heat Vulnerability of Harare Metropolitan City Using Multispectral Remote Sensing and Socio-Economic Data. J. Spat. Sci. 2017 , 63 , 173–191. [ Google Scholar ] [ CrossRef ]
• Khan, A.; Sudheer, M. Machine Learning-Based Monitoring and Modeling for Spatio-Temporal Urban Growth of Islamabad. Egypt. J. Remote Sens. Space Sci. 2022 , 25 , 541–550. [ Google Scholar ] [ CrossRef ]
• Tuczek, M.; Degirmenci, K.; Desouza, K.; Watson, R.; Yigitcanlar, T.; Breitner, M. Mitigating urban heat with optimal distribution of vegetation and buildings. Urban Clim. 2022 , 44 , 101208. [ Google Scholar ] [ CrossRef ]
• Liu, Y.; Yigitcanlar, T.; Guaralda, M.; Degirmenci, K.; Liu, A. Spatial Modelling of Urban Wind Characteristics: Review of Contributions to Sustainable Urban Development. Buildings 2024 , 14 , 737. [ Google Scholar ] [ CrossRef ]
• Liu, Y.; Guaralda, M.; Yigitcanlar, T.; Limb, M.; Garcia-Hansen, V. Navigating urban climate design implementation challenges: Insights from Brisbane’s built environment experts. J. Urban Des. 2024 , 1–20. [ Google Scholar ] [ CrossRef ]
• Guo, F.; Zheng, R.; Zhao, J.; Zhang, H.; Dong, J. Framework of street grid-based urban heat vulnerability assessment: Integrating entropy weight method and BPNN model. Urban Clim. 2024 , 56 , 102067. [ Google Scholar ] [ CrossRef ]
• Kitchley, J.L. A framework to assess the contextual composite heat vulnerability index for a heritage city in India-A case study of Madurai. Sustain. Cities Soc. 2024 , 101 , 105119. [ Google Scholar ] [ CrossRef ]
• Wu, H.; Xu, Y.; Zhang, M.; Su, L.; Wang, Y.; Zhu, S. Spatially explicit assessment of the heat-related health risk in the Yangtze River Delta, China, using multisource remote sensing and socioeconomic data. Sustain. Cities Soc. 2024 , 104 , 105300. [ Google Scholar ] [ CrossRef ]
• Wang, H.; Yang, J.; Chen, G.; Ren, C.; Zhang, J. Machine Learning Applications on Air Temperature Prediction in the Urban Canopy Layer: A Critical Review of 2011–2022. Urban Clim. 2023 , 49 , 101499. [ Google Scholar ] [ CrossRef ]

Click here to enlarge figure

Share and Cite

Li, F.; Yigitcanlar, T.; Nepal, M.; Thanh, K.N.; Dur, F. A Novel Urban Heat Vulnerability Analysis: Integrating Machine Learning and Remote Sensing for Enhanced Insights. Remote Sens. 2024 , 16 , 3032. https://doi.org/10.3390/rs16163032

Li F, Yigitcanlar T, Nepal M, Thanh KN, Dur F. A Novel Urban Heat Vulnerability Analysis: Integrating Machine Learning and Remote Sensing for Enhanced Insights. Remote Sensing . 2024; 16(16):3032. https://doi.org/10.3390/rs16163032

Li, Fei, Tan Yigitcanlar, Madhav Nepal, Kien Nguyen Thanh, and Fatih Dur. 2024. "A Novel Urban Heat Vulnerability Analysis: Integrating Machine Learning and Remote Sensing for Enhanced Insights" Remote Sensing 16, no. 16: 3032. https://doi.org/10.3390/rs16163032

Article Metrics

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

Subscribe to receive issue release notifications and newsletters from MDPI journals

Paper Highlights How Climate Change Challenges, Transforms Agriculture

As the climate continues to change, the risks to farming are only going to increase.

That's the key takeaway from a recent paper published by a team that included UC Merced researchers. The paper dives into what those challenges are, how farmers are working to address them and what should come next.

"Climate Smart Agriculture: Assessing Needs and Perceptions of California's Farmers" was first authored by Samuel Ikendi, academic coordinator, with engineering research Professor Tapan Pathak  as a corresponding author. Pathak is also a project director of National Institute of Food and Agriculture-funded project "Multifaceted Pathways to Climate-Smart Agriculture through Participator Program Development and Delivery," which supported this study. The study appeared in the open access journal Frontiers in Sustainable Food Systems .

The needs assessment was designed to understand farmers' perceptions and experiences with climate change exposures; the risk management practices they currently use; and what tools and resources would assist them in making strategic decisions.

Of the farmers surveyed, roughly two-thirds agree climate change is occurring and requires action. Even more said they are interested in learning more about the impacts of climate change on the agricultural industry. Most respondents said they experience greater climate change impacts on their farms today compared with 10 years ago.

Farmers were very concerned with water-related issues, with those in the San Joaquin Valley, Central Coast and Inland Empire areas particularly worried about a reduction in the availability of groundwater. Increased drought severity was a very significant concern among farmers in the Inland Empire, Central Coast and Southern regions. Farmers in the North Coast and Southern regions were concerned about increased damage to crops due to wildfire.

Closely related were temperature-related issues, including crop damage due to extreme heat.

Those who farm vegetables were more concerned about water availability for irrigation, while fruit farmers were more concerned about increased crop/water stress and increased crop damage due to extreme heat.

Many respondents said they are implementing climate adaption practices including managing water resources, maintaining soil health and making more use of renewable energy sources. They are seeking insurance and government help to pay for these adaptations and increase their agricultural resilience, the researchers wrote.

The farmers expressed interest in learning more about measures they might take to mitigate climate change. But they cited significant barriers to this work, including government regulations, high implementation cost, labor access/cost, access to water and the availability of money to pay for it.

"Climate change is significantly altering California's highly diverse agricultural landscape, posing challenges such as increased water stress, heat stress, and shifting growing seasons," Pathak said. "Climate-smart agriculture practices can alleviate some of those stresses."

But, he said, research and UC Extension efforts only have value if they lead to enhanced climate-informed decision-making at the local level.

"Assessing their level of knowledge, perception and needs will help in tailoring research and extension activities that are most relevant to farmers on the ground," Pathak said. "Results from this study could also provide important policy insights on financial incentives and technical assistance."

Patty Guerra

Media Contact

Office: (209) 769-0948

[email protected]

Video Spotlight

Researchers Accurately Model Animals’ Hunting, Scavenging Behavior

A group of UC Merced researchers modeled predation behaviors, as well as changes in those behaviors, among large carnivores, developing a new theory that...

Psychological Bias Links Good Deeds with Religious Belief, UC Merced Research Says

UC Merced research finds that people who perform good deeds are far more likely to be thought of as religious believers than atheists. Further, this psychological bias appears to be global in scale.

Researcher's Work Aimed at Helping People Live Better Lives

"I like research on the Internet of Things because it solves problems in people's lives," said Shijia Pan , a professor of computer...

