40 Earths: NCAR's Large Ensemble reveals staggering climate variability
Data set an instant hit with climate and Earth system researchers
Sep 28, 2016 - by Laura Snider
Sep 28, 2016 - by Laura Snider
Over the last century, Earth's climate has had its natural ups and downs. Against the backdrop of human-caused climate change, fluctuating atmosphere and ocean circulation patterns have caused the melting of Arctic sea ice to sometimes speed up and sometimes slow down, for example. And the back-and-forth formation of El Niño and La Niña events in the Pacific has cause d some parts of the world to get wetter or drier while some parts get warmer or cooler, depending on the year.
But what if the sequence of variability that actually occurred over the last century was just one way that Earth's climate story could have plausibly unfolded? What if tiny — even imperceptible — changes in Earth's atmosphere had kicked off an entirely different sequence of naturally occurring climate events?
"It's the proverbial butterfly effect," said Clara Deser, a senior climate scientist at the National Center for Atmospheric Research (NCAR). "Could a butterfly flapping its wings in Mexico set off these little motions in the atmosphere that cascade into large-scale changes to atmospheric circulation?"
To explore the possible impact of miniscule perturbations to the climate — and gain a fuller understanding of the range of climate variability that could occur — Deser and her colleague Jennifer Kay, an assistant professor at the University of Colorado Boulder and an NCAR visiting scientist, led a project to run the NCAR-based Community Earth System Model (CESM) 40 times from 1920 forward to 2100. With each simulation, the scientists modified the model's starting conditions ever so slightly by adjusting the global atmospheric temperature by less than one-trillionth of one degree, touching off a unique and chaotic chain of climate events.
The result, called the CESM Large Ensemble, is a staggering display of Earth climates that could have been along with a rich look at future climates that could potentially be.
"We gave the temperature in the atmosphere the tiniest tickle in the model — you could never measure it — and the resulting diversity of climate projections is astounding," Deser said. "It's been really eye-opening for people."
The dataset generated during the project, which is freely available, has already proven to be a tremendous resource for researchers across the globe who are interested in how natural climate variability and human-caused climate change interact. In a little over a year, about 100 peer-reviewed scientific journal articles have used data from the CESM Large Ensemble.
Running a complex climate model like the CESM several dozen times takes a vast amount of computing resources, which makes such projects rare and difficult to pull off. With that in mind, Deser and Kay wanted to make sure that the data resulting from the Large Ensemble were as useful as possible. To do that, they queried scientists from across the community who might make use of the project results — oceanographers, geochemists, atmospheric scientists, biologists, socioeconomic researchers — about what they really wanted.
"It took a village to make this ensemble happen and for it to be useful to and usable by the broad climate community," Kay said. "The result is a large number of ensemble members, in a state-of-the-art climate model, with outputs asked for by the community, that is publicly available and relatively easy to access — it's no wonder it's getting so much use."
Scientists have so far relied on the CESM Large Ensemble to study everything from oxygen levels in the ocean to potential geoengineering scenarios to possible changes in the frequency of moisture-laden atmospheric rivers making landfall. In fact, so many researchers have found the Large Ensemble so useful that Kay and Deser were honored with the 2016 CESM Distinguished Achievement Award, which recognizes significant contributions to the climate modeling community.
The award citation noted the pair was chosen because "the Large Ensemble represents one of NCAR's most significant contributions to the U.S. climate research community. … At a scientific level, the utility of the Large Ensemble cannot be overstated."
Clearly, the CESM Large Ensemble is useful for looking forward: What is the range of possible futures we might expect in the face of a changing climate? How much warmer will summers become? When will summer Arctic sea ice disappear? How will climate change affect ocean life?
But the Large Ensemble is also an extremely valuable tool for understanding our past. This vast storehouse of data helps scientists evaluate observations and put them in context: How unusual is a particular heat wave? Is a recent change in rainfall patterns the result of global warming or could it be from solely natural causes?
