The biggest source of climate uncertainty is white and fluffy
UP IN THE AIR Clouds can both warm and cool the planet. How clouds will respond to global warming is the largest source of uncertainty in climate change predictions.
From space, clouds appear to perform an intricate and never-ending ballet. Thin streaks dance at the poles, vast storms plow across the jet streams, spinning cyclones get tossed up in the tropics and deep convecting monsters churn near the equator. Clouds whip and curl and billow, materializing seemingly out of nowhere and dissipating just as mysteriously.
The mystery deepens when scientists try to understand how clouds influence climate. Clouds lead a sort of double life, both trapping and deflecting planet-warming energy. Their molecules, like all water in the atmosphere, contribute to the greenhouse effect by lapping up infrared radiation emitted by Earth and redirecting some of that energy back toward the planet’s surface. But clouds’ white tops also reflect, collectively, almost a quarter of the solar radiation that reaches them, in effect shading the planet.
All told, clouds cool through reflection far more than they warm through the greenhouse effect. Without them, Earth’s surface would be, on average, about 5 degrees Celsius warmer. “Clouds are really at the heart of the climate system,” says Sandrine Bony, a climate scientist at the Université Pierre et Marie Curie in Paris.
That clouds both warm and cool is established. But how the global balance between those two effects will shift as the climate heats up is not. Even seemingly minor shifts in clouds’ behavior could substantially dampen or accelerate global warming.
Early predictions suggested that clouds might work to counteract rising temperatures: As oceans absorb more heat, they add more water vapor to the air. This, the thinking went, would create more sunlight-reflecting clouds, which would help cool the planet. In climate speak, this is known as a negative feedback. Research over the last two decades suggests, however, that the cloud feedback is more complicated and likely to result not in cooling but in added warming.
But no one knows how much additional warming, if any, to expect. The United Nations Intergovernmental Panel on Climate Change, which represents the collective knowledge of the world’s climate scientists, considers cloud feedbacks the top source of uncertainty in climate change prediction. This uncertainty is reflected in the reports that the panel releases every five to seven years. In its 2007 report, the panel estimated that if the concentration of carbon dioxide in the atmosphere were to double from its preindustrial level — a likely outcome by the end of this century — global average temperature would rise between 2 and 4.5 degrees Celsius. The panel’s latest report, officially published January 30, estimates a temperature rise of 1.5 to 4.5 degrees with carbon dioxide doubling. In other words, seven years later, the uncertainty has actually grown.
Researchers nevertheless insist that they understand clouds much better than they did in 2007. “We’ve moved from the unknown unknowns to the known unknowns,” says Leo Donner, a scientist who develops climate simulations at the National Oceanic and Atmospheric Administration’s Geophysical Fluid Dynamics Laboratory in Princeton, N.J. “I would argue that in fact there has been very significant progress, though [it’s] correct that the bottom line is still not changed.”
Scientists like Donner are increasingly convinced that cloud feedbacks are not going to diminish greenhouse gas warming. But to really put this issue to bed, they say they need more sophisticated cloud observations extending over decades. Whether researchers will get this crucial data record is far from certain.
Data collected by a new generation of cloud-gazing satellites such as NASA’s CloudSat (shown in an artist’s conception) are helping researchers pin down the basic properties of clouds and fine-tune climate simulations. JPL/NASA
Getting clouds’ numbers
Until recently, researchers had no way to monitor clouds on a global scale. Ground-based observatories could see only the bottoms of clouds. And while scientists could send balloons and airplanes through individual clouds to gather more complete profiles, these methods provided only local snapshots. Things improved in 1999 when NASA launched the first of its two MODIS instruments, which circle the planet and look down on cloud tops. But these data, too, are limited: Clouds and sea ice are nearly indistinguishable from above, and a several-kilometer-thick cloud can hide large variations in its interior.
In 2006, NASA launched the satellites CloudSat and CALIPSO. These sister orbiters fly in close formation and send out beams — radar for CloudSat and lidar (the laser version of radar) for CALIPSO — that penetrate deep into clouds and bounce off water droplets and airborne particles called aerosols before returning to the satellites. Zipping around the planet roughly every hour and a half, the satellites send down a continuous trove of data that scientists can access almost in real time.
CloudSat and CALIPSO are “a revolution in observing technology,” says Ulrike Lohmann, a cloud physicist at the Swiss Federal Institute of Technology in Zurich. Previously, scientists didn’t always know even basic things such as clouds’ altitudes, volume and how often they produce rain. Researchers have been surprised to learn how much of the water in clouds is frozen, says Lohmann. “The amount of ice in the atmosphere seems to exceed the amount of liquid almost everywhere.”
Frozen clouds seem to have a stronger greenhouse effect than liquid clouds, says Graeme Stephens, a researcher at NASA’s Jet Propulsion Laboratory in Pasadena, Calif., and chief scientist on the CloudSat mission.
