Cloud cover decreases in a warming planet.

Cloud Cover in a Warming Earth

I’ve looked at clouds from both sides now

And up and down, and still somehow

It’s clouds illusions I recall

I really don’t know clouds at all

 - Joni Mitchell

  (this post has been updated on 28/7/13)

Summary

The contribution of clouds to climate sensitivity is uncertain and difficult to evaluate. Consideration of the effect of rising temperature on cloud cover may help us to avoid and elucidate the uncertainties. It is shown here that as air temperatures increase, the quantity of cloud cover decreases and its quality alters in the direction of providing a positive feedback. 

Climate Sensitivity

We know for sure that human activity has been adding greenhouse gases (GHGs) to the earth’s atmosphere, and that the climate will be committed eventually to increased temperatures of ~1.2°C arising from these GHGs if CO2 doubles, irrespective of feedbacks. At present the CO2 has increased by some 40%.

Climate Sensitivity is the degree to which the earth climate will change in response to a doubling of CO2 as a result of feedbacks. It is generally considered that the impact of this increase on the global climate will be a further 3ºC, (+/– 1.5ºC), primarily through positive feedback effects on albedo and water vapour. However, the feedback effect of clouds remains difficult to ascertain, with man-made global warming (AGW) sceptics attempting to show that clouds have a negative feedback effect. They are at pains to show that there are several influences (cosmic rays, dimethyl sulphide and other particles) which will increase cloud formation in a warmer climate.  Their aim is to show that clouds provide a natural negative feedback on global temperature increases, and therefore the additional greenhouse gases that the human economy is injecting into the atmosphere is not a cause for concern.

In scientific terms, the sceptic hypothesis is that climate sensitivity is low, and they depend heavily on clouds to try to sustain their case.

The Complex Effect of Clouds on Climate

Seen from below, clouds reflect infra-red radiation back to the surface, and have a warming effect. Seen from above, clouds reflect short wave radiation back into space, and have a cooling effect. This latter effect is called albedo. Cloud albedo can vary greatly with water drop size, liquid water content, ice content and thickness of the cloud.

Low clouds have a net cooling effect on the planet, while high clouds have a net warming effect. Polar clouds both warm the surface below them, and decrease moisture transfer to higher levels, resulting in a neutral effect.

So evaluating the precise effect of cloud at any point in time and place is a brutally complex and challenging task. However, evaluating their behaviour in time as temperature changes is far more simple and straightforward, and throws useful light on the complexities.

[update 26.9.2012]
I have been correspondence with a climate scientist who says:

You have to be careful with cloud feedback and cloud cover, because the radiative effects of clouds (and therefore cloud feedbacks) also include:

•   cloud optical thickness : thin vs thick cloud, like wispy cirrus (thin) versus heavy stratus (thick),
•   cloud particle size (changes the relationship of cloud particle size to the wavelength of light and therefore to the spectral wavelength effects of clouds), 
•   cloud height (difference between temperature of the cloud and the surface control its greenhouse warming effect), and 
• cloud emissivity (the infrared version of visible cloud optical depth).  

So a direct relationship between cloud cover and cloud feedback only exists if all other cloud properties remain constant.

Hovever, if you use the effect of clouds on radiative fluxes (e.g. CERES cloud radiative forcing for solar, infrared, and combined net) then you avoid the need to handle individual properties since you are getting the full radiative impact, which is most closely related to the feedback (see for example the radiative kernel paper by Soden et al. Journal of Climate 2008 .  This is the paper that clearly shows how to unscramble feedbacks in climate models, but would work for the earth as well.  

Method

On Climate4you (a website by a climate sceptic, although, refreshingly, it provides data rather than rhetoric) we find a page on climate + clouds which contains this figure:

Fig 1

The commentary on the figure, by the website author Ole Humlum reads (emphasis added):

Scatter diagram showing the total monthly global cloud cover plotted versus the monthly global surface air temperature, since July 1983. High values of global cloud cover is associated with low global temperatures, demonstrating the cooling effect of clouds. A simple linear fit model suggests that an increase in global cloud cover of 1 percent corresponds to a global temperature decrease of about 0.07^oC. From a simple statistical point of view, this model explains about 28 percent of observed spread of surface air temperature in the diagram. .... The association between low clouds and global temperature is almost similar, as can be seen by clicking here. Data sources: The International Satellite Cloud Climatology Project and University of East Anglia's Climatic Research Unit. (comment by site owner, Ole Humlum)

Ole's view that this demonstrates the cooling effect of clouds is too simplistic, as I will show below.
It depends on the direction of the arrow of causality: do more clouds cause cooling, or does more warming cause less clouds?

Climate contrarians like Lindzen and Spencer have tried to show that changes in cloud cover cause changes in temperature. Dessler has refuted them, as is explained here.

From the slope of the graph, we learn that a warming of 0.9C would cause a 14% reduction of global cloud cover.

