By Kevin Krajick
When climate scientists talk about natural climate swings that came before humans started messing with the system, many invoke two epochs. During the Medieval Warm Period, roughly from 800 to 1200 AD, temperatures rose a few degrees above average. That warming has been connected to improved crop yields in parts of Europe, and the temporary Viking occupation of Greenland. During the following Little Ice Age, which lasted roughly from 1300 to 1850, the Greenland Vikings disappeared, glaciers from California to the European Alps advanced, and New York harbor froze, enabling people to walk from Manhattan to New Jersey without benefit of the George Washington Bridge.
For a long time, many took on faith the idea that these phenomena were global. But that assumption has been undermined in the past decade or so by studies from widespread areas (including parts of Greenland) suggesting that in fact temperatures in many places did not line up with one or the other periods. Some regions appear to have been warming when they were supposed to be cooling, and vice versa. The same goes for two lesser-known, more vaguely defined earlier swings, known as the Roman Warm Period (ca. 100-300 AD) and the Dark Ages Cold Period (ca.400-800).
A new study puts together the evidence on a global scale for the first time. Based on this, the authors say that the supposed warm and cold epochs may represent, more than anything, regional variations that can be explained by random variability. Published in the leading journal Nature this week, the study analyzes paleoclimate data from across the world, using multiple statistical methods and many sources: tree rings, glacial ice cores, corals, lake sediments. It does not suggest that the periods of high or low temperatures observed during the named epochs did not exist in certain places; rather that they did not exist everywhere at the same time, and thus probably were not caused by some kind of planetary driver.
That said, the study does find one very coherent period: an unprecedented warm one extending over 98 percent of the globe, starting in the 20th century. This is almost certainly caused by us.
We spoke with coauthor Nathan Steiger of Columbia University’s Lamont-Doherty Earth Observatory about climates of the past and present, and what we can learn from them.
What are the most important conclusions of your study?
We show that previously named climate epochs of the Common Era were not coherent phenomena across the globe. This goes against the widespread notion that periods like the Little Ice Age or Medieval Warm Period were global periods of cold or warmth. We’re not the first to point out that there are problems with this idea, but our study is the first to rigorously test the hypothesis on a global scale. In contrast to this, we see that current global warming is remarkably coherent.
How do you tell what temperatures were doing in various parts of the world during these past times?
We rely on proxies. Trees, for example, can be very sensitive to annual changes in temperature and moisture, and the width and density of their annual rings reflect those year-to-year changes. We can then sample hundreds of trees all over the world along with other natural archives to infer what climate was like in the past. For this study in particular we used several different statistical methods that combine all of these proxies to produce global maps of temperature change going back 2,000 years.
Have scientists been too narrow-minded in their geographical focus? I mean, the very name “Medieval” calls up part of European history–a period that didn’t exist in Asia, the Americas or Africa.
Paleoclimate is like many fields of study. There are historical biases in where data is collected, and how the stories about the data are developed. The first paleoclimate data were largely collected from Europe by Europeans, and so it’s not terribly surprising that the stories that try to make sense of such data are Euro-focused. Another problem is that until recently, people have been reluctant to share data and to create narratives that include more than a single, or perhaps a few, time series. If you’re a scientist who has spent a lot of time and money in producing a particular proxy time series, then there’s a tendency to emphasize the importance of that particular time series and to develop a story explaining it. The simplest story to develop is one that corresponds to a traditional understanding of what the climate “should” be doing going back in time. It’s only been in the past few years that scientists from across the paleoclimate community have begun to publicly collate a wide range of data types from all over the globe.
Are you recommending that scientists stop using terms like the “Little Ice Age”?
Not necessarily. In general, having simplified conceptual models of natural phenomena can be very useful and even essential in the pursuit of scientific understanding. It’s when the conceptual models get in the way of accurate science that problems arise. For example, when one labels any Common Era proxy time series with terms like the “Medieval Climate Anomaly,” they are usually implicitly assuming that such epochs were global, and over well-defined time intervals. Our results show that both of these assumptions are incorrect.
Not that there isn’t already plenty of evidence, but does this study add to the argument that humans are causing global warming?
We show that conditions during medieval times or during the Little Ice Age are expected to occur naturally. But the large spatial consistency of the present warm phase cannot be explained by natural variability. This result corroborates many existing studies that have shown that humans are causing global temperatures to rise since the beginning of the industrial period.
Are there limitations to your study?
Yes. Paleoclimate proxies can be used to infer past temperatures, but are not thermometers per se, and so they include non-temperature “noise.” We have therefore tried to use as many proxies as feasible for our study, but we are limited by where the data exists and the quality of the data. Uncertainties are usually largest in places without good quality proxy data. But for the particular hypothesis we’re testing, we don’t think these uncertainties significantly impact the results. We find the same results regardless of which proxy networks or which statistical methodologies we use. So we’re pretty confident in the results.
The study’s lead author is Raphael Neukom, University of Bern, Switzerland. Other coauthors are Juan José Gomez-Navarro, University of Murcia, Spain; Jianghao Wang, MathWorks, Natick, Mass.; and Johannes Werner, Bjerknes Center for Climate Research, Norway.