“Active” instruments like radar and lidar provide “the least ambiguous view of clouds,” he says. “Everything else gives us just glimpses.” He has called the CloudSat-CALIPSO era a “golden age” of cloud observations. And these observations are helping researchers home in on some of the biggest unknowns in climate science.
Modeling the future
The first good satellite observations of clouds showed that the supercomputer-driven climate simulations used to predict future warming got a lot right. But these simulations tended to produce too few clouds and to make them too reflective. The models also produced too many storms and failed to reproduce important weather patterns.
Many of the problems stem from simplifications that scientists have to make when constructing climate models. Modelers start by dividing Earth’s surface into squares. Each square becomes the base of a stack of rectangular boxes extending up through the atmosphere. Scientists decide on a set of physical variables to sum up atmospheric conditions, and then use real-world data to assign a starting value for each factor in each box. Modelers then write computer code to make neighboring boxes interact based on the laws of physics and fluid dynamics. Finally, researchers run time forward in discrete steps and study how their digitized planet changes.
Early climate models had grid squares around 500 kilometers on a side. In today’s climate models, squares have shrunk to 100 kilometers wide. Vertical and time resolution have also improved. Still higher resolutions are possible, but going to that level of detail for long-term climate simulations would gobble up too much time even on today’s fastest computers, says David Randall, an atmospheric scientist at Colorado State University.
Resolution has been a major problem for simulating clouds, which can be as small as tens of meters across. And the processes by which clouds form — water nucleating around aerosol particles — occur at scales of micrometers.
The results of models that run at the scale of individual clouds can be fed into climate models. But this means that global climate simulations themselves must approximate or ignore important cloud-related processes.
To get around this limitation, Randall has helped pioneer a technique called superparameterization that embeds small-scale cloud models inside global climate models. The method uses a lot of computing power, but less than trying to shrink global models’ grid cells to the size of clouds. “The superparameterization approach is a kind of a compromise,” Randall says, that better reproduces certain local processes like day-night rainfall patterns and the annual monsoon cycle in the Indian Ocean. He predicts the technique will become more common.
Higher resolution helps, says Anthony Del Genio, who develops simulations for NASA’s Goddard Institute for Space Studies in New York City. But it is not the only, or even the best way to make more realistic models. “In terms of the big issues in climate change, resolution is not necessarily the answer,” he says. “Better physics is the answer.”
Today’s climate models represent cloud-related processes with dozens of variables, including humidity, amount of water condensed as water and as ice, number of droplets falling at different speeds and concentrations and sizes of various aerosol particles. Models that used to ignore or drastically oversimplify aerosols now allow them to interact with water and form droplets. Also more nuanced are climate models’ simulations of convection, the complex physical process that moves water upward in the atmosphere, causing much of the planet’s rainfall.
These improvements, notes cloud physicist Lohmann, give a more realistic representation of nature’s complexity, but not necessarily better predictive power. “It tends to be that the more we know, the larger the uncertainty gets,” she says.
Bjorn Stevens of the Max Planck Institute for Meteorology in Hamburg, who works on a model called ECHAM, emphasizes how crucial it is for climate simulations to get what is known about clouds right. He recently found that ECHAM was representing clouds in an unrealistically crude way: Instead of allowing cloudiness to vary smoothly from 0 (perfectly clear) to 1 (overcast), the value was forced to occupy one of the extremes. When Stevens and colleagues changed their computer code to allow fractional cloudiness, the model’s prediction for future temperature rise doubled.
Closing in on the cloud feedback
Before every IPCC report, the world’s 20 or so major climate modeling groups run simulations of the future climate assuming various greenhouse gas emissions projections, from an uncontrolled increase to a stabilization or decrease. Scientists then compare the simulations’ outputs to see where models agree and where they diverge. The most recent comparison showed, among other things, that nearly all models predict that as the world warms, clouds will change in ways that further increase warming. “I don’t think anybody’s really constructed a full model with a significantly negative cloud feedback,” says Andrew Gettelman, a scientist at the National Center for Atmospheric Research in Boulder, Colo. “If you discovered that you had a negative feedback and could build a self-consistent model of that, you’d become famous.”
Gettelman stresses that he and his colleagues aren’t just relying on models. They also have reasons based on established physics to think that clouds will amplify, or at least not dampen, global warming. One of these mechanisms results from the fact that the atmosphere’s lowest layer, the troposphere, is becoming taller as the climate warms. Many clouds extend to the top of the troposphere, meaning their tops are also rising. This makes the clouds’ tops colder, so they radiate less energy into space.
A related pro-warming effect results as warming at the equator expands the rotating cells of dry air that maintain Earth’s low-latitude deserts. Scientists expect these expanding cells to push midlatitude storm tracks toward the poles, widening the low-latitude cloudfree belts where most planet-warming sunlight strikes. “Your cloud is essentially giving you less bang for your buck,” explains Mark Zelinka, who studies cloud feedbacks at the Lawrence Livermore National Laboratory in California. A recent analysis of more than 30 years of weather data by Zelinka’s colleagues Kate Marvel and Céline Bonfils provided evidence that storm tracks are in fact moving poleward (SN Online: 11/11/13).
