18 | 03 | 2016

Report: Lecture by Prof Salby 7th Nov 2013

Download this file (salby01-20131107_u2.pdf)salby01-20131107_u2.pdf[Report: Lecture by Prof Salby 7th Nov 2013]1420 Kb

Climate: What we know and what we don't

downloadable version

Professor Salby giving his presentation on 7th November 2013
to the Scottish Climate & Energy Forum.
produced by Mike Haseler BSc. MBA


In order to understand the importance of the evidence presented by Salby it is necessary to understand the case for attributing the recent rise in CO2 to human emissions. This starts with the assertion that man-made, rather than natural, emissions of CO2 can be shown to be the cause of the recent rise in atmospheric CO2 because:

  1. The recorded rise in CO2 from 1958 of about 100ppm is larger than anything apparent in the proxy record. This “unprecedented” rise is seen as a fingerprint of recent human activity

  2. The ratio of carbon 13 to carbon 12 in the atmosphere has decreased since 1830. This was thought to be due to the burning of fossil fuels which have a lower ratio of carbon 13. As such the reduction in the ratio was thought to be the “fingerprint” of man-made emissions.

And then it is argued that this rise in CO2 is causing global warming because:

  1. CO2 and temperature move together in an apparent relationship in the proxy records

In his lecture Salby showed:

  1. Whilst there is a good fit in the ancient record from proxy ice-cores, the measurements of recent global temperature is poorly correlated with the measured level of CO2.

  2. Instead, net emissions of CO2 (not the level) is more closely related to temperature.

  3. If we model surface conditions with temperature & humidity in the atmosphere:

    • net emissions of CO2 can be predicted from surface conditions

    • net emissions of Methane can be predicted from surface conditions

    • net emissions of Carbon 13 can be predicted from surface conditions

  1. Evidence shows that the sources of atmospheric CO2 (as shown by areas with highest concentration) are not related to man-made emissions from burning fossil fuels.

  2. The evidence shows Carbon 13 is not a fingerprint of human emissions.

  3. The IPCC are wrong to say: “all ... increases [in CO2] are caused by human activity.” or “the increased atmospheric CO2 concentration is known to be caused by human activities”.

  4. In significant part, changes in the level of CO2 are controlled by global temperature.

  5. Furthermore he proposed a mechanism to explain the anomaly between the behaviour of CO2 in the actual atmosphere and that seen in the proxy record from the ice core. This was that there was a non-conservative damping mechanism such as diffusion or loss in removal of the ice core.

  6. Non-conservative influences would cause past atmospheric CO2 to be significantly underestimated, so it is likely that the recent rise in CO2 is not unprecedented.

  7. All the recent history of CO2 can be explained from surface conditions alone.


On 7th November 2013, Professor Salby gave two presentations in Edinburgh. The first was hosted by Murdo Fraser MSP in the Scottish Parliament, the second by the Scottish Climate & Energy Forum at the Links Hotel.

This report summarises & explains the arguments & evidence presented at the lecture. Its purpose is to inform members of the Scottish Climate & Energy Forum & other invited guests who could not attend with an aim of highlighting the significant implication on public policy.

Please note:

Whilst every care has been taken compiling this report, it is not intended to be a reproduction of the lecture which was largely scientific in nature. Any omissions or mistakes are likely the responsibility of the author.

In line with the convention for scientific work we will refer to Professor Salby in this work as Salby.


ppm: Parts per million. 100ppm = 0.01%

Correlation & coherence: When we measure two things and their measured values increase and decrease together they are said to be correlated. An example is children's age and height. The better they are correlated, the more a change in one is proportional to a change in the other.

Lag (when referring to correlation): Often, although two things are related, they do not occur at the same time. For example, freezing conditions cause burst water pipes – but usually after the pipes thaw. So, the weather has to warm before the effect of the severe cold shows. There is a time delay between the cause and the measured effect. So, if we plotted burst pipes with temperature, we would find a poor correlation between the number of call outs to plumbers and the current temperature, but perhaps a much better one between the number of call outs and the temperature the day before. This is the lag.

Proxy: We do not have reliable direct measurements of temperature or CO2 before 1958. However we can infer the level of these in the past by measuring other things. A proxy is something that isn't a direct measurement but something from which we estimate the value of something else.

