The Century after Tomorrow
The Ultimate Long-Range Forecast
Will our planet's climate a few decades from now look familiar, or will it seem more like a threatening stranger? How quickly might New York City's climate turn into something resembling Washington's? Policy makers and the public need answers as they look for affordable and effective ways to address global climate change. In response, climate scientists at NCAR continue to refine their portraits of our past, present, and future atmosphere, using some of the world's most powerful computers.
A new spin on ice ages
James Croll's insights were vast in both space and time. The self-educated Croll labored as a tea merchant, hotelier, janitor, and apprentice wheelwright before finding work at Scotland's geological society.
Encouraged by correspondence with leading scientists, he penned an 1864 essay and an 1875 book (Climate and Time, in their Geological Relations) on his notion that ice ages were related to variations in Earth's orbit. Croll plotted long-term changes in the orbit's eccentricity (the degree to which it is elliptical rather than circular). He theorized that high eccentricity could trigger ice ages by prolonging and intensifying North American winters. Hotly debated at first, Croll's ideas then languished for lack of supporting data. In the 1940s, Serbian mathematician Milutin Milankovitch strengthened the theory by adding variations in Earth's tilt to the mix of variables influencing glaciation.
AT THE LABORATORIES
Putting sea ice in Earth's modeled climate
"Simulating global climate is an arduous job," says Elizabeth Hunke (Los Alamos National Laboratory).
Hunke should know: she's spent more than a decade building and refining one of the world's premier models for depicting sea ice. The resulting software is now part of the Community Climate System Model. Los Alamos, with its powerhouse array of scientists and supercomputers, has collaborated with NCAR for years on tough modeling problems. The lab's ocean model was incorporated into the CCSM early on. Hunke was hired at Los Alamos in 1994 as a postdoctoral researcher and charged with launching a sea ice model. NCAR served as a liaison for the effort. "We developed the dynamic component of the model, the University of Washington contributed the thermodynamic part, and then our colleagues at NCAR tested and validated them together," explains Hunke. "The group wasn't so big that it got bogged down, but it was big enough that we could actually tackle the problem and finish it." Improving the treatment of sea ice in global climate models is crucial not only for its own sake, with Arctic ice being depleted at a record pace, but as part of the models' overall energy balance. "Sea ice keeps the heat in the ocean from reaching the atmosphere, and it also insulates the ocean from incoming solar radiation," says Hunke. Thus, "the extent of the ice coverage is extremely important."
Web structures to foster Earth system research
"Linking enormous amounts of data to the Internet doesn't make them readily accessible," says NCAR's Don Middleton.
As head of the center's strategic initiative on cyberinfrastructure, Middleton oversees several efforts to make a vast array of climate and other geophysical data more easily findable, browsable, and downloadable. "When people would ask if NCAR had certain data, that's been a hard question to answer," says Middleton. Now the Community Data Portal (CDP) serves nonscientists as well as specialists with a more comprehensive and navigable environment. For NCAR groups providing especially large amounts of data to the research community, including climate researchers, the CDP provides one-stop help. Rather than whip up a Web site for scientific data from the ground up, or pour resources into a fancier one, says Middleton, "a scientist or a team with data can simply publish them onto the CDP, and then they get a wide variety of services—a search interface, catalogs, links to other sites—without investing redundant effort." Most importantly, he adds, "They're still in control of their data."
On a lazy afternoon in your backyard, a temperature rise of 3 degrees Celsius (5 degrees Fahrenheit) could mean the difference between long sleeves and a T-shirt. The same increase on a global scale would hasten the melting of glaciers thousands of years old. Depending on its pace, that change would also affect the range and robustness of uncounted species of flora and fauna.
When NCAR was founded in 1960, such concerns had yet to be addressed by most scientists and politicians. Now climate change has taken center stage, thanks to an accelerating jump in global temperature and a deeper awareness of the intimate links among elements of the Earth system.
The detailed outlooks society needs—exactly where will it warm the most, and when? where will drought rule, or flood reign?—aren't yet in scientists' pockets. But real progress is taking shape. The software that simulates climate is gaining in power and sophistication. Multiple projections on more powerful computers are helping scientists build consensus reports that are at once more far reaching and more specific. These reports, and the research on which they're based, incorporate once-disparate specialties more fully than ever. From an ancient lakebed in Tibet to the whorls of seawater spun off by the Gulf Stream, every part of Earth is relevant to the environmental question of the new century: What's happening to our climate?
The next step in climate modeling
One of the most valuable resources NCAR provides to the world of university research is roughly half a million lines of code. This mammoth software package is the Community Climate System Model (CCSM). It's one of only about a dozen global climate models in the world.
Each of these is molded through years of work at a leading laboratory. Each model has a distinctive approach, with its own strengths and weaknesses in simulating millennia of past climate and projecting the future. But the CCSM is the only one designed expressly for academia. Researchers from universities and federal labs work side by side with NCAR scientists to steer development of each model component. Professors can download the code, call on help from NCAR as needed, and collaborate across the modeling community as the topic dictates.