Additional Links

• Executive Leadership
• University Library
• School of Engineering
• School of Natural Sciences
• School of Social Sciences, Humanities & Arts
• Ernest & Julio Gallo Management Program
• Division of Graduate Education
• Division of Undergraduate Education

Administration

• Office of the Chancellor
• Office of Executive Vice Chancellor and Provost
• Equity, Justice and Inclusive Excellence
• External Relations
• Finance & Administration
• Physical Operations, Planning and Development
• Student Affairs
• Research and Economic Development
• Office of Information Technology

University of California, Merced 5200 North Lake Rd. Merced, CA 95343 Telephone: (209) 228-4400

• © 2024
• About UC Merced
• Privacy/Legal
• Site Feedback
• Accessibility

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: 19 November 2020

Predicting global patterns of long-term climate change from short-term simulations using machine learning

• L. A. Mansfield   ORCID: orcid.org/0000-0002-6285-6045 1 , 2 ,
• P. J. Nowack   ORCID: orcid.org/0000-0003-4588-7832 1 , 3 , 4 , 5 ,
• M. Kasoar   ORCID: orcid.org/0000-0001-5571-8843 1 , 3 , 6 ,
• R. G. Everitt   ORCID: orcid.org/0000-0002-0822-5648 7 ,
• W. J. Collins   ORCID: orcid.org/0000-0002-7419-0850 8 &
• A. Voulgarakis   ORCID: orcid.org/0000-0002-6656-4437 1 , 6 , 9  

npj Climate and Atmospheric Science volume  3 , Article number:  44 ( 2020 ) Cite this article

17k Accesses

36 Citations

25 Altmetric

Metrics details

• Atmospheric science
• Climate change
• Climate-change impacts
• Climate-change mitigation
• Projection and prediction

Understanding and estimating regional climate change under different anthropogenic emission scenarios is pivotal for informing societal adaptation and mitigation measures. However, the high computational complexity of state-of-the-art climate models remains a central bottleneck in this endeavour. Here we introduce a machine learning approach, which utilises a unique dataset of existing climate model simulations to learn relationships between short-term and long-term temperature responses to different climate forcing scenarios. This approach not only has the potential to accelerate climate change projections by reducing the costs of scenario computations, but also helps uncover early indicators of modelled long-term climate responses, which is of relevance to climate change detection, predictability, and attribution. Our results highlight challenges and opportunities for data-driven climate modelling, especially concerning the incorporation of even larger model datasets in the future. We therefore encourage extensive data sharing among research institutes to build ever more powerful climate response emulators, and thus to enable faster climate change projections.

Similar content being viewed by others

A Large Ensemble Global Dataset for Climate Impact Assessments

Training machine learning models on climate model output yields skillful interpretable seasonal precipitation forecasts

A machine learning approach to rapidly project climate responses under a multitude of net-zero emission pathways

Introduction.

To achieve long-term climate change mitigation and adaptation goals, such as limiting global warming to 1.5 or 2 °C, there must be a global effort to decide and act upon effective but realistic emission pathways 1 . This requires an understanding of the consequences of such pathways, which are often diverse and involve changes in multiple climate forcers 1 , 2 , 3 . In particular, different emission scenarios of, for example, greenhouse gases and aerosols are responsible for diverse changes in regional climate, which are not always well captured by a metric such as global temperature-change potential 4 , 5 , 6 , 7 , 8 , 9 . Exploring more detailed relationships between emissions and multiregional climate responses still requires the application of Global Climate Models (GCMs) that allow the behaviour of the climate to be simulated under various conditions (e.g. different atmospheric greenhouse gas and aerosol concentrations or emissions fields) 10 , 11 , 12 on decadal to multi-centennial timescales (e.g. refs. 5 , 13 , 14 , 15 , 16 ). However, modelling climate at increasingly high spatial resolutions has significantly increased the computational complexity of GCMs 2 , a tendency that has been accelerated by the incorporation and enhancement of a number of new Earth system model components and processes 17 , 18 , 19 , 20 . This high computational cost has driven us to investigate how machine learning methods can help accelerate estimates of global and regional climate change under different climate forcing scenarios.

Our work is further motivated by studies that have suggested links between characteristic short-term and long-term response patterns to different climate forcing agents 5 , 21 , 22 . Here, we seek a fast ‘surrogate model’ 23 to find a mapping from short-term to long-term response patterns within a given GCM (Fig. 1 ). Once learned, this surrogate model can be used to rapidly predict other outputs (long-term responses) given new unseen inputs (short-term responses i.e. the results of easier to perform short-term simulations). While data science methods are increasingly used within climate science (e.g. refs. 24 , 25 , 26 , 27 , 28 , 29 , 30 ), no study has attempted the application we present here, i.e. to predict the magnitude and patterns of long-term climate response to a wide range of global and regional forcing scenarios.

a Global mean surface temperature response of a GCM (HadGEM3) to selected global and regional sudden step perturbations, e.g. to changes in long-lived greenhouse gases (CO 2 , CH 4 ), the solar constant and short-lived aerosols (SO 4 , BC). b Example of the short-term and long-term surface temperature response patterns for 2xCO 2 scenario, defined as an average over the first 10 years and years 70–100, respectively. c Process diagram highlighting the training and prediction stages. In the training stage, a regression function is learned for pairs of short-term and long-term response maps, where the data are obtained from existing HadGEM3 simulations. In the prediction stage, the long-term response for a new unseen scenario is predicted by applying the already learned function to the short-term response to this new scenario, which is cheaper to obtain (here only 10 climate model years).

Building surrogate climate models

To train our learning algorithms, we take advantage of a unique set of GCM simulations performed in recent years using the Hadley Centre Global Environment Model 3 (HadGEM3). In these, step-wise perturbations were applied to various forcing agents to explore characteristic short- and long-term climate responses to them 5 , 7 , 8 , 14 , 16 , 31 , 32 , 33 , 34 . The set of simulations includes global perturbations of long-lived greenhouse gases such as carbon dioxide (CO 2 ) and methane (CH 4 ), as well as global and local perturbations to key short-lived pollutants such as sulfate (SO 4 ) and black carbon (BC) particles, amongst others (Supplementary Table 1 ). A key difference between these two types of perturbations is that long-lived forcers are homogeneously distributed in the atmosphere so that the region of emission is effectively inconsequential for the global temperature response pattern. In contrast, the response pattern does depend on the region of emission for short-lived forcers.

The evolution of the GCM’s global mean temperature response to some example forcing scenarios is highlighted in Fig. 1a . All scenarios show an initial sudden response in the first few years, which we label the ‘short-term response’. The global mean temperature then converges towards a new (approximately) equilibrated steady state, which we label the ‘long-term response’. We are interested in not just the global mean response but, more importantly, in the global response patterns, such as the example shown in Fig. 1b for the 2xCO 2 scenario.

In essence, GCMs map the initial state of the climate system and its boundary conditions, such as emission fields, to a state of the climate at a later time, using complicated functions representing the model physics, chemistry, and biology 17 . Our statistical model approximates the behaviour of the full GCM for a specific target climate variable of interest; here we choose surface temperature at each GCM grid cell, a central variable of interest in climate science and impact studies. This model is trained on simulations from the full global climate model (supervised learning 35 ), in order to predict the long-term surface temperature response of the GCM from the short-term temperature responses to perturbations (Fig. 1c ). Then we can make effectively instantaneous predictions using results from new short-term simulations as input so that repeated long GCM runs can be avoided. Based on the available GCM data, we define the ‘long-term’ as the quasi-equilibrium response after removing the initial transient response (first 70 years) and averaging over the remaining years of the simulations, similarly to previous studies (see Methods) 5 , 14 , 36 . We define ‘short-term’ as the response over the first 10 years of each simulation.

The task is to learn the function \(f({\mathbf{x}})\) that maps these short-term responses ( \({\mathbf{x}}\) ) to the long-term responses ( \({\mathbf{y}}\) ) (‘TRAINING’ in Fig. 1c ). We use an independent regression model of the long-term response for each grid cell. Each one depends on the short-term response at all grid cells, so that predictions are not only based on local information but can also draw predictive capability from any changes in surface temperature worldwide. We present Ridge regression 37 and Gaussian Process Regression (GPR) 38 with a linear kernel (see Methods) as approaches for constructing this mapping. Then, the learned regression functions can be used to predict the long-term response for new, unseen inputs ( \({\mathbf{x}}^ \ast\) ), (‘PREDICTION’ in Fig. 1c ). We choose Ridge regression and GPR, because these two methods handle well the limited sample size (number of simulations available) for training, which also limits how effectively the number of free parameters for other approaches such as deep learning, including convolutional neural networks, could be constrained. Future data collaborations, discussed below, could make the adaptation of our methodology to incorporate deep leaning an option. For the learning process, we use all but one of the available simulations at a time for training and cross-validation. The trained model is then used to make a temperature response prediction for the simulation that was left out each time. Finally, we assess the prediction skill of our machine learning models by comparing the predicted response maps \(f({\mathbf{x}}^ \ast )\) to the results of the complex GCM simulations. This is repeated so that each simulation is predicted once based on the information learned from all other independent simulations (Methods).