Two proxies are commonly used. It has been found that there are bubbles trapped within ice-cores in areas like Greenland where it has been continually cold, long enough for ice to accumulate over long periods. It is possible to measure the amount of CO2 in the air in these bubbles and this is believed to tell us the amount of CO2 present when the snow that formed the ice fell.

The other proxy is the ratio of Oxygen 16 and Oxygen 18 in the water. Since the ratio of these varies with temperature, this ratio can be used to estimate temperature when the snow formed.

The Case for human emissions causing the recent rise in CO2 & global warming.

Since 1958 when good records of CO2 begin, both CO2 and global temperatures have risen. Two questions arise:

  1. Did the rise in CO2 cause the rise in global temperatures.

  2. Is the CO2 rise caused by humans.

If both are true then it is reasonable to conclude that that humans are causing global warming through man-made emissions of CO2. If not, then this conclusion is false.

The evidence CO2 causes warming: the proxy ice cores.

The main evidence for CO2 causing global warming lies in the proxy measurements of CO2 & temperature derived from polar ice cores.


Fig 1: Graph of temperature (blue), CO2 (green) & CH4 (purple) from 450,000BC to present

This plot shows the proxy values of CO2, methane (CH4) and temperature over time derived from measurements from ice-cores. But note, these proxy values are not the actual value of CO2, temperature or methane, instead they are indirect measurements of the original value. These indirect measurements appear to show that global temperature is high when CO2 & methane are high, and that global temperature is low when CO2 & methane is low. If the CO2 in the ice cores were the same as the atmosphere, it would show that global temperature is related to the level of methane and CO2 in some way.

Therefore it was assumed that there is a direct relationship: that CO2 level causes the temperature to respond. However, more detailed analysis showed that temperature causes the response in the level of CO2.

Moreover if true, this proxy record shows that CO2 levels change from a low of 180 parts per million

in the depths of ice ages to about 280 during the warm periods. This 100ppm change would make the recent change from around 300ppm before industrialisation to 400ppm “unprecedented”. So small range of CO2 in the ice core samples is seen as evidence that the recent change in CO2 is highly unusual. This is taken as proof that the change must be due to human activity.

Carbon 13 - the “fingerprint” of Mankind on CO2

One of the key pieces of evidence for human causation for the CO2 rise is that there have been changes to the carbon isotopes in the atmosphere. Isotopes are different forms of atoms with the same chemical behaviour but with different masses. Carbon has three main isotopes: carbon 12, carbon 13 & carbon 14 (For shorthand these are often written: 12C, 13C and 14C). Carbon 12 is the most common.

Fig 2: Recent evolution of CO2 (green) & the ratio of Carbon 13 to carbon 12 (δ13C in red). This shows CO2 has risen whilst the ratio of Carbon 13 to carbon 12 has dropped

Figure 2 (above) shows that whilst the level of CO2 has risen in the atmosphere, the ratio of carbon 13 to carbon 12 (δ13C) has dropped.

About 1% of carbon in atmospheric CO2 is carbon 13. But, the level in CO2 produced in fossil fuels, burning forests or rotting vegetation, has less carbon 13 than the atmosphere. This is because plants have a preference for the lighter isotope of carbon 12. So things like wood or fossil fuels which are formed from plant material have a lower level of carbon 13. If CO2 from these materials is released into the atmosphere, the the ratio of Carbon 13 to Carbon 12 in the atmosphere would decrease. So figure 2 is thought to show the “fingerprint” of humans: that as we burn fossil fuel, with lower levels of carbon 13, the proportion of Carbon 13 in the atmosphere drops.

Summary of the case

Taken together, the dropping carbon 13 levels, the unprecedented rise in CO2 & the correlation of CO2 and temperature are all believed to be conclusive proof that humans have caused man-made warming. The carbon 13 & apparently unprecedented rise appear to show the CO2 is man-made. The correlation is believed to show that the CO2 rise in turn caused global temperatures to rise; the so called smoking gun.

Are humans responsible for the rise in CO2 and does CO2 drive temperature change?

Are recent increase in temperature driven by CO2?

There were two key graphs in the presentation. The first was this:

Fig 3: Observed record (atmospheric measurements) Absolute value of CO2 in green and global temperature in blue. This shows poor correlation.

