The cloud-climate conundrum

by Judith Curry
Four new papers remind us of the very large uncertainties surrounding cloud-climate feedbacks.

CERN
CERN’s CLOUD Experiment recently published three new papers:

Science has an overview article: Earth’s climate may not warm as quickly as expected, suggest new cloud studiesExcerpts (my bold):
Clouds need to condense around small particles called aerosols to form, and human aerosol pollution—primarily in the form of sulfuric acid—has made for cloudier skies. That’s why scientists have generally assumed Earth’s ancient skies were much sunnier than they are now. But today, three new studies show how naturally emitted gases from trees can also form the seed particles for clouds. The results not only point to a cloudier past, but they also indicate a potentially cooler future: If Earth’s climate is less sensitive to rising carbon dioxide (CO2) levels, as the study suggests, future temperatures may not rise as quickly as predicted. 
It’s been long thought that sulfuric acid is really the key player [in cloud formation],” says atmospheric chemist Chris Cappa of the University of California, Davis, who was not involved in the research. The studies “show pretty convincingly that we don’t need sulfuric acid around to allow new particles to grow.”
The new research, however, suggests that the past may have been cloudier than scientists realized. To simulate ancient atmospheric conditions, one research group used CLOUD (Cosmics Leaving OUtdoor Droplets), a controlled chamber at CERN, Europe’s particle physics facility near Geneva, Switzerland. Nearly as big as a bus, the chamber was filled with synthetically produced air, allowing precisely controlled chemical conditions. Jasper Kirkby, a CERN particle physicist, and his colleagues introduced a mixture of natural oxidants present in the air and an organic hydrocarbon released by coniferous plants. The hydrocarbon was rapidly oxidized. The only other ingredient allowed in the chamber was cosmic rays, high energy radiation from outer space, which made the molecules clump together into aerosols. Sulfuric acid was not required. In fact, even when the researchers introduced low concentrations of sulfuric acid to the chamber such as might be found in unpolluted air, the aerosol formation rate was unaffected. In a second CLOUD experiment published simultaneously in Nature, researchers showed these same oxidized molecules could rapidly grow the particles to sizes big enough to seed cloud droplets.
In search of a pristine atmospheric environment, a second group of researchers made atmospheric measurements of aerosol formation at the Jungfraujoch high altitude research station, 3500 meters up in the Swiss Alps to confirm that this process really occurs in nature. Over the course of a year, they measured the changing concentrations of sulfuric acid and organic molecules in the air. They found more aerosols formed with more organic molecules around, and—crucially—observed formation of organic particles without sulfuric acid. They used exactly the same instruments as at CLOUD to analyze the aerosols: “The clusters were formed mainly by organics,” says atmospheric chemist Federico Bianchi of the Paul Scherrer Institute in Villigen, Switzerland, who led the Jungfraujoch research published today in Science.
All the researchers stress sulfuric acid is still a major contributor to cloud formation on Earth today. “Today the purely plant-based pathway is much less important than it was preindustrially,” Kirkby explains. Crucially, however, the result means climate modelers can’t assume that the ancient past was much less cloudy simply because there was less sulfur dioxide. If ancient cloud cover was closer to today’s levels, the increase in the cloud-cooling effect due to human pollution could also be smaller—which means that Earth was not warming up so much in response to increased greenhouse gases alone. In other words, Earth is less sensitive to greenhouse gases than previously thought, and it may warm up less in response to future carbon emissions, says Urs Baltensperger of the Paul Scherrer Institute, who was an author on all three papers. He says that the current best estimates of future temperature rises are still feasible, but “the highest values become improbable.”  The researchers are currently working toward more precise estimates of how the newly discovered process affects predictions of the Earth’s future climate.
Nature News also has an article on the papers: Cloud-seeding surprise could improve climate predictionsExcerpts:
The findings run contrary to an assumption that the pollutant sulphuric acid is required for a certain type of cloud formation — and suggest that climate predictions may have underestimated the role that clouds had in shaping the pre-industrial climate.
If the results of the experiments hold up, predictions of future climate change should take them into account, says Reto Knutti, a climate modeller at the Swiss Federal Institute of Technology. For 20 or more years, clouds have been the largest source of uncertainty in understanding how manmade emissions affect the atmosphere, he says.
In addition to releasing carbon dioxide, burning fossil fuels indirectly produces sulphuric acid, which is known to seed clouds. So, climate scientists have assumed that since pre-industrial times, there has been a large increase in cloud cover, which is thought to have an overall cooling effect by reflecting sunlight back into space. And they have assumed that this overall cooling effect has partially masked the climate’s underlying sensitivity to rising carbon dioxide levels.