Results and discussion

Overall method performance.

We evaluate the performance of the two different machine learning methods (Ridge, GPR) by benchmarking them against a traditional pattern scaling approach 36 , 39 , often used for estimating future patterns of climate change 40 , 41 , 42 . The latter relies on multiplying the long-term response pattern for the 2xCO 2 scenario by the relative magnitude of global mean response for each individual climate forcer. This is approximated as the ratio of global mean effective radiative forcing (ERF) between the forcer and the 2xCO 2 scenario (Methods) 36 . Alternative approaches are discussed in Methods and Supplementary.

We compare the predictions of long-term regional surface temperature changes with those produced by the complex GCM. From analysis at a grid-cell level, both Ridge regression and GPR capture some broad features that pattern scaling is also known to predict effectively, such as enhanced warming over the Northern Hemisphere, particularly over land, and Arctic amplification 43 (Supplementary Figs. 1 and 2 ). However, the key advantage of both machine learning methods is that they capture regional patterns and diversity in the response not predicted by pattern scaling. In particular, aerosol forcing scenarios show highly specific regional imprints on surface temperature due to the spatial heterogeneity of the emissions and their short lifetimes 4 , 7 , 33 . It is the ability to learn these patterns that gives data-driven methods the edge over any pattern scaling method for such predictions. The example in Fig. 2 shows the distribution of predicted temperature responses over all individual grid boxes for one short-lived and one long-lived forcing scenario. For the long-lived forcings all three types of model predictions produce a similar distribution of surface temperature responses to the GCM. However, for short-lived forcing scenarios, the range and variability of responses is highly underestimated in the case of pattern scaling. This is consistent across short-lived forcing scenario predictions (Supplementary Fig. 3 ) and exists because pattern scaling is constrained to the same pattern, regardless of the scaling factor used to estimate the global mean response (Methods, Supplementary Fig. 4 ).

The central vertical boxes indicate the interquartile range shown on a standard box plot, the horizontal line shows the median and the black point shows the mean. The horizontal width shows the distribution of temperature values overall grid points, i.e. the wider regions highlight that more grid points have this value of predicted temperature response. Note the different vertical scales.

In the following, we quantify how well the two machine learning models and pattern scaling perform on different spatial scales. At the grid-scale level, we calculate the Root Mean Squared Error (RMSE) by comparing the prediction and GCM response at every grid point (Methods). We highlight that grid-scale error metrics need to be interpreted with care because they can present misleading results, particularly for higher resolution models. For example, they penalize patterns that—as broad features—are predicted correctly but displaced marginally on the spatial grid 44 . This issue is necessarily more prevalent for the machine learning approaches where smaller scale patterns are more frequently predicted, while pattern scaling predicts more consistently smooth, cautious patterns with reduced spatial variability (Supplementary Fig. 1 ). This consideration is a key reason why predictions for larger scale domains are often selected in impact studies 11 , 12 . We therefore also compare the absolute errors in global mean temperature and in regional mean temperature over ten broad regions (Fig. 3 ); four of which are the main emission regions (North America, Europe, South Asia, and East Asia) and the remaining cover primarily land areas where responses affect the majority of the world’s population. The boxplots in Fig. 3 show how these errors are distributed overall predicted scenarios for each regression method.

RMSE at grid-cell level and global/regional absolute errors in °C for all scenarios, calculated by averaging the predicted response over each region and taking the difference between the GCM output and the prediction using three methods: R = Ridge regression, G = Gaussian Process Regression, and P = Pattern scaling. Boxplots show the distribution of errors across scenario predictions. Boxes show the interquartile range, whiskers show the extrema, lines show the medians and black diamonds show the mean. The dots indicate the errors for each individual scenario. Note the different scale for the Arctic and that points exceed the scale in Arctic (9.5), Northwest Asia (4.7), East Asia (3.7) and the Grid RMSE (3.8).

Both Ridge and GPR generally outperform the pattern scaling approach, but we find that, in most cases, it is GPR errors that are lowest. Note that scenario-specific pattern scaling errors are necessarily dependent on the approach chosen to scale the global CO 2 -response pattern (Methods, Supplementary Fig. 4 ), but all pattern scaling approaches share their fundamental limitation in predicting spatial variability (Fig. 2 ). The large spread in absolute errors in Fig. 3 is due to the large spread in response magnitude for the different scenarios. Specifically, the large errors (e.g. 1–2 °C for the machine learning models and >3 °C for pattern scaling) come mostly from regions/scenarios with a large magnitude of response, which expectedly tend to be for strong forcings (e.g. strong solar or greenhouse gas forcings), but these errors can be small relative to the overall magnitude of scenario response. In contrast, small absolute errors can be large relative to the magnitude of response (Supplementary Fig. 5 ), making prediction more challenging for weakly forced scenarios. This is also consistent with the finding that regional aerosol perturbations, with typically weaker forcings, are more difficult to predict compared to long-lived pollutant perturbations (Fig. 2 ).

Learning early indicators

As well as advancing our predictability skills, the machine learning methods inform us about regions that experience the earliest indicators of long-term climate change in the GCM. By assessing the structure of learned Ridge regression coefficients, we find patterns in the short-term response that consistently indicate the long-term temperature response (Supplementary Fig. 6 ). In some regions (e.g. East Asia) the dominant coefficients appear in regions close to the predicted grid cell, whereas in other regions (e.g. Europe) predictions are strongly influenced by the short-term responses over relatively remote areas, such as sea-ice regions over the Arctic. This highlights the fact that climate model response predictability varies strongly depending on the region of interest, and often involves interactions with regions very far from the region of interest as well as from the emission region.

We also examine which areas are overall the most influential for long-term predictability, by averaging magnitude of coefficients across all grid cells to find a global mean coefficient map (Supplementary Fig. 6c, f ). This coefficient map mimics warming patterns seen in previous studies (enhanced at high latitudes, over land and over the subtropics) 14 but also shows amplified coefficient weights in sea-ice regions, high-altitude regions, primary emission regions and mid-latitude jet stream regions. Arctic and high-altitude regions are known to warm more rapidly due to ice and snow albedo feedbacks 45 and faster upper tropospheric warming 11 , 46 respectively. These regions exhibit accelerated warming in the simulation compared to their surroundings, making them robust harbingers of long-term change within the model. We highlight the implications for future studies that attempt to interpret already observed warming patterns from a climate change perspective.

Data constraints and future directions

We identify more extensive training data (additional simulations and forcing scenarios) as key to further improving the skill of our machine learning methods. In Fig. 4 it is demonstrated that as the number of data training samples increases, the mean prediction accuracy significantly increases and becomes more consistent. We therefore expect significant potential for further improvements in predictions with even more training data. More simulations would better constrain parameters of the statistical models and improve the chances that a predicted scenario contains features previously seen by the statistical model (e.g. refs. 38 , 47 , Methods).

Mean of absolute errors in °C across all predicted scenarios against number of training simulations, with each line representing a different region (Fig. 3 ). RMSE at the grid-scale level is also shown in black with white dots. For a fixed number of training data points, the process of training and predicting is repeated several times over different combinations of training data to obtain multiple prediction errors for each scenario. Full boxplots showing the distribution of errors across scenario predictions given these different combinations of training simulations can be found in Supplementary Fig. 7 .