With only a single model run, scientists are limited in what they can conclude when an observation doesn't match up with a model's projection. For example, if the Arctic sea ice extent were to expand, even though the model projected a decline, what would that mean? Is the physics underlying the model wrong? Or does the model incorrectly capture the natural variability? In other words, if you ran the model more times, with slightly different starting conditions, would one of the model runs correctly project the growth in sea ice?
The Large Ensemble helps answer that question. Armed with 40 different simulations, scientists can characterize the range of historic natural variability. With this information, they can determine if observations fit within the envelope of natural variability outlined in the model, instead of comparing them to a single run.
Creating an envelope of what can be considered natural also makes it possible to see when the signal of human-caused climate change has pushed an observation beyond the natural variability. The Large Ensemble can also clarify the climate change "signal" in the model. That's because averaging together the 40 ensemble members can effectively cancel out the natural variability — a La Niña in one model run might cancel out an El Niño in another, for example — leaving behind only changes due to climate change.
"This new ability to separate natural internal variability from externally driven trends is absolutely critical for moving forward our understanding of climate and climate change," said Galen McKinley, a professor of atmospheric and oceanic sciences at the University of Wisconsin–Madison.
McKinley used the Large Ensemble — which she called a "transformative tool" — to study changes in the ocean's ability to take up carbon dioxide in a warming climate.
The CESM Large Ensemble is not the first ensemble of climate simulations, though it is perhaps the most comprehensive and widely used. Scientists have long understood that it makes sense to look at more than one model run. Frequently, however, scientists have done this by comparing simulations from different climate models, collectively called a multi-model ensemble.
This method gives a feel for the diversity of possible outcomes, but it doesn't allow researchers to determine why two model simulations might differ: Is it because the models themselves represent the physics of the Earth system differently? Or is it because the models have different representations of the natural variability or different sensitivities to changing carbon dioxide concentrations?
The Large Ensemble helps resolve this dilemma. Because each member is run using the same model, the differences between runs can be attributed to differences in natural variability alone. The Large Ensemble also offers context for comparing simulations in a multi-model ensemble. If the simulations appear to disagree about what the future may look like—but they still fit within the envelope of natural variability characterized by the Large Ensemble—that could be a clue that the models do not actually disagree on the fundamentals. Instead, they may just be representing different sequences of natural variability.
This ability to put model results in context is important, not just for scientists but for policy makers, according to Noah Diffenbaugh, a climate scientist at Stanford University who has used the Large Ensemble in several studies, including one that looks at the contribution of climate change to the recent, severe California drought.
“It’s pretty common for real-world decision makers to look at the different simulations from different models, and throw up their hands and say, 'These models don't agree so I can't make decisions,'" he said. "In reality, it may not be that the models are disagreeing. Instead, we may be seeing the actual uncertainty of the climate system. There is some amount of natural uncertainty that we can't reduce — that information is really important for making robust decisions, and the Large Ensemble is giving us a window that we haven’t had before.”
Deser agrees that it's important to communicate to the public that, in the climate system, there will always be this "irreducible" uncertainty.
"We’re always going to have these two components to the climate system: human-induced changes and natural variability. You always have to take both into account," Deser said. "In the future, it will all depend on how the human-induced component is either offset — or augmented — by the sequence of natural variability that unfolds."
Authors: J. E. Kay, C. Deser, A. Phillips, A. Mai, C. Hannay, G. Strand, J. M. Arblaster, S. C. Bates, G. Danabasoglu, J. Edwards, M. Holland, P. Kushner, J.-F. Lamarque, D. Lawrence, K. Lindsay, A. Middleton, E. Munoz, R. Neale, K. Oleson, L. Polvani, and M. Vertenstein
Journal: Bulletin of the American Meteorological Society, DOI: 10.1175/BAMS-D-13-00255.1
National Science Foundation
U.S. Department of Energy