"It's a good substrate for university research," says William Collins, the CCSM chair at NCAR. Collins and colleagues spent much of the last several years exhaustively testing the CCSM3, the model's third generation, which was released in June 2004. Among its many improvements, the CCSM3 now accommodates three different levels of detail. The lowest-resolution version places the least demands on computing time, so it's well suited for tracking past climate over extremely long intervals. The intermediate version is the "workhorse for new scientific development," says Collins, while the high-resolution CCSM3 illuminates patterns from continent to continent with new fidelity, a boon to assessments of 21st-century climate.
Even as they wrap up work on a new edition, the CCSM's creators are looking ahead to the next one. They plan to build and test versions that will project each of the types of aerosols—dust, sulfates, soot from wildfires, and other airborne particles—that have complex and critical effects on regional and global climate. They want to capture the chemical reactions that transfer carbon among atmosphere, land, and ocean. And eventually they'd like to depict clouds and their climate-altering properties in much higher detail than is now possible.
Some of these processes, only dimly understood, are still being tackled in basic research. And many take place too quickly or on too small a scale to be captured in a century-spanning simulation based on today's computing resources. Still, the CCSM team is optimistic. "We'll be moving the model toward much more detailed representations of processes in the atmosphere," says Collins. "The model development never ends."
A room with a view of climate
The circuitry that plays host to CCSM calculations is located in a single ballroom-sized space in the basement of NCAR's Mesa Laboratory. Here, under tight security and strictly controlled temperature and humidity, a collection of supercomputers and storage silos carries out the computations that will help guide global climate policy.
The flagship of NCAR's climate computing fleet is Blue Sky, an IBM cluster with 1,600 processors scattered across more than 100 nodes. Shortly after its delivery in late 2002, Blue Sky placed 10th in the Top 500 rating of computational power among the world's supercomputers. An upgrade in 2003 brought its speed to a peak level of more than eight trillion calculations per second.
About half of that capacity serves university scientists in general; the rest is allocated to intensive global modeling. With academia in mind, NCAR began testing a new IBM machine in 2004 that uses the Linux operating system, a favorite among university users for its versatility and low cost. A cadre of experienced NCAR consultants is on hand for community users accessing Blue Sky and other NCAR machines.
The oft-cited Moore's Law holds that the fastest computers will double their speed about every 18 months. That's not enough to meet the needs of global climate simulation, where the detail desired in a model far outstrips the available computing speed and power. Once the computing capacity arrives, modelers are ready to create even more intricate products. With regular hardware upgrades—and the possibility of an enlarged computing facility elsewhere in Boulder—NCAR is working to meet that need.
The next assessment
Throughout 2004, computers at NCAR and elsewhere around the world hummed with calculations in support of the upcoming fourth assessment from the Intergovernmental Panel on Climate Change. Hundreds of experts spend years examining output from climate models and observations to craft a consensus report for the Intergovernmental Panel on Climate Change (or IPCC). The first three reports, issued in 1990, 1995, and 2001, gained widespread press coverage and helped shape national and global policies at the highest levels.
"The previous IPCC reports have been very comprehensive and encyclopedic," says Kevin Trenberth, one of several NCAR scientists playing major roles in the current IPCC process. "You really want to assess the state of knowledge—how you got there and what you don't know, including all of the certainties and uncertainties."
For the 2007 report, Trenberth is a convening lead author of the chapter on observations of the atmosphere and surface. That task involves coordinating with fellow convening lead Philip Jones (University of East Anglia) and 10 other lead authors, including 5 from developing countries. NCAR lead authors on other chapters include William Collins, Elisabeth Holland, Linda Mearns, and Gerald Meehl.
In preparing the report, says Trenberth, "you get brought up to date on a very broad range of topics." One emerging thread this time is global dimming, which hadn't been considered in detail in previous IPCC studies. Solar radiation at Earth's surface decreased by several percent in the latter part of the 20th century. Two related factors appear to be the culprits: an increase in aerosols, and changes in the extent and type of cloud cover. The dimming isn't enough for laypeople to notice when compared to the sharp contrasts between day and night, but it's a potential influence on climate—one of many the authors will need to synthesize.
In the latter part of 2005, Jones and Trenberth expect the first draft to be complete and available for comment by "any scientists anywhere," says Jones. "This is when the scientists who'll think we've misrepresented or ignored their views get a chance to tell us. We have to respond to all." Jones expects the chapter to include recent work on some perennially hot topics, such as how urban heat-island effects and other land-use changes are accounted for in global temperature analyses. Climate extremes will get more attention than ever—"there is a lot more information out there," says Jones—and the authors plan to endorse long-range initiatives of the Global Climate Observing System.
Not only must IPCC authors digest a staggering amount of research and produce a 1,000-page report, they have to fit this task into their everyday workloads. "It can be quite demanding," says Trenberth, "but it's also very beneficial in bringing scientists up to date, as well as providing a basis for policy decisions." Jones points out another benefit: "It produces an excellent book for teaching, when combined with earlier assessments."