Since obtaining training data from the GCM is expensive, sensible choices can also be made about how to increase the dataset by choosing which new scenarios will benefit the accuracy of the method the most, e.g. to address some complex regional aspects of the responses to short-lived pollutants. We recommend increasing the dataset to include more short-lived pollutant scenarios, noting that those with large forcings may reduce the noise in the training data so as to better constrain learned relationships (e.g. Supplementary Fig. 5 ). Some regions stand out as particularly challenging for our machine learning approaches, with Europe being a prominent example (Supplementary Fig. 2 ). This is partly due to large variations in the long-term response across the training data over Europe relative to other regions, which means predictions are less well constrained and would benefit more from increased training data. Additionally, the variability in the GCM-predicted temperature time series is generally larger over Europe compared to other regions in both the control and perturbation simulations (Supplementary Fig. 8 ). This gives rise to a weaker signal-to-noise ratio for both short- and long-term responses in this region, increasing the difficulty of learning meaningful predictive relationships. It is also noteworthy that Ridge regression predictions for Europe depend strongly on remote parts of the Arctic where the short-term response is stronger but also highly variable (Supplementary Figs. 5 and 6 ). This points to the issue that internal variability can introduce noise to the inputs and outputs of the regression. This is partially addressed with multidecadal averages in the definitions of the short- and long-term responses, under the limitation that we have only a single realization of each simulation available. If, in future work, we have available an ensemble of simulations for each perturbation, an average over these would more effectively separate the internal variability from the response. The use of several diverse simulations in the training dataset also allows the noise in the inputs and outputs to be treated as random noise in the regression, which would be even better determined with increased training data.

A key challenge of working with the climate model information here is its high dimensionality (27,840 grid cells) given the small scenario sample size of 21 simulations. We note that we tried sensible approaches to dimension reduction for decreasing the number of points in both inputs and outputs, including physical dimension reduction by regional averaging, and statistical dimension reduction with principal component analysis (PCA) 47 . However, the resulting regressions generated larger prediction errors (Supplementary Fig. 9 ). Furthermore, we explored the use of different variables as the short-term predictors, such as air temperature at 500 hPa, geopotential height at 500 hPa (as an indicator of the large-scale dynamical responses), radiative forcing or sea level pressure. Surface temperature consistently outperforms other predictors, although a similar degree of accuracy is achieved with 500 hPa air temperature and geopotential height, suggesting the information encoded by these is similar (Supplementary Fig. 10 ). Throughout, we have selected the first 10 years of the GCM simulations as the inputs to our regression, but we find promising results for even shorter periods, e.g. the first 5 years (Supplementary Fig. 11 ). Finally, we also tested other linear (e.g. LASSO 47 ) and nonlinear (e.g. Random Forest) methods for the same learning task. However, these provided weaker results so that we focused our discussion on Ridge and GPR here. We have explored the use of these methods in the context of predicting temperature responses; however, we leave open the topic of predicting other variables such as precipitation, which we expect to be more challenging due to its spatial and temporal variability 48 , 49 , but for which pattern scaling approaches are well-known to perform particularly poorly 36 , 41 , 43 , 50 .

We also wish to highlight another long-term perspective in which the framework presented here could be useful. ‘Emulators’ that approximate model output given specific inputs, are a popular tool of choice for prediction, sensitivity analysis, uncertainty quantification and calibration and have great potential for climate prediction and impact studies 23 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 . However, long-term, spatially resolved climate prediction for diverse forcings has not yet been addressed due to the cost of training such emulators. A major implication of the approach presented here is that it can catalyse designing long-term climate emulators, by using a combination of the short-term/long-term relationships presented here and trained emulators of the short-term climate response to different forcings (i.e. multilevel emulation 52 , 59 ). Training an emulator that predicts the spatial patterns of long-term response to a range of forcings would be an extremely challenging task, as it would require tens of simulations, all of them multidecadal in length, in order to train the emulator. Our method drastically accelerates this process by reducing the length of such simulations to be of the order of 5–10 years, with subsequent use of the relationships presented here for translating short-term responses to long-term responses.

Our study made use of existing simulations from a single global climate model. However, it opens the door for similar approaches to be taken with datasets from other individual climate models. The same GCMs are typically run by several different research centres across the world so that additional simulation data should be an effort of (inter)national collaboration. We therefore encourage widespread data sharing to test the limits of our approach as an important part of future research efforts in this direction. We hope that our work will catalyse developments for coordinated efforts in which carefully selected perturbation experiments will be performed in a multi-model framework. Increased availability of training datasets through model intercomparison exercises, along with increasing access to powerful computing hardware can only help with this endeavour, leading to further advances in climate model emulation.

Available simulations

To learn the regression models, we use data from long-term simulations from the Hadley Centre Global Environment Model 3 (HadGEM3) HadGEM3, a climate model developed by the UK Met Office 17 . HadGEM3 is a GCM for the atmosphere, land 18 , ocean 19 , and sea-ice 20 . In the configuration used here, the horizontal resolution is 1.875° by 1.25°, giving grid boxes ~140 km wide in the mid-latitudes 17 . The simulations were run in previous academic studies and model intercomparison projects, namely the Precipitation Driver and Response Model Intercomparison Project (PDRMIP) 16 , 31 , 32 , Evaluating the Climate and Air Quality Impacts of Short-lived pollutants (ECLIPSE) 7 , 8 , 33 and Kasoar et al. (2018) 5 , 14 , 34 . There are 21 such simulations for a range of forcings, including long-lived greenhouse gas perturbations (e.g. carbon dioxide (CO 2 ), methane (CH 4 ), CFC-12), short-lived pollutant perturbations (e.g. sulfur dioxide emissions (SO 2 , the precursor to sulfate aerosol (SO 4 )), black carbon (BC), organic carbon (OC)) and a solar forcing perturbation. For the short-term pollutants, regional perturbations exist, to account for the influence of emission region to the response 4 , 60 .

The long-lived greenhouse gas (CO 2 , CH 4 , CFC-12) simulations were performed by altering the atmospheric mixing ratios. The short-lived pollutant experiments were performed by abruptly scaling present-day emission fields in simulations performed by ECLIPSE 7 , 8 , 33 and Kasoar et al. (2018) 5 , 14 , 34 or by scaling multi-model mean concentration fields in PDRMIP 16 , 31 , 32 . The solar forcing experiment was performed by changing the solar irradiance constant 31 . The GCM is run until it converges towards a new climate state, to reach an approximate equilibrium (70–100 years). The response is calculated by differencing this with its corresponding control simulation (independent control simulations were run for each project 5 , 7 , 8 , 14 , 16 , 31 , 32 , 33 , 34 ). For the long-term response, we discard the transient response and average from year 70–100 for PDRMIP and Kasoar et al. (2018) to smooth out internal variability over the 30-year period 36 . For the 5 ECLIPSE simulations, we average from year 70 to year 80, since this is the full temporal extent of ECLIPSE simulations. For the short-term response, we average over the first 10 years of the simulation to reduce the influence of natural variability of the GCM 36 .

The experiments from PDRMIP consist of simulations with a doubling of CO 2 concentration, tripling of CH 4 concentration, a 10× increase in CFC-12 concentration, a 2% increase in total solar irradiance, 5× increase in sulfate concentrations (SO 4 ), a 10× increase in black carbon (BC) concentrations, a 10× increase in SO 4 concentrations over Europe only, a 10× increase in SO 4 concentrations over Asia only, and a reduction to preindustrial SO 4 concentrations 16 , 31 . From ECLIPSE project simulations, we use a 20% reduction in CH 4 emissions, a doubling in CO 2 concentration, a 100% reduction in BC emissions, 100% reduction in SO 2 emissions, and a 100% reduction in carbon monoxide (CO) emissions 7 , 8 , 33 . The simulations performed by Kasoar et al. (2018) consist of a 100% reduction in SO 2 over the Northern Hemisphere mid-latitudes (NHML), a 100% reduction in BC over the NHML, a 100% reduction in SO 2 over China only, a 100% reduction in SO 2 over East Asia, a 100% reduction in SO 2 over Europe and a 100% reduction in SO 2 over US 5 , 14 , 34 . Additional simulations had also been performed by the groups, but we only consider simulations where the global mean response exceeds natural variability, calculated as the standard deviation among the control simulations. This is because we want to limit the noise in the small dataset we have. Scenarios that we did not use for this reason were the global removals of organic carbon, volatile organic compounds and nitrogen oxides (ECLIPSE 7 , 8 , 33 ) and the removal of SO 2 over India (Kasoar et al. (2018) 5 , 14 , 34 ).