It shows a plot of the level of CO2 in the atmosphere (green) and global temperature (blue). This is usually interpreted as the rise in CO2 driving temperature. If true, we would expect the temperature plot to follow the plot of CO2. In contrast, except for the general upward trend, the two have no features in common. The CO2 plot curves upward, in contrast, with no general general warming after 1998, the temperature plot curves downward. The CO2 plot has a yearly cycle which hides other features. The temperature curve is full of sharp variations which cannot be seen in the plot of CO2.

Finding 1: Global temperature is poorly correlated with CO2.

However, the large yearly cycle of CO2 may be hiding fine detail. We can see if there is any detail in the CO2 curve by by averaging over a year to remove yearly changes and then plotting the rate of change. This removes the overall trend and magnifies any small rapid changes. This produces the blue plot in figure 4 below. In contrast to the level of CO2 shown in figure 3, the plot showing the rate of change of CO2 is a very close match to the global temperature curve in green (which has been similarly smoothed).

Fig 4: Plot of CO2 inter-yearly CO2 net emission rate against global temperature


This time, the two show a remarkable high level of correlation: they both share the same constant upward trend which is consistent over the whole period from 1964 to 2008, and almost every peak and trough in CO2 emissions has a corresponding peak or trough in global temperature. So is this the fingerprint showing that changes in CO2 drive global temperature?

No! The reason is that this time we have not shown the level of CO2 ,but net emissions of CO2 . When temperature is higher, there are more net emissions of CO2. When temperature is lower, there is less net emissions of CO2. This is very strong evidence that the rate of emissions is linked to global temperature. This is not what we would expect if CO2 were driving global temperatures.

Finding 2: Measurements of the atmosphere show that it is the rate of change of CO2 , not the level, which is closely related to global temperature.

Finding 3: Measurements of the atmosphere show a very different relationship between CO2 and temperature than indirect estimates of ancient CO2 and temperature from ice-core proxies.

Could human activity changing in response to temperature explain the relationship between net CO2 emissions & temperature?

It is possible that there is some kind of indirect relationship between net global emissions of CO2 and temperature. For example, when the weather is cold we turn on fires. So we would expect emissions to increase when the weather is cold. However the relationship is that emissions increase in warmer periods, so perhaps warmer conditions allow more building or other economic activity that produces more CO2 ? Could human emissions be responding to changes in global temperature and could this be responsible for the increase in emissions when temperatures rise?

Fig 5: World production of CO2.


As figure 5 above shows, the plot of human emissions does not look anything like (the) net change of atmospheric CO2 in figure 4. So human emissions cannot explain the change in atmospheric CO2. The plot is remarkably smooth with nothing to suggest any link to the changes in emission we see in figure 2. Indeed natural fluxes should be more important because as figure 6 (below) shows the human emission of 29 gigatons is a minor flux of only 4% of total emissions. So, even a small imbalance in natural fluxes will dwarf the small human emissions, so one would look at these first for the cause of any change in net CO2 emissions.

Fig 6: Global carbon fluxes of carbon dioxide in gigatons
Source: (Figure 7.3, IPCC AR4).

Finding 4: Net emissions of CO2 to the atmosphere do not show a relationship to global consumption of fossil fuels because natural fluxes are more important.

Is human produced CO2 increasing atmospheric CO2 levels?

As human we burn large amounts of fossil fuel which produces CO2 it would be reasonable to assume that as we burn more or less that these changes would lead to a corresponding change in the level of CO2 , but apparently not. Although Salby did not say so himself, the following two maps more or less demolish this assumption. Figure 7 shows calculated human emissions of CO2 (in red). Figure 8 shows the actual concentration of CO2 as observed by satellite, with likely sources (in red) and likely sinks (in blue).  It is clear that the two maps are very different.


Fig 7: Map showing production of CO2 with highest areas in blue and then red.(source EDGAR)

Fig 8: Map showing concentration of CO2 (highest in red)

Man-made CO2 emissions are highest in the mid-northern latitudes, in a band going from the East USA through Europe down to India and up to China. In contrast, the satellite image shows almost no high concentrations of CO2 in these areas. Instead the high concentrations of CO2 are in a band at the equator. Indeed, the only area of high emissions near the equator in Africa is Nigeria which has conspicuously low levels of CO2. One could even suggest that these maps show that rather than humans increasing CO2, the data would better suit a hypothesis that humans are decreasing CO2! There is clearly no fingerprint of human emissions from fossil fuels in its distribution.