Low-altitude clouds create the most headaches for researchers trying to pin down the total cloud impact. Many scientists once thought that evaporating water in the tropics would form more highly reflective low clouds, which would act as a brake on climate change. But as scientists have come to better understand convection processes near the ocean surface, their view has changed. Most now suspect that low clouds will decrease as temperatures warm.
Using global satellite data, scientists are beginning to directly measure cloud feedbacks. In 2010, Texas A&M University climate researcher Andrew Dessler analyzed radiation data collected by satellites over the previous decade and found that the way clouds responded to natural temperature fluctuations tended to increase warming (SN Online: 12/9/10). Although he acknowledges his data series was too short to show a trend from human-caused climate change, Dessler thinks scientists already have enough evidence to rule out a large climate-saving effect from clouds. “We don’t see any evidence … that clouds are this big negative stabilizing feedback that acts to prevent warming.”
Not all scientists are ready to concede that the cloud feedback is positive, though. Stephens thinks climate models still vary too much among themselves to allow scientists to make a definitive statement. “There’s agreement that of the feedbacks that they know, those feedbacks are most likely to be positive,” he says. “It does not mean that the feedbacks overall should be positive.” Partly due to Stephens’ influence, the IPCC report states that the total cloud feedback is “likely positive” (emphasis original), leaving room for a neutral or slightly negative effect.
The key to nailing the cloud feedback, everybody agrees, is extending the data record. But this record is in jeopardy. CloudSat has already lost its nighttime observing capability because of battery problems, and failing energy supplies will cause CALIPSO to start to depart from its orbit over the next few years. The European Space Agency’s EarthCARE satellite, which will also beam radar and lidar at clouds, may temporarily fill the gap. But EarthCARE is slated to fly only from 2015 to 2018, and no one has announced a follow-on mission that could keep the data coming.
Losing the opportunity to compile a long-term record on clouds could severely hamper scientists’ attempts to observe and predict climate change, Gettelman says. The CloudSat and CALIPSO series, though valuable, are far too short to show the effects of global warming. Scientists say they need at least 20 to 30 years of data to average out natural fluctuations and determine whether clouds respond to global warming the way simulations predict. And Gettelman says he is frustrated that NASA has no plan for gathering this record. “It’s a political and bureaucratic failure to take the observations we have and make sure they continue.”
Clouds on the horizon
In January 2014, scientists analyzed how climate models simulate convection and found that many simulations get the process wrong. As a result, the team reported in Nature, these simulations produce too many low, sunlight-reflecting clouds. Models that get convection right predict, on average, substantially more warming over the next century. The study authors, who include Bony, concluded that doubling carbon dioxide should raise temperatures by 3 to 4.5 degrees, the upper half of the IPCC’s current range.
But not all evidence points in that direction. Since 1998, Earth’s surface temperature has remained roughly constant, a substantial shift after three decades of rapid warming (SN: 10/5/13, p. 14). If the climate were really as sensitive to greenhouse gases as Bony and her colleagues think, warming should have continued apace, or even accelerated. Studies of past climate changes also hint that greenhouse gases may have less impact on global temperature than many models predict. Reconciling this evidence with scientists’ latest findings on clouds is one of the main challenges facing the field today.
Aerosols could also play a joker in the climate game. In pre-industrial times, clouds nucleated around natural aerosols like salt from sea spray, volcanic sulfates and desert dust. These days, however, human-caused emissions from power plants, factory chimneys and wood stoves supplement the natural aerosol load. With more particles in the air, cloud droplets become smaller and more numerous, and therefore reflect more sunlight.
When it comes to cloud impacts, climate simulations vary widely. Modeling a 4-degree-Celsius increase in temperature, four climate models came to different conclusions on whether clouds on a simplified planet would have a warming (red) or cooling (blue) effect. B. Stevens and S. Bony/Science 2013
So clouds are almost certainly cooling the planet a bit more — and possibly a lot more — than they would without human-made aerosols.
Pinpointing how clouds have adjusted to the changing aerosol potpourri is a major unsolved problem that scientists are furiously working to solve. Writing in Science in January, Donner and other climate scientists called for a new satellite-based instrument that would measure not only the amounts of aerosols, as CALIPSO does, but also the weather patterns that move aerosols through the atmosphere. The team argues that such observations are crucial for removing remaining uncertainties from global warming predictions.
Even with those satellite measurements, however, it will be at least a couple of decades before long-term datasets and enhanced computing power allow scientists to elucidate how cloud feedbacks and cloud-aerosol interactions influence climate. If governments haven’t acted to reduce greenhouse gas emissions by then, the world will be committed to what almost all scientists consider a dangerous amount of warming. “The models aren’t perfect,” says Steven Sherwood, a climate researcher at the University of New South Wales in Sydney and a coauthor on both the January Nature and Science papers. But, he says, “it never makes sense to use uncertainty as an excuse not to do anything. You certainly wouldn’t do that if you were running a business … or in any other aspect of your life.”