The latest experiments suggest that it may have been cloudier in pre-industrial times than previously thought. If this is so, then the masking effect, and in turn the warming effects of carbon dioxide, might have been overestimated, says Jasper Kirkby, a physicist at the CERN, Europe’s particle-physics laboratory near Geneva, Switzerland, who led one of the experiments.
But Kirkby adds that it is too early to say whether this is true in practice, or by how much, because there are so many factors that play into such projections. “There are many uncertainties; we are only talking about one,” says Kirkby. Knutti says the results will probably not affect the most likely projections of warming, as laid out by the Intergovernmental Panel on Climate Change. “Our best estimate is probably still the same,” he says.
Until recently, atmospheric scientists thought that only sulphuric acid vapour, which can be produced by volcanic emissions or by burning fossil fuels, could trigger this process. As a result, it was thought that pre-industrial skies were somewhat less cloudy than present ones because they contained less of this pollutant, says Kirkby.
In addition to feeding into climate predictions, the findings have another potential implication, says atmospheric scientist Bjorn Stevens of the Max Planck Institute for Meteorology in Hamburg, Germany. Some scientists have warned that measures such as scrubbing sulphur dioxide from coal-plant emissions could remove some of the beneficial cooling effect of clouds and boost global warming, but this may now be less of a concern because trees can seed clouds too. “What it means is, we don’t have to fear clean air,” says Stevens.
It is also interesting to speculate whether trees emit these compounds in part because there is a benefit to them in making their own climate, Kirkby says. “This really does touch on the Gaia hypothesis,” he says, referring to the theory that Earth’s life behaves as a single organism that tends to preserve itself. “It’s a beautiful mechanism for trees to control their environment.”
JC comments:
There are some interesting results here in terms of confirming Svensmark’s ideas, and the experiments (both laboratory and in nature) seem to be well conceived and executed. However, the authors have made some bizarre and incorrect statements in the press release and in interviews, including the headlines.
Authors: Until recently, atmospheric scientists thought that only sulphuric acid vapour, which can be produced by volcanic emissions or by burning fossil fuels, could trigger this process.
Sulphate particles are the primary source of atmospheric cloud condensation nuclei (CCN). Burning fossil fuels and volcanoes produce sulphuric acid, which is transformed into sulphate particles (most frequently, ammonium sulphate). However, pollution and volcanoes are far from the only source of sulphate particles, and may not even be the dominant source. There are also marine sources of suphate particles, primarily from dimethylsulfide. Further, there are non-sulphate natural sources of CCN (e.g. sea salt, volatile organic compounds, soil). Biogenic cloud condensation nuclei (from organic aerosol or secondary organic aerosol formed from chemical reactions in the atmosphere) have long been known to be a source of CCN, although the exact mechanisms continue to be studied. A very good article on this is by Carslaw et al. 2013 Large contribution of natural aerosols to uncertainty in indirect forcing.  Key excerpt from the abstract:
Here we perform a sensitivity analysis on a global model to quantify the uncertainty in cloud radiative forcing over the industrial period caused by uncertainties in aerosol emissions and processes. Our results show that 45 per cent of the variance of aerosol forcing since about 1750 arises from uncertainties in natural emissions of volcanic sulphur dioxide, marine dimethylsulphide, biogenic volatile organic carbon, biomass burning and sea spray. Only 34 per cent of the variance is associated with anthropogenic emissions.
The headline for the CERN press release
CLOUD shows pre-industrial skies were cloudier than we thought
seems quite bizarre and unsupported by their research. They seem to infer that the only source of CCN acknowledged by climate scientists is pollution aerosol (from burning fossil fuels)- which is most definitely incorrect – and then somehow infer that pre-industrial times were less cloudy. No one to my knowledge has previous asserted, much less actually provided any evidence, that pre-industrial skies were less cloudy. CCN is not a limiting factor in cloud formation; there is plenty of natural CCN even in pristine, remote environments. Whether or not a cloud forms does not depend on the number of CCN, although the number of CCN can influence the concentration and size of the droplets, and hence the cloud radiative characteristics and whether or not the cloud forms precipitation – both of which can influence the lifetime of the cloud.
The observational problems surrounding the historical cloudiness records are described in this paper by Dai et al.  The only study that I am aware of that looks at trends in the global cloud data is a paper by Eastman and Warren that analyzed land observations during the period 1971-2009.  The paper concluded: “Global-average trends of cloud cover suggest a small decline in total cloud cover, on the order of 0.4% per decade.”  Eastman et al. 2011 analyzed observations over oceans from 1954 to 2008, and concluded: “Given the subtle long-term variation in cloud cover shown on the global-scale, spurious variation makes finding trends on a large scale a perilous pursuit. Looking at smaller regions (adjusted for the long-term global variation), a possible increase in total cloud cover is observed in the central Pacific, while possible declines are seen in stratiform cloud cover in regions of persistent marine stratocumulus clouds”. Note, pollution aerosol (generated mostly in NH land regions) does not have a big signal on marine clouds.