Measuring cloud is a challenging task, because there may be more than one layer of cloud at some points. However, the slope of the graph is pretty clear.



[Update: Eastman and Warren have found a reduction in cloud cover from 1971 to 2009. Global temperatures increased by 0.6C in that time. However, the reduction they found was far less than that predicted above. They found 1.6% reduction, whereas the graph above would expect a 10% fall. This discrepancy may be due in part to the fact that they were looking only at marine cloud, which differs from land cloud, being lower (and therefore having more cooling effect). Average cloud cover at sea is 68%, while cover over land is only 54%.

The figure shown above relates to total global cloud cover.

Interestingly, Eastman and Warren found an increase in cumulonimbus cloud, which is consistent with global warming theory, and consistent also with the flooding we are experiencing, since cu-nims cause more intense, localised rainfall.]



Discussion

As well as demonstrating the cooling effect of clouds, the figure demonstrates that cloud cover is inversely proportional to temperature: in a warmer climate, cloud cover is less. Which means that, if there is less cloud, there is more sunshine, so that more surface warming takes place during the day. But at night, more heat will escape out into space.

A 0.5ºC increase in temperature will reduce the amount of cloud cover by about 7%.

Which way does the causality lie? Do more clouds cool the earth, or does a cooler earth mean more cloudiness?

Certainly, physics points towards the latter explanation: in a cooler atmosphere, more water vapour will condense into droplets. It is well known that clouds form when moist air masses are cooled.

I looked closely at the peaks of temperature and cloudiness in Figure 2 below by drawing vertical lines. There seemed to be no significant lead or lag either way. This matters to the climate action delayers ("skeptics") because they claim that clouds must cause temperature variations. 

It does not really matter from a long term point of view. What we learn from this relationship is that in a warming world there will be less cloud.

This relationship is borne out by this graph, again from Climate4you:

Fig 2

Green line is tropical cloud cover, blue is the surface temperatures.

It is clear that the trends of tropical cloud cover and temperature are opposed, so that as global surface temperatures rise, cloud cover becomes less.

The composition of the clouds also change with rising temperatures, as is shown in Figure 3:

Fig 3

Figure 3 shows the trends since 1983 of different types of cloud. Confusingly, the position of low, middle and high clouds are reversed in the graph. Bearing in mind that the overall temperature trend since 1983 is upward, with a plateau since 2000, it is clear that low level clouds (which have a cooling effect) are decreasing as temperatures rise, and that middle level and high level clouds (which have a warming effect) are increasing as temperatures rise.

Therefore the net effect of increasing global temperatures is to push cloud cover towards having a warming effect.

Putting together the effects of the changes in clouds in a warming planet, we find:

1. Total albedo will decrease                            >   warming effect
2. Low cloud will decrease as % of total cloud  >    warming effect
3. High and middle cloud will increase as % do. >   warming effect
4. Cirrus cloud cover will decrease                             cooling effect       

This is support for the findings of Dr Andrew Dessler, who studied the transmission of radiation from clouds and concluded that for every degree C of warming, clouds trap 0.54 +/-  0.74 W/m2 of heat.

Effects 2 and 3 will be lessened by the fact that the total cloud cover will be decreasing in a warmer planet.

There are other factors, such as thickness, roughness, and colour of clouds which could have an effect on the final outcome.

The interesting thing is that it is very difficult to determine what clouds are doing to climate sensitivity at any one point in time, and much effort has rightly gone in to sorting this out. 

However, looking at what happens over a period of time changes the situation, and makes the net effect much more clear.

The Paradox

Consideration of the long term trends of cloud formation has a paradoxical effect on the debate within climate science about clouds.

The consensus view is that that clouds probably have a net positive feedback, so that clouds will tend to amplify any warming that takes place. As the planet warms, this positive feedback will tend to decrease cloud warming effect as total cloud cover decreases, although it will be compensated to some extent by a trend towards warming cloud. These conflicting trends will need to be entered into the models.

On the other hand, the new consideration has a severe impact on the way that opponents of anthropogenic climate change view clouds.

The hypothesis of AGW sceptics is that climate sensitivity is low. In order to support this view, they have focussed on the uncertainty surrounding clouds, and attempted to show that clouds will react to increases in temperature by damping down the increases, acting as negative feedbacks.

There have been two papers this year attempting  to support this argument, one by Lindzen and Choi, and one by Spencer and Braswell. Both papers are criticised here.

Irrespective of the quality of their scientific work, the data given in this post  refutes the sceptics’ hypothesis. If, for the sake of argument, clouds are given a negative feedback, as the observed temperature increases occur, cloud cover, and therefore the negative feedback, will diminish. At the same time the quality of the cloud moves from negative feedback towards positive feedback. 

The conclusion is that the hypothesis that climate sensitivity is low cannot be sustained by arguing that clouds cover will increase to cool the planet.