Before global warming, was Earth cooling down or heating up?
Climate models and natural climate archives are at odds over whether Earth gradually warmed or cooled over the past few thousand years. The most comprehensive evaluation of evidence supports a cooling trend that lasted until industrialization kicked off the current warming trend.  

NAU Communications and University Communications
Feb. 20, 2023


The authors of the paper, Ellie Broadman and Darrell Kaufman, assisting on a lake coring platform in northwest Alaska, August 2022
The authors of the paper, Ellie Broadman and Darrell Kaufman, assisting on a lake coring platform in northwest Alaska, August 2022. Victor O. Leshyk/Ecoss

The effects of global warming accelerated by human burning of fossil fuels have been widely established. To predict future climate change, researchers often consult information about past climate and environmental change, to demonstrate the sensitivity of the climate system and to understand the impacts of previous temperature changes. To do this, they rely on sophisticated computer models and evidence of past climates recorded in Earth's geological record.

There is only one problem: These two lines of evidence don't agree on the temperature trend over the last millennia. While climate model simulations of the last 6,000 years generally show gradual warming, geologic data stored in natural archives on Earth, such as sediments, tree rings and corals, generally show gradual cooling. This global "temperature conundrum" between models and natural archives is the subject of a review article published in Nature. The paper is authored by Darrell Kaufman, a Regents Professor in Northern Arizona University's School of Earth and Sustainability, and Ellie Broadman, a postdoctoral researcher in the University of Arizona Laboratory of Tree-Ring Research.

The paper addresses a debate in the study of past climate, or paleoclimatology, that has gone on for at least a decade, Broadman said.

"It seems a relatively straightforward question at first, but our two major lines of evidence conflict," said Broadman, who, while earning her doctorate at NAU, helped analyze a broad swath of available data from the last 12,000 years. "Many studies have addressed it in recent years but reach contrasting conclusions, and the debate remains largely unresolved."

The authors provide the most comprehensive evaluation of evidence on the topic to date, calling on evidence from natural archives, climate models and climate theory. Their study incorporates work that was included in the latest major climate report by the Intergovernmental Panel on Climate Change that Kaufman helped lead. Specifically, the authors look at whether global average temperature 6,500 years ago was warmer, as indicated by natural archives of past climate information, known as proxy records, or colder, as simulated by models, in comparison to the late 19th century, when the Industrial Revolution led to a significant increase in human-caused warming.


A schematic illustrating various natural climate archives containing evidence of past temperature changes.
Natural archives on land and in the ocean contain evidence of past temperature changes. These include biological, chemical and physical indicators, or proxies, preserved in a variety of settings that extend back for thousands of years. Victor O. Leshyk/Ecoss

"We conclude that the majority of existing evidence supports a warmer period about 6,500 years ago, followed by a cooling trend that ended in the 1800s, as suggested by most proxy records," Broadman said. "This finding is significant in part because it reveals ways we can improve climate models and refine both our understanding of the past and our predictions for the future."

Accurate climate models play a critical role in climate science and policy, helping to inform policy- and decision-makers throughout the world as they consider ways to slow the deadly effects of a warming planet and to adapt to changes already in progress. To test their accuracy, models are programmed to simulate past climate to see if they agree with geologic evidence. The model simulations can conflict with the evidence. How can we know which is correct?

What we know

Scientists know more about the climate of the Holocene, which began after the last major ice age ended 12,000 years ago, than any other multi-millennial period. There are published studies from a variety of natural archives that store information about historical changes that occurred in the atmosphere, oceans, snow and ice packs, and on land; studies that look at the forces that drove past climate changes, such as Earth's orbit, solar radiation, volcanic eruptions and greenhouse gases; and climate model simulations that demonstrate the influence of those forces on global temperature. All these types of studies were included in this review.

Also known – whether the numbers trend up or down – is that the change in global average temperature in the past 6,500 years has been gradual, probably less than 1 degree Celsius (1.8 degrees Fahrenheit). This is less than the warming already measured in the last 100 years, most of which humans have caused. However, because global temperature change of any magnitude is significant, especially in response to changing greenhouse gases, knowing whether temperatures were higher or lower 6,500 years ago is important, Kaufman said.

"Tracing changes in global average temperature is important because it's the same metric used to gauge the march of human-caused warming and to identify internationally negotiated targets to limit it," he said.

Ellie and Emmy Kelly Lake.jpg

Ellie Broadman holds a sediment core from a lake on Alaska’s Kenai Peninsula.
Ellie Broadman holds a sediment core from a lake on Alaska’s Kenai Peninsula. Emily Stone

What we don't know

This study highlighted uncertainties in the climate models. If the authors' interpretation – that recent global warming was preceded by 6,500 years of global cooling – is correct, then scientists can improve the way that natural climate forcings and feedbacks are represented in models. If they're incorrect, then scientists need to improve their ability to interpret temperature signals in Earth's natural records, and further develop analytical tools to capture these trends on a global scale.
The authors cautioned that uncertainty still exists despite recent studies that claimed to have resolved the conundrum. Their paper, they say, exposes how little is known about slow-moving changes in the climate system – those that play out over multiple centuries or millennia. These changes occur naturally, and more have been set into motion by human-caused global warming. The extent to which climate models can accurately simulate these long-term changes is still an open question, but the review suggests that they are missing important climate feedbacks that can amplify global warming.

Why it matters

Broadman, whose work includes a focus on science communication, created the figures that accompany the research. This is a critical way of communicating difficult-to-understand results to audiences – and in climate science, the audiences are diverse and include educators, policymakers, nonprofits and scientists throughout the world.

"One interesting takeaway is that our findings demonstrate the impact that regional changes can have on global average temperature," Broadman said. "Environmental changes in some regions, like declining Arctic Sea ice or changing vegetation cover in what are now vast deserts, can cause feedbacks that influence the planet as a whole. With current global warming, we already see some regions changing very quickly. Our work highlights that some of those regional changes and feedbacks are really important to understand and capture in climate models."

Additionally, Kaufman said, accurately reconstructing the details of past temperature change offers insights into how the climate responds to various causes, both natural and human caused. The results serve as benchmarks to test how well climate models simulate Earth's climate system.

"Climate models are the only source of detailed quantitative climate predictions, so their fidelity is critical for planning the most effective strategies to mitigate and adapt to climate change," Kaufman said. "Our review suggests that climate models are underestimating important climate feedbacks that can amplify global warming."


Resources for the media

Media contact(s)

Daniel Stolte

Science Writer, University Communications