Author: If ancient cloud cover was closer to today’s levels, the increase in the cloud-cooling effect due to human pollution could also be smaller—which means that Earth was not warming up so much in response to increased greenhouse gases alone. In other words, Earth is less sensitive to greenhouse gases than previously thought, and it may warm up less in response to future carbon emissions.
I would be very interested to see if there is a trend in 20th century cloudiness as simulated by climate models. A quick google search doesn’t turn up anything of interest, but maybe I am missing something. But I sincerely doubt that modeled cloudiness prior to 1950 is much different from cloudiness following 1950.
The aerosol indirect effects – which relates to the effect of aerosols on cloud optical properties, phase (liquid or ice) and precipitation formation – has only a very indirect effect on cloudiness (fractional coverage of clouds). The magnitude of the aerosol indirect effect is the subject of much controversy and uncertainty, with the most recent estimate by Bjorn Stevens suggesting that it is much smaller than has been assumed by the AR4 and AR5 (and CMIP3, CMIP5).
While I agree with their conclusion “Earth is less sensitive to greenhouse gases than previously thought”, I don’t see that their research adds any support to that conclusion. I simply can’t imagine what their line of reasoning is, other than to assume that they are making an embarrassing assumption that there is a dearth of natural CCN and hence very little CCN in preindustrial times that somehow meant fewer clouds formed. Historical records of rainfall suggests that indeed clouds did exist in abundance in preindustrial times.
Pollution aerosol makes some stratiform clouds more reflective, and hence has a cooling effect. An abundance of natural aerosol can have the same effect. However, this affect impacts only certain cloud types with relatively small optical depths; at larger optical depths, any increase in the number of particles will have a diminishing impact on the cloud reflectivity. Their papers don’t change our understanding of this relationship.
Climate models represent the aerosol indirect effects in different ways, with varying magnitudes. Climate models don’t currently include interactive aerosol chemistry (modeling sources and sinks), although a few are beginning to introduce a sulfur cycle in experimental mode.
I found this statement by Bjorn Stevens to be significant:
Some scientists have warned that measures such as scrubbing sulphur dioxide from coal-plant emissions could remove some of the beneficial cooling effect of clouds and boost global warming, but this may now be less of a concern because trees can seed clouds too. “What it means is, we don’t have to fear clean air,” says Stevens.
I agree with this statement – we don’t need to fear clean air for eliminating this cooling effect. There are plenty of natural sources of CCN – I think Carslaw’s estimate of 34% anthropogenic is about right.
I also find this statement by Kirby to be intriguing:
“This really does touch on the Gaia hypothesis. It’s a beautiful mechanism for trees to control their environment.”
This reminds us that there are undoubtedly many feedbacks in the climate system that are not yet accounted for, including possibly stabilizing (negative feedbacks).
Cloud phase feedback
Another relevant paper was recently published in Science by Tan et al.:

The paper is described in an article in ScienceDaily Climate models underestimate global warming by exaggerating cloud brightening.  Excerpts:
As the atmosphere warms, clouds become increasingly composed of liquid rather than ice, making them brighter. Because liquid clouds reflect more sunlight back to space than ice clouds, this “cloud phase feedback” acts as a brake on global warming in climate models.
But most models’ clouds contain too much ice that is susceptible to becoming liquid with warming, which makes their stabilizing cloud phase feedback unrealistically strong. Using a state-of-the-art climate model, the researchers modified parameters to bring the relative amounts of liquid and ice in clouds into agreement with clouds observed in nature. Correcting the bias led to a weaker cloud phase feedback and greater warming in response to carbon dioxide.
“We found that the climate sensitivity increased from 4 degrees C in the default model to 5-5.3 degrees C in versions that were modified to bring liquid and ice amounts into closer agreement with observations,” said Yale researcher Ivy Tan, lead author of the paper.
In nature, clouds containing both ice crystals and liquid droplets are common at temperatures well below freezing. As the atmosphere warms due to carbon dioxide emissions, the relative amount of liquid in these so-called mixed phase clouds will increase. Since liquid clouds tend to reflect more sunlight back to space than ice clouds, this phase feedback acts to reduce global warming. The icier the clouds to begin with, the more liquid is gained as the planet warms; this stabilizing feedback is stronger in models containing less liquid relative to ice at sub-freezing temperatures.