Climatic drama hidden in mild times
Climate change leapt into movie theaters in 2004. Instead of Godzilla or killer bees, The Day after Tomorrow featured abrupt climate change as the villain. Scientists agreed that the film's flash-freezing of New York City was unrealistic. However, the film's modus operandi—a sudden infusion of meltwater from the Canadian Arctic that pinches off warm currents over the North Atlantic—is a process some climate scientists see as plausible, given the right set of conditions and a time span of years rather than days.
Leading the way to a clearer picture of long-ago climate change are paleoclimatologists. They study past climates with a geologist's eye for clues trapped in ice, earth, and fossils, while using the latest computer models to interpret and extend the geologic record.
Relative warmth was the rule for tens of millions of years before ice ages came to dominate the last million years. Those early epochs, and more recent warm periods between ice ages, could hold valuable clues to how this century's climate might unfold.
At Purdue University, Matthew Huber is using the CCSM3 to study the Eocene, a post-dinosaur era from about 54 to 38 million years ago. "I find unsettling the fact that the warm climates that dominated the past 90 million years are largely unexplained," says Huber. The Eocene apparently featured polar winters far warmer than ours (averaging above freezing) and tropical temperatures similar to those at present.
The CCSM's ability to produce realistic El Niño events, even for ancient climates, gave Huber and Rodrigo Caballero (now at the University of Chicago) a new angle on the Eocene. By comparing the model's results to data from lake sediments in Wyoming and Germany, Huber and Caballero found that El Niño events in the Eocene were similar in timing and structure to those observed today. Their finding serves as a hint that 21st-century warming may not affect El Niño, and thus the tropical ocean-atmosphere system, as much as previously thought.
Through an NSF grant, Huber will be helping other university scientists to use the CCSM3 for paleoclimate work. Huber and graduate student Ryan Sriver are building a series of simulations to show the model's versatility at handling a variety of idealized paleoclimates. These range from a land-free "waterworld" to one with a single, narrow land mass that stretches from pole to pole ("banana planet" ). Huber will then visit four campuses, helping researchers carry out their own CCSM3 depictions of ancient climate. "The goal is to enable paleoclimate modeling to evolve to the next level of sophistication," says Huber, "while easing the process of performing fully coupled climate modeling for all periods in Earth's history."
While some paleoclimatologists examine the Eocene, others are delving more deeply into the Holocene era, which spans the end of the last ice age (about 10,000 years ago) to the present. NCAR's Carrie Morrill gathered sediments from a lakebed in Tibet and analyzed them for her doctorate at the University of Arizona. By using isotopes of oxygen trapped within the sediment as tracers for the source region of rainfall, she found hints that the Indian monsoon weakened over a century's time, around 4,700 years ago.
"For a while, people thought the Holocene was just a flat line, climatically," says Morrill. "Now we know that abrupt climate changes happened." Since joining NCAR's paleoclimatology group in 2002, Morrill has been scrutinizing data from ice cores, pollen, and windblown silt gathered on several continents. Thus far, the evidence bolsters her case for a hemispheric—possibly global—series of climate shifts in the mid-Holocene that may have profoundly affected several civilizations, including the Indus in India and Pakistan.
"The number of paleoclimate records being generated is amazing," says Morrill. "That's enabling us to look more globally." However, she adds, "Lake sediments will tell you how climate changed, but not why. That's where models come in—they're really good for the 'why' questions."
Visits that make science happen
UOP's Visiting Scientist Programs place several researchers a year at key labs nationwide on behalf of NOAA's Postdoctoral Program in Climate and Global Change. Baylor Fox-Kemper is stationed at NOAA's Geophysical Fluid Dynamics Laboratory, where he's applying his expertise in ocean circulation to improve how global climate models portray the sea.
Fox-Kemper's focus is eddies, the circulations 10 to 100 kilometers (6–60 miles) wide that churn alongside warm currents like the Atlantic's Gulf Stream and the Pacific's Kuroshio. After spinning off, they help create vast pools of well-mixed water that gather carbon from the atmosphere and descend to the bottom of the ocean's thermocline (the zone between upper and deep oceans). Discovered in a field experiment, these pools could play a major role in Earth's carbon budget, which shapes and is shaped by climate change.
"The global climate models don't portray this water very well," says Fox-Kemper. Because of limits on computer time, most global models parameterize eddies—they make educated guesses at where and when the eddies will occur, rather than simulating them directly. In a small-scale test, Fox-Kemper found that a high-resolution global model should be able to capture not only the eddies but the subsequent pools of carbon-absorbing water. However, it might take decades before computers are fast enough for such a detailed modeling effort.
"If you try to resolve eddies for the kind of calculations we do for IPCC, it would take something like 50 years of improvements for computers to be fast enough for those calculations. We need to know the answer now, so we can predict what's going to happen 50 years from now," says Fox-Kemper. He's now working on other ways to capture the ins and outs of oceanic carbon and the flow of oceanic heat from the Equator to the poles.
Fox-Kemper says that working at a NOAA lab through a UCAR program gives him contacts and perspectives he wouldn't have otherwise. "It's different everywhere you go. Diversity of location is really diversity of thought."