Inside UVA

Finding meaningful patterns in climate

Between the mid-1600s and the early 1700s, European winter temperatures were reduced by 1.8 to 2.7 degrees Fahrenheit — and that difference was enough to routinely freeze canals solid in Holland, as well as lock up Greenland in ice. Rivers in Europe that were typically ice-free, froze over and inspired ‘Frost Fairs’ as well as activities like skating and even golf as seen here in“Sports on a Frozen River,”ca. 1660, by Aert van der Neer (Dutch, 1603/4–1677).

Photo by Tom Cogill

Environmental Sciences professor Michael Mann has used a variety of creative strategies to create a record that can function as a baseline for creating climate change models able to distinguish between ordinary oscillations in temperature and those produced by global warming. He has tapped the information contained in such natural archives as tree rings, ice cores and coral reefs.

By Charles Feigenoff

Predicting the weather a day at a time is hard enough. The problem that Michael
Mann, assistant professor of environmental sciences, has created for himself is even more difficult: to project the weather back year by year for a millennium and beyond. His interest is not simply historical, though Mann is clearly motivated in part by his curiosity about the past. Instead, his purpose is to create a record that is lengthy enough to function as a baseline for creating climate change models that can distinguish between ordinary oscillations in temperature and those produced by global warming.

Mann has had to use a variety of creative strategies to produce such a record. First of all, he has tapped the information contained in a variety of such natural archives as tree rings, ice cores and coral reefs. He also looks at anecdotal records found in old letters, books and other documents. The strategy of using different sources has many benefits. Most obviously, it helps Mann create an unbroken record that goes back 1,000 years. “We only use records where the chronology is exact,” he says.

This approach also helps compensate for the limitations inherent in each type of data. For instance, ice cores reflect temperatures only in arctic regions and higher elevations when snow fell. The layers of calcium carbonate found in coral reefs provide an indication of climate conditions that is limited to tropical seas. Tree rings are found everywhere except in tropical forests. The same limits can be seen in the historical record. For instance, most historical accounts of the “Little Ice Age,” a 400-year period from approximately 1450 to 1850, were produced by people living in Europe and eastern North America. Mounting evidence suggests that these observations do not reflect conditions elsewhere.

Climate change can occur regionally

Be glad you weren’t living on Greenland in the late 17th century. The island often was surrounded by ice and cut off from nearby lands.

One example of climatologist Michael Mann’s research in patterns of climate change shows that only some regions were affected during the period known as the “Little Ice Age.” This resulted from a decrease in the sun’s activity, which occurs periodically, causing a shift in winds.

Mann, assistant professor of environmental sciences at U.Va., and his colleagues at NASA’s Goddard Institute for Space Studies and at the University of Massachusetts, compared results of a computer climate simulation with actual climate data to estimate climate and atmospheric conditions during the peak of the Little Ice Age. Their work shows that climactic changes during the period were concentrated more regionally than globally.

Changes in the sun’s energy used to be one of the biggest factors influencing climate change. Since the industrial revolution, however, greenhouse gasses have become the biggest catalysts.

Between the mid-1600s and the early 1700s, the Earth’s surface temperatures in the Northern Hemisphere appear to have been equal to or near the lowest levels during the 20th century, and only about one degree Fahrenheit colder than today. In contrast, European winter temperatures were reduced by 1.8 to 2.7 degrees Fahrenheit — and that difference was enough to routinely freeze canals solid in Holland, as well as lock up Greenland in ice.

“In an inverse way, this has implications for the global warming debate,” Mann said. “It could suggest that when global temperature warms even slightly, some regions would experience potentially significant temperature increases, which could affect people, crops and ecosystems.”

“By using records from a variety of sources, we can assemble a much more comprehensive, global picture,” Mann says. By matching these data with well-understood weather patterns like El Niño, we can make reasonable surmises about conditions in areas for which we have no data.

For Mann, creativity involves moving deeper and deeper into a problem. He found the level of accuracy attained by simply compiling data was inadequate because it does not account for local anomalies — conditions that might produce unusual readings in a specific year or contaminate the data. Consequently, he chose to apply the tools of multivariate statistical analysis, which enables him to compare groups of data over any given year and compare them with established patterns. He can then identify the most reliable information in the available data.

“We turn to the instrumental record established during the past 100 years to identify a dozen or so influential patterns,” he says. “In essence we are using the 20th century to train the proxy data from previous centuries.” To ensure that he doesn’t bias his analysis toward the 20th century, he assumes that the frequency and amplitude of these patterns might be different in the past.

Mann’s creativity and persistence have paid off. Conclusions about yearly temperature produced by his statistical methods have been confirmed by climate models that incorporate such factors as variations in solar energy, including the effects of these variations on the jet stream.

This validation, in turn, helps give the modelers the assurance they need to apply their models to predicting changes in the future.

Main story reprinted from the Winter 2002 issue of Explorations, produced by the Office of the Vice Provost for Research and Public Service and the University Development Office of Corporate and Foundation Relations. Online at www.virginia.edu/researchandpublicservice/explorations

Clarification: All of the articles reprinted from Explorations in the Feb. 8 issue of Inside UVA were written by Charles Feigenoff.