Regression methods

We construct the mapping between short-term temperature response ( \(x\) ) and long-term temperature response ( \(y\) ) described in Fig. 1b using Ridge regression 37 and Gaussian Process Regression (GPR) 38 . These were found to be strongest from a range of machine learning methods tested, including Random Forest and Lasso.

Ridge regression

Given output variable \(y\) and input variable \(x\) , linear regression uses the mapping

where there are \(p\) predictors, indexed by \(j = 1, \cdots ,p\) . The parameters to fit are the intercept, \(\beta _0\) , and the coefficients, \(\beta _j\) , associated with each predictor \(x_j\) . The method of least squares is used to fit the parameters by minimising the sum of the residual squared error for the training data pairs \((x_i,y_i)\) for grid points \(i = 1, \cdots ,N\) :

When the number of samples exactly equals the number of parameters, \(N = p + 1\) , this can be minimised to give a unique solution. When \(N\, > \,p + 1\) the parameters are overdetermined and this is an optimisation problem in \(\beta _j\) . In contrast, when \(N\, < \,p + 1\) , there are more free parameters, \(\beta _j\) , than there are observed data points to constrain them 47 . There are many possible values of \(\beta _j\) that satisfy (2) equal to zero, making this an under-determined problem. Our problem falls under this regime since we have many predictors (one for each grid point, i.e. \(p = 27,840\) ) but few training simulations \((N = 20)\) . This is why we introduce a regularisation constraint which penalises large values of \(\beta _j\) . Thus, we minimise 47 , 61 :

The last term shrinks many of the \(\beta _j\) coefficients close to zero, so that the remaining large coefficients can be viewed as stronger predictors of \(y\) . This introduces a bias but lowers the variance 5 . The regularisation parameter λ controls the amount of shrinkage and is chosen through cross-validation, described below. Once \(\beta _0\) and \(\beta _j\) have been learned, we can use (1) to make predictions. We carried out the regression with and without inputs \(x\) normalised to zero mean and unit variance with very little difference in results. We use Python package scikit-learn to implement Ridge regression and cross-validation 62 .

Cross-validation

Cross-validation is used here to estimate the best value of λ for prediction based on the available training data. First, we split the training dataset (of size \(N\) ) into a chosen number of subsets of size \(N_{CV}\) . We use three subsets so \(N_{CV}\) is around 6–7. Then, we iterate through a list of possible values of \(\lambda\) , and for each one, the following steps are taken.

Set \(\lambda\) from list.

Set aside one of the smaller datasets as the validation data (size \(N_{CV}\) ).

Train the regression model with the remaining data \((N - N_{CV})\) by minimising (3).

Use the inputs of the validation dataset on the trained model to make predictions on the outputs using (1) and call this \({\boldsymbol{y}} \ast\) .

Compare these predictions with the true outputs of the validation dataset using an error metric such as root-mean-squared error (RMSE), accounting for all grid cells \(i = 1, \ldots ,p\) and weighting by the grid-cell area, \(w_i\) ,

Repeat steps a-d for other subsets of validation data (we use 3 in total).

Calculate the cross-validation score as the mean RMSE for this value of \(\lambda\) for all three subsets.

This process is repeated for all values of λ in the list. The value of λ that produces the lowest \(RMSE_\lambda\) is selected as the parameter for use in the final stage of training of the model, where all training data is used.

Gaussian Process Regression

Rather than learning the parameters \(\beta _0\) and \(\beta _j\) , Gaussian Process Regression is a non-parametric approach, where we seek a distribution over possible functions that fit the data. This is done from a Bayesian perspective, where we define a prior distribution over the possible functions. Then after observing the data, we use Bayes’ theorem to obtain a posterior distribution over possible functions. The prior distribution is a Gaussian process,

where \(\mu _0\) is the prior mean function, which we assume to be linear with slope \(\beta\) , \(\mu _0\left( x \right) = \beta x\) , and \(C_0\left( {x,x^{\prime}} \right)\) is the prior covariance function, which describes the covariance between two points, \(x\) and \(x^{\prime}\) 38 . We choose the following squared exponential covariance function,

where \(\sigma ^2\) and \(l\) are the output variance and lengthscale, respectively, which reflect the sensitivity of the outputs to changes in inputs 38 .

The prior Gaussian process is combined with the data using Bayes’ Theorem to obtain a posterior distribution over functions. This is another Gaussian process, with an updated mean function, \(\mu ^ \ast (x)\) , and covariance function, \(C^ \ast (x,x^{\prime})\) ,

The details can be found in relevant textbooks 38 . Predictions of the output can then be made at unseen values of \(x\) , where the Gaussian process provides both an expected value and the variance around this value. Since the prediction is effectively built on correlations between the new inputs and the training data inputs, this variance will be lower for predictions at values of \(x\) that are closer to values already seen in training data. We follow these steps with the framework provided by GPy in Python. The values of \(\beta\) , \(\sigma ^2\) , and \(l\) are learned through optimisation (the L-BGFS optimiser) in GPy 63 .

Pattern scaling

We benchmark our machine learning models against pattern scaling, a traditional method for obtaining spatial response patterns to forcings without running a full GCM 36 , 39 . It has been widely used for conducting regional climate change projections 40 , 41 , 42 in impact studies 64 and to extend simplified models to predict spatial outputs 58 , 65 . Pattern scaling requires one previous GCM run to obtain the long-term response of the variable of interest for a reference scenario. Typically, a strong greenhouse gas perturbation, such as a doubling of CO 2 is used as this reference response pattern on the longitude-latitude grid, \(V_{{\mathrm{ref}}}\left( {{\mathrm{lat}},{\mathrm{lon}}} \right)\) . We use the 2xCO 2 scenario from PDRMIP (since more than half of the simulations are from PDRMIP we expect this to be a more valid reference pattern than the 2xCO 2 ECLIPSE scenario) 16 , 31 , 32 . Then, the variable of interest is estimated at each grid point for a new scenario, \(V^ \ast \left( {{\mathrm{lat}},{\mathrm{lon}}} \right)\) by multiplying the reference pattern by scaler value s , i.e.

The scaler value s is the ratio of long-term global mean temperature response between the prediction and reference scenario. This can be derived from either a simplified climate model, such as a global energy balance model 43 , 66 ; a statistical model 58 ; or a mathematical relationship, such as the assumed linear relationship between long-term temperature response and effective radiative forcing (ERF) 64 , 67 . We take the latter approach due to the availability of variables required to calculate ERF for the relevant perturbations studied here.

ERF is defined as the energy imbalance between the surface and the top of the atmosphere in a GCM run in which the atmosphere is allowed to respond, while sea-surface temperatures are kept fixed (i.e. no ocean coupling) 1 , 5 , 8 , 33 . These simulations were run for 5 years in previous studies 5 , 7 , 8 , 14 , 16 , 31 , 32 , 33 , 34 and therefore we average over the first 5 years of the simulations to reduce noise in the estimate of global mean ERFs.

Pattern scaling is generally considered as a fair approximation 36 , 43 , 66 but it assumes that the magnitude of the response scales linearly with the amount of radiative forcing, which is not necessarily true, particularly for climate forcings of a different type to the reference scenario 36 . Furthermore, it cannot necessarily predict the highly inhomogeneous effects of certain types of climate forcings such as from aerosol emissions.

There are alternative approaches for obtaining a sensible scaler value s such as using the ratio of short-term temperature response between the predicted and reference scenarios (see Supplementary Fig. 4 ). We note that such a method can sometimes achieve a higher performance in predicting the mean response in some regions than our machine learning approach. However, it suffers the same limitations as the method presented here, in that the spatial variability in the response is not captured, particularly for short-lived pollutants (Supplementary Fig. 3 ). This limitation will be true regardless of the choice of scaler value, since the spatial variability is fixed based on the reference pattern.