“Most climate models are a little too eager to glaciate below freezing, so they are likely exaggerating the increase in cloud reflectivity as the atmosphere warms,” said LLNL coauthor Mark Zelinka. “This means they may be systematically underestimating how much warming will occur in response to carbon dioxide.”
These results add to a growing body of evidence that the stabilizing cloud feedback at mid- to high latitudes in climate models is overstated.
“The evidence is piling up against an overall stabilizing cloud feedback,” concluded Zelinka. “Clouds do not seem to want to do us any favors when it comes to limiting global warming.”
JC comments:
Quantifying the cloud phase feedback is elusive owing to the complexities of heterogenous ice nucleation – the freezing of water droplets between temperatures of about -4C to -40C that are mediated by aerosol particles (ice forming nuclei, or IFN, have different chemical compositions from CCN and are mainly dust, soil, soot.) Further, there are multiple mechanisms of ice nucleation. Most climate models use a simple temperature relationship for ice nucleation, although a few models include more sophisticated parameterizations.
The clouds that are susceptible to this feedback are the stratiform clouds of mid and high latitudes. However, for relatively thin clouds, there is also a longwave effect, so maintaining supercooled water clouds at colder temperatures increased the downwelling longwave radiation, and hence has a surface warming effect. In fact, the longwave effect dominates over the shortwave effect at high latitudes (I have published numerous papers on this topic). I suspect that the climate model does not include such effects in the longwave radiative transfer parameterization (but I haven’t looked at the most recent version of the CCSM radiation code).
The nature of the heterogenous ice nucleation parameterization (summarized in my 2012 publication) will determine the nature of the feedbacks – including only a temperature dependence (and not a supersaturation dependence) will provide a feedback that is quantitatively and qualitatively different from one with a supersaturation dependence.
JC reflections 
Clouds and their feedbacks remain the biggest uncertainty in climate models, and this hasn’t change since the IPCC FAR in 1990. The cloud problem is conceptually divided into two parts – the dynamics of clouds (i.e. how, when and where they form) and the microphysics of clouds (i.e. what happens to the cloud particles after the cloud forms, including nucleating new particles, particle growth, particle phase, and precipitation). Aerosols modulate the cloud microphysics, but only have a minor impact on cloud dynamics.
Since about 2001, focus (i.e. funding in the U.S.) has focused more on cloud microphysics. While a challenging problem, it is much more tractable problem than cloud dynamics. I refer you to our textbook Thermodynamics, Kinetics and Microphysics of Clouds for background.
Climate models are in their infancy with regards to parameterizing cloud-aerosol interactions. A 2006 RealClimate post describes the state of the art about 10 years ago. The sophistication has increased somewhat, but to really understand this we need climate models with interactive atmospheric chemistry and aerosol sources and sinks. Some regional and global climate models are experimenting with such interactions, but they are computationally very expensive.
Climate models are extremely sensitive to details of the cloud microphysical parameterizations, as reflected by the Tan et al. paper. My assessment is that this sensitivity is an artifact of missing degrees of freedom in the model that are needed to regulate the interactions among aerosols, clouds, radiation and temperature. Because of the nonlinearities in the model, we don’t really know how to make good choices about simplifying the processes until we can test understanding with a complete model having all the appropriate degrees of freedom. In the absence of this understanding, it might be a better choice to simplify the cloud microphysical parameterizations and avoid introducing new feedbacks into a model that doesn’t have the appropriate degrees of freedom.
After working on this general topic for decades, it is my ‘hunch’ that the aerosol-cloud indirect feedbacks are relatively small, with negative feedbacks in the system (e.g. aerosol removal processes, ocean and land emissions of volatiles). To first or even second order, I would say that these processes could be ignored in terms of their impact on the climate system.
The real challenge is getting the cloud dynamics and associated feedbacks correct. In fact, the cloud dynamics problem is probably an order of magnitude more important than microphysical processes in terms of W/m2. The regulating effects associated with cloud dynamics are potentially very substantial – Lindzen and Spencer have contributed to our understanding of this, but there is so much more that we don’t understand. I am hopeful that the (relatively) new cloud satellites (e.g. CloudSat and Calipso) will provide the raw data needed to understand the cloud dynamical feedbacks in the climate system.
The bottom line is that by focusing only on a single piece of the puzzle, there is plenty of scope for observational and modeling studies concerning clouds and climate to obtain dramatically high or low values of climate sensitivity. I can only hope that climate scientists studying these problems won’t ‘overplay their hand’ by proclaiming that climate sensitivity is higher or lower based on their latest study of a single puzzle piece.
 Filed under: Sensitivity & feedbacks

Source