Finding 5: The sources of atmospheric CO2 (as shown by areas with highest concentration) are not related to emissions from burning fossil fuels.

If changes in CO2 levels are not related to human activity what can explain the changes in CO2?

When two things happen together, it is not always easy to work out which is causing which. For example, does the dog wag its tail because it gets a treat, or does the mistress give the dog a treat because it wags its tail?

Common sense can often help, but not always. One possible explanation for rising CO2 is that human emissions are the cause, but this does not fit the evidence of figure 4 & 5 which show that net global emissions of CO2 into the atmosphere are related to temperature and not to human emissions.

So, we need to look for some way to explain the change in CO2. Salby proposes that soil conditions must play a part in the same way that field measurements show that natural CO2 emissions increase when temperatures rise. This is fairly simple to understand. Fungi & plants, like all living things, prefer warmer conditions. So when it is warm, there is not only more growth but faster breakdown of organic material which releases CO2.

Because decomposition in anaerobic conditions tend to produce methane with CO2 , then if decomposition is responsible for the production of CO2 then we would expect to see that the level of methane (CH4.) is also affected by surface conditions.

To validate this hypothesis Salby shows that he can reconstruct the emissions of: CO2 (figure 9); Methane (figure 10), and Carbon 13 (figure 11), very accurately using only two features of surface condition, namely temperature and humidity.

Table 1 (below) shows that proxy CO2 (from ice cores) correlates well with proxy temperature with a figure of about 0.85. But when the actual recent measurements of CO2 are compared with measured temperature the correlation is poor (about 0.5). The correlation improves when we compare net CO2 emissions and temperature (0.63) and very good when we include soil moisture to model surface conditions with correlations between modelled surface conditions and CO2 of 0.93, methane/CH4. of 0.94 and the ratio of carbon 13 to carbon 12 of 0.88.

So this is very strong evidence that it is soil conditions that are causing the changes in net global emissions of CO2 , methane and the change in the ratio of Carbon 13.






Proxy CO2 level

Proxy Temperature


at <1000 years lag

CO2 level



at 10 month lag

CO2 change




CO2 change

Temperature & soil moisture



Methane change

Temperature & soil moisture



Carbon 13

Temperature & soil moisture


Negatively correlated.


Table 1: Correlation of proxy CO2, measured CO2 & change of CO2 with temperature. Correlation of methane and Carbon 13 with temperature.



Fig 9: CO2 emissions compared to global model for surface conditions (i.e. surface temperature and soil moisture). Correlation is 0.93

Figure 9 (above) shows that the emission of CO2 is closely related to surface conditions (surface temperature & soil humidity) achieving a much better correlation than that between actual CO2 & temperature or even proxy CO2 and proxy temperature.

Fig 10: Methane (CH4) emissions compared to global model for surface conditions (i.e. surface temperature and soil moisture). Correlation is 0.94

The correlation of methane with soil conditions shows that variation in the soil conditions are the main cause of the observable change in methane. Because methane is not produced by burning fossil fuels but by decomposition this shows that soil conditions cause these emissions.

Finding 6: There is compelling evidence that the change in CO2 is driven (in whole or part) by temperature & surface conditions.

Fig 11: Carbon 13 emissions compared to global model for surface conditions (surface conditions based on surface temperature and soil moisture are shown inverted).

Finally the correlation of carbon 13 with soil conditions (shown inverted) shows that changes in carbon 13 are a result of native sources affected by soil properties.

Figure 11 is particularly important because it shows that changes in the proportion of carbon 13 in the atmosphere are strongly related to surface conditions. This contradicts the assertion that the change in carbon 13 is caused by man-made CO2 emissions. As the decrease in carbon 13 in the atmosphere in figure 2 was taken as proof of human emissions, this graph shows that this assertion is wrong.

Finding 7: Carbon 13 is not a fingerprint of human emissions. Instead its production (at least in part) relates to surface conditions.

Is there other evidence the change in CO2 is caused by global temperature?

If one thing causes another, then it follows that if there is a time delay between the two, then the cause will occur before the effect. Figure 12 below shows how well the temperature measurements match the increase in CO2 if the two are compared for various time differences.

Fig 12: Correlation between CO2 and temperature in the observed record showing maximum correlation where temperature leads CO2 level by 