Prediction errors

We predict long-term climate response, \({\boldsymbol{y}}^ \ast\) for each scenario following the three methods described above. We calculate the Root Mean Squared Error (RMSE) at the grid-cell level with

where subscript \({i} = {1}, \ldots , {p}\) indexes the grid cell and \({w}_{i}\) is the normalised weight of grid cell \({{i}}\) . We note that measuring errors at these scales can introduce unintended biases in the evaluation of our methods. For example, even small spatial offsets in climate response patterns can lead to large, nonphysical quantitative errors 44 . We also show the absolute error in mean response over ten world regions that cover a broader spatial scale (Fig. 3 ). These are the four main emission regions; North America, Europe, South Asia and East Asia, as defined in the Hemispheric Transport of Air Pollution experiments 68 ; and six remaining regions; the Arctic, Northwest Asia, Northern Africa, Southern Africa, South America and Australia. These cover the land regions where climate responses are of interest due to societal relevance. Here we defined the prediction error as the absolute difference between the predicted response in each region, \({\boldsymbol{y}}_{r}^ \ast\) , and the response from the complex GCM in the same region, \({\boldsymbol{y}}_{{r}}\) :

where subscript \(r\) indicates the mean response overall grid boxes in that region, weighted by the grid box area. We also calculate the absolute error for the global mean response in the same way. These RMSE, regional and global error metrics are presented in Fig. 3 for all prediction methods.

Data availability

Data used in this manuscript were originally produced in previous studies 5 , 7 , 8 , 14 , 16 , 31 , 32 , 33 , 34 . Postprocessed data used to produce results in this study is available at 10.5281/zenodo.3971024.

Code availability

Code to produce results is publicly available on github.com/lm2612/Ridge_3 and github.com/lm2612/GPRegression. Use of the HadGEM3-GA4 climate model was provided by the Met Office through the Joint Weather and Climate Research Programme, and the model source code is not generally available. For more information on accessing the model, see http://www.metoffice.gov.uk/research/collaboration/um-collaboration .

IPCC Climate Change 2014: Synthesis Report (eds. Pachauri, R. K. & Meyer, L. A.) (Cambridge Univ. Press, 2015).

Collins, M. et al. Quantifying future climate change. Nat. Clim. Change 2 , 403 EP (2012).

Article   Google Scholar  

Rogelj, J., Mccollum, D. L., O'Neill, B. C. & Riahi, K. 2020 emissions levels required to limit warming to below 2 °C. Nat. Clim. Change 3 , 405–412 (2013).

Shindell, D. & Faluvegi, G. Climate response to regional radiative forcing during the twentieth century. Nat. Geosci. 2 , 294 EP (2009).

Kasoar, M., Shawki, D. & Voulgarakis, A. Similar spatial patterns of global climate response to aerosols from different regions. npj Clim. Atmos. Sci. 1 , 12 (2018).

Shine, K. P., Fuglestvedt, J. S., Hailemariam, K. & Stuber, N. Alternatives to the Global Warming Potential for comparing climate impacts of emissions of greenhouse gases. Clim. Change 68 , 281–302, (2005).

Baker, L. H. et al. Climate responses to anthropogenic emissions of short-lived climate pollutants. Atmos. Chem. Phys. 15 , 8201–8216 (2015).

Aamaas, B., Berntsen, T. K., Fuglestvedt, J. S., Shine, K. P. & Collins, W. J. Regional temperature change potentials for short-lived climate forcers based on radiative forcing from multiple models. Atmos. Chem. Phys. 17 , 10795–10809 (2017).

Collins, W. J. et al. Global and regional temperature-change potentials for near-term climate forcers. Atmos. Chem. Phys. 13 , 2471–2485 (2013).

Bitz, C. M. & Polvani, L. M. Antarctic climate response to stratospheric ozone depletion in a fine resolution ocean climate model. Geophys. Res. Lett. 39 , L20705 (2012).

Nowack, P. J., Braesicke, P., Luke Abraham, N. & Pyle, J. A. On the role of ozone feedback in the ENSO amplitude response under global warming. Geophys. Res. Lett. 44 , 3858–3866 (2017).

Hartmann, D. L., Blossey, P. N. & Dygert, B. D. Convection and climate: what have we learned from simple models and simplified settings? Curr. Clim. Chang. Rep. 5 , 196–206 (2019).

Persad, G. G. & Caldeira, K. Divergent global-scale temperature effects from identical aerosols emitted in different regions. Nat. Commun. 9 , 3289 (2018).

Shawki, D., Voulgarakis, A., Chakraborty, A., Kasoar, M. & Srinivasan, J. The South Asian monsoon response to remote aerosols: global and regional mechanisms. J. Geophys. Res. Atmos. 123 , 11585–11601 (2018).

Conley, A. J. et al. Multimodel surface temperature responses to removal of U.S. sulfur dioxide emissions. J. Geophys. Res. Atmos. 123 , 2773–2796 (2018).

Liu, L. et al. A PDRMIP Multimodel study on the impacts of regional aerosol forcings on global and regional precipitation. J. Clim. 31 , 4429–4447 (2018).

Williams, K. D. et al. The Met Office Global Coupled Model 3.0 and 3.1 (GC3.0 and GC3.1) Configurations. J. Adv. Model. Earth Syst. 10 , 357–380 (2018).

Walters, D. et al. The met Office Unified Model Global Atmosphere 7.0/7.1 and JULES Global Land 7.0 configurations. Geosci. Model Dev. 12 , 1909–1963 (2019).

Storkey, D. et al. UK Global Ocean GO6 and GO7: a traceable hierarchy of model resolutions. Geosci. Model Dev. 11 , 3187–3213 (2018).

Ridley, J. K. et al. The sea ice model component of HadGEM3-GC3.1. Geosci. Model Dev. 11 , 713–723 (2018).

Ceppi, P., Zappa, G., Shepherd, T. G. & Gregory, J. M. Fast and slow components of the extratropical atmospheric circulation response to CO 2 forcing. J. Clim. 31 , 1091–1105 (2017).

Persad, G. G., Ming, Y., Shen, Z. & Ramaswamy, V. Spatially similar surface energy flux perturbations due to greenhouse gases and aerosols. Nat. Commun. 9 , 3247 (2018).

Ryan, E., Wild, O., Voulgarakis, A. & Lee, L. Fast sensitivity analysis methods for computationally expensive models with multi-dimensional output. Geosci. Model Dev. 11 , 3131–3146 (2018).

Bracco, A., Falasca, F., Nenes, A., Fountalis, I. & Dovrolis, C. Advancing climate science with knowledge-discovery through data mining. npj Clim. Atmos. Sci. 1 , 20174 (2018).

Kretschmer, M., Runge, J. & Coumou, D. Early prediction of extreme stratospheric polar vortex states based on causal precursors. Geophys. Res. Lett. 44 , 8592–8600 (2017).

Nowack, P. et al. Using machine learning to build temperature-based ozone parameterizations for climate sensitivity simulations. Environ. Res. Lett. 13 , 104016 (2018).

Sippel, S. et al. Uncovering the forced climate response from a single ensemble member using statistical learning. J. Clim. 32 , 5677–5699 (2019).

Runge, J. et al. Inferring causation from time series in Earth system sciences. Nat. Commun. 10 , 2553 (2019).

Knüsel, B. et al. Applying big data beyond small problems in climate research. Nat. Clim. Change 9 , 196–202 (2019).

Nowack, P., Runge, J., Eyring, V. & Haigh, J. D. Causal networks for climate model evaluation and constrained projections. Nat. Commun. 11 , 1415 (2020).

Myhre, G. et al. PDRMIP: a precipitation driver and response model intercomparison project-protocol and preliminary results. Bull. Am. Meteorol. Soc. 98 , 1185–1198 (2017).

Samset, B. H. et al. Fast and slow precipitation responses to individual climate forcers: a PDRMIP multimodel study. Geophys. Res. Lett. 43 , 2782–2791 (2016).

Stohl, A. et al. Evaluating the climate and air quality impacts of short-lived pollutants. Atmos. Chem. Phys. 15 , 10529–10566 (2015).

Kasoar, M. et al. Regional and global temperature response to anthropogenic SO 2 emissions from China in three climate models. Atmos. Chem. Phys. 16 , 9785–9804 (2016).

Bishop, C. M. Pattern Recognition and Machine Learning (Information Science and Statistics) (Springer-Verlag, 2006).

Mitchell, T. D. Pattern scaling: an examination of the accuracy of the technique for describing future climates. Clim. Change 60 , 217–242 (2003).

Hoerl, A. E. & Kennard, R. W. Ridge regression: applications to nonorthogonal problems. Technometrics 12 , 69–82 (1970).

Rasmussen, C. E. & Williams, C. K. I. Gaussian Processes for Machine Learning. Cambridge MA: MIT Press (2006).

Santer, B. D., Wigley, T. M. L., Schlesinger, M. E. & Mitchell, J. F. B. Developing Climate Scenarios from Equilibrium GCM Results (1990) Max Planck Institut für Meteorologie, Report 47, Hamburg.

Hulme, M., Raper, S. C. B. & Wigley, T. M. L. An integrated framework to address climate change (ESCAPE) and further developments of the global and regional climate modules (MAGICC). Energy Policy 23 , 347–355 (1995).

Murphy, J. M. et al. A methodology for probabilistic predictions of regional climate change from perturbed physics ensembles. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 365 , 1993–2028 (2007).

Watterson, I. G. Calculation of probability density functions for temperature and precipitation change under global warming. J. Geophys. Res. Atmos 113 , D12106 (2008).

Tebaldi, C. & Arblaster, J. M. Pattern scaling: Its strengths and limitations, and an update on the latest model simulations. Clim. Change 122 , 459–471 (2014).

Rougier, J. Ensemble averaging and mean squared error. J. Clim. 29 , 8865–8870 (2016).

Hall, A., Cox, P., Huntingford, C. & Klein, S. Progressing emergent constraints on future climate change. Nat. Clim. Change 9 , 269–278 (2019)

Fu, Q., Manabe, S. & Johanson, C. M. On the warming in the tropical upper troposphere: models versus observations. Geophys. Res. Lett. 38 , L15704 (2011).

Google Scholar  

Hastie, T., Tibshirani, R. & Friedman, J. The Elements of Statistical Learning (Springer New York Inc., 2001).

Pendergrass, A. G., Knutti, R., Lehner, F., Deser, C. & Sanderson, B. M. Precipitation variability increases in a warmer climate. Sci. Rep. 7 , 17966 (2017).

Pendergrass, A. G. & Knutti, R. The uneven nature of daily precipitation and its change. Geophys. Res. Lett. 45 , 11980–11988 (2018).

Pendergrass, A. G., Lehner, F., Sanderson, B. M. & Xu, Y. Does extreme precipitation intensity depend on the emissions scenario? Geophys. Res. Lett. 42 , 8767–8774 (2015).

Williamson, D., Blaker, A. T., Hampton, C. & Salter, J. Identifying and removing structural biases in climate models with history matching. Clim. Dyn. 45 , 1299–1324 (2015).

Cumming, J. & Goldstein, M. Bayes linear uncertainty analysis for oil reservoirs based on multiscale computer experiments. In Oxford Handbook of Applied Bayesian Analysis , Oxford: Oxford University Press, (eds. O’Hagan, A., & West, M.) pp. 241–270 (2010).

McNeall, D. J., Challenor, P. G., Gattiker, J. R. & Stone, E. J. The potential of an observational data set for calibration of a computationally expensive computer model. Geosci. Model Dev. Discuss. 6 , 1715–1728 (2013).

Salter, J. M. & Williamson, D. A comparison of statistical emulation methodologies for multi-wave calibration of environmental models. Environmetrics 27 , 507–523 (2016).

Rougier, J., Sexton, D. M. H., Murphy, J. M. & Stainforth, D. Analyzing the climate sensitivity of the HadSM3 climate model using ensembles from different but related experiments. J. Clim. 22 , 3540–3557 (2009).

Lee, L. A., Reddington, C. L. & Carslaw, K. S. On the relationship between aerosol model uncertainty and radiative forcing uncertainty. Proc. Natl Acad. Sci. USA. 113 , 5820–5827 (2016).

Edwards, T. L. et al. Revisiting Antarctic ice loss due to marine ice-cliff instability. Nature 566 , 58–64 (2019).

Castruccio, S. et al. Statistical emulation of climate model projections based on precomputed GCM runs. J. Clim. 27 , 1829–1844 (2014).

Tran, G. T. et al. Building a traceable climate model hierarchy with multi-level emulators. Adv. Stat. Climatol. Meteorol. Oceanogr. 2 , 17–37 (2016).

Shindell, D. T., Voulgarakis, A., Faluvegi, G. & Milly, G. Precipitation response to regional radiative forcing. Atmos. Chem. Phys. 12 , 6969–6982 (2012).

Tibshirani, R. Regression shrinkage and selection via the lasso. J. R. Stat. Soc. Ser. B 58 (1), 267–288 (1996).

Pedregosa, F. et al. Scikit-learn: machine learning in python. J. Mach. Learn. Res. 12 , 2825–2830 (2011).

GPy. GPy: A gaussian process framework in python . http://github.com/SheffieldML/GPy (2014).

Huntingford, C. & Cox, P. M. An analogue model to derive additional climate change scenarios from existing GCM simulations. Clim. Dyn. 16 , 575–586 (2000).

Harris, G. R. et al. Frequency distributions of transient regional climate change from perturbed physics ensembles of general circulation model simulations. Clim. Dyn. 27 , 357–375 (2006).

Ishizaki, Y. et al. Temperature scaling pattern dependence on representative concentration pathway emission scenarios. Clim. Change 112 , 535–546 (2012).

Gregory, J. M. et al. A new method for diagnosing radiative forcing and climate sensitivity. Geophys. Res. Lett. 31 , L03205 (2004).

Sanderson, M. G. et al. A multi-model study of the hemispheric transport and deposition of oxidised nitrogen. Geophys. Res. Lett. 35 , L17815 (2008).

Download references

Acknowledgements

L.A.M.’s work was funded through EPSRC grant EP/L016613/1. P.J.N. is supported through an Imperial College Research Fellowship. A.V. is partially funded by the Leverhulme Centre for Wildfires, Environment and Society through the Leverhulme Trust, grant RC-2018-023. Simulations with HadGEM3-GA4 were performed using the MONSooN system, a collaborative facility supplied under the Joint Weather and Climate Research Programme, which is a strategic partnership between the Met Office and the Natural Environment Research Council.

Author information

Authors and affiliations.

Department of Physics, Imperial College London, South Kensington Campus, London, SW7 2BW, UK

L. A. Mansfield, P. J. Nowack, M. Kasoar & A. Voulgarakis

School of Mathematics and Statistics, University of Reading, Whiteknights, Berkshire, RG6 6AX, UK

L. A. Mansfield

Grantham Institute, Imperial College London, South Kensington Campus, London, SW7 2AZ, UK

P. J. Nowack & M. Kasoar

Climatic Research Unit, Data Science Institute, Imperial College London, South Kensington Campus, London, SW7 2AZ, UK

P. J. Nowack

School of Environmental Sciences, University of East Anglia, Norwich, Norfolk, NR4 7TJ, UK

Leverhulme Centre for Wildfires, Environment and Society, Department of Physics, Imperial College London, South Kensington Campus, London, SW7 2BW, UK

M. Kasoar & A. Voulgarakis

Department of Statistics, University of Warwick, Coventry, CV4 7AL, UK

R. G. Everitt

Department of Meteorology, University of Reading, Whiteknights, Berkshire, RG6 6ET, UK

W. J. Collins

School of Environmental Engineering, Technical University of Crete, Chania, Crete, 73100, Greece

A. Voulgarakis

You can also search for this author in PubMed   Google Scholar

Contributions

This work was initiated by A.V. L.A.M. carried out the analyses and wrote the manuscript. A.V. and P.J.N. supervised and contributed to writing. P.J.N. and R.G.E. advised on statistical methods. M.K. and B.C. performed the simulations.

Corresponding author

Correspondence to L. A. Mansfield .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

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

Supplementary information

Supplementary information, 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 license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Mansfield, L.A., Nowack, P.J., Kasoar, M. et al. Predicting global patterns of long-term climate change from short-term simulations using machine learning. npj Clim Atmos Sci 3 , 44 (2020). https://doi.org/10.1038/s41612-020-00148-5

Download citation

Received : 17 March 2020

Accepted : 11 October 2020

Published : 19 November 2020

DOI : https://doi.org/10.1038/s41612-020-00148-5

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

Using sequences of life-events to predict human lives.

• Germans Savcisens
• Tina Eliassi-Rad
• Sune Lehmann

Nature Computational Science (2023)

Regional differences in surface air temperature changing patterns from 1960 to 2016 of China

• Quanzhi Yuan
• Shaohong Wu

Climate Dynamics (2021)

Quick links

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

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Español (Spanish)
• Français (French)
• Bahasa Indonesia (Indonesian)
• Brasil (Portuguese)
• India (English)
• हिंदी (Hindi)
• Feature Stories
• Explore All
• Subscribe page
• Submissions
• Privacy Policy
• Terms of Use
• Advertising
• Wild Madagascar
• Selva tropicales
• Mongabay.org
• Tropical Forest Network

Reforestation to capture carbon could be done much more cheaply, study says

Share this article

If you liked this story, share it with other people.

• New research shows that a mix of natural forest regrowth and tree planting could remove up to 10 times more carbon at $20 per metric ton than previously estimated by the IPCC, the U.N.’s climate science panel.
• The study found that natural regeneration is more cost-effective in 46% of suitable areas, while tree planting is better in 54%, suggesting a tailored approach could maximize carbon capture.
• Researchers estimate that using the most cost-effective method in each location could remove 31.4 billion metric tons of CO2 over 30 years at less than $50 per metric ton.
• While the findings are promising, experts caution that reforestation alone can’t solve the climate crisis and emphasize the need to consider biodiversity and other ecological factors alongside cost-effectiveness.

Trees are allies in the struggle against climate change, and regrowing forests to capture carbon may be cheaper than we thought. According to new research published in Nature Climate Change , a strategic mix of natural regrowth and tree planting could be the most cost-effective way to capture carbon.

Researchers analyzed reforestation projects in 138 low- and middle-income countries to compare the costs of different reforestation approaches. They found it’s possible to remove 10 times more carbon at $20 per metric ton, and almost three times more at $50, compared to what the Intergovernmental Panel on Climate Change (IPCC) had previously estimated .

Neither natural regeneration nor tree planting consistently outperforms the other. Instead, the most cost-effective method varies depending on local conditions. Natural regeneration, which involves letting forests regrow on their own, is cheaper in about 46% of suitable areas. Tree planting, on the other hand, is more cost-effective in 54% of areas.

“Natural regeneration is more cost-effective in areas where tree planting is expensive, regrowing forests accumulate carbon more quickly, or timber infrastructure is distant,” said lead author Jonah Busch, who conducted the study while working for Conservation International. “On the other hand, plantations outperform in areas far from natural seed sources, or where more of the carbon from harvested wood is stored in long-lasting products.”

The research team estimates that by using the cheapest method in each location, we could remove a staggering 31.4 billion metric tons of carbon dioxide from the atmosphere over 30 years, at a cost of less than $50 per metric ton. This is about 40% more carbon removal than if only one method was used universally.

“It’s exciting that the opportunity for low-cost reforestation appears much more plentiful than previously thought; this suggests reforestation projects are worth a second look by communities that might have prejudged them to be cost prohibitive,” said Busch. “While reforestation can’t be the only solution to climate change, our findings suggest it should be a bigger piece of the puzzle than previously thought.”

To reach these conclusions, the research team gathered data from hundreds of reforestation projects and used machine-learning techniques to map costs across different areas at a 1-kilometer (0.6-mile) resolution. This detailed approach allowed them to consider crucial factors such as tree growth rates and potential species in different regions.

Ecologist Robin Chazdon, who wasn’t involved in the research, praised the comprehensive approach but highlighted important considerations beyond cost-effectiveness.

“These eye-opening findings add nuance and complexity to our understanding of the net costs of carbon storage for naturally regenerating forests and monoculture plantations,” Chazdon said. However, she emphasized that “the relative costs of carbon storage should not be the only factor to consider regarding spatial planning of reforestation.”

Chazdon pointed out some of the ecological trade-offs involved in different reforestation methods. Monoculture tree plantations, while potentially cost-effective in certain areas, often create excessive water demand and provide poor opportunities for native biodiversity conservation. In contrast, naturally regenerating forests typically offer a wider range of ecosystem services and better support local biodiversity.

“Ultimately, these environmental costs and benefits — which can be difficult to monetize — need to be incorporated in decisions regarding how and where to grow plantations or foster natural regeneration,” Chazdon said.

The study’s authors acknowledge these limitations and suggest several directions for future research. They propose extending the analysis to high-income countries and exploring other forms of reforestation, such as agroforestry or planting patches of trees and allowing the rest of an area to regrow naturally.

Additionally, the researchers emphasize the need to integrate their findings on cost-effectiveness with data on biodiversity, livelihoods and other societal needs to guide reforestation efforts in different contexts.

While the study’s findings are promising, the researchers caution that reforestation alone won’t solve the climate crisis. Even at its maximum potential, reforestation would only remove as much carbon dioxide in 30 years as eight months of current global emissions.

Reforestation is very important, but it won’t solve climate change on its own, Busch said. Ultimately, “we still need to reduce emissions from fossil fuels.”

Banner image of two men planting trees in the Yokadouma Council Forest, Cameroon. Image courtesy WWF.

Liz Kimbrough  is a staff writer for Mongabay and holds a Ph.D. in ecology and evolutionary biology from Tulane University, where she studied the microbiomes of trees. View more of her reporting  here .

How to pick a tree-planting project? Mongabay launches transparency tool to help supporters decide

Busch, J., Bukoski, J. J., Cook-Patton, S. C., Griscom, B., Kaczan, D., Potts, M. D., … Vincent, J. R. (2024). Cost-effectiveness of natural forest regeneration and plantations for climate mitigation.  Nature Climate Change , 1-7. doi: 10.1038/s41558-024-02068-1

FEEDBACK :  Use this form  to send a message directly to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.

To wipe or to wash? That is the question

Toilet paper: Environmentally impactful, but alternatives are rolling out

Rolling towards circularity? Tracking the trace of tires

Getting the bread: What’s the environmental impact of wheat?

Consumed traces the life cycle of a variety of common consumer products from their origins, across supply chains, and waste streams. The circular economy is an attempt to lessen the pace and impact of consumption through efforts to reduce demand for raw materials by recycling wastes, improve the reusability/durability of products to limit pollution, and […]

Free and open access to credible information

Latest articles.

The promise of ecocide law, from boardroom to courtroom (commentary)

Blockchain as a carbon market fix: Interview with William ten Zijthoff, Flávia De Souza Mendes & Maximilian Rösgen

‘Everything is a being’ for South Africa’s amaMpondo fighting to protect nature

Brazil’s ‘Mothers of the Mangroves’ protect an ecological and cultural heritage

A national park and its rangers in Bolivia endure persisting road construction, illegal mining

Acre’s communities face drinking water shortage amid Amazon drought

Sri Lanka’s blue carbon ecosystems at risk as government seeks way out of economic crisis

Deal ends environmental agents’ strike in Brazil, but grievances fester

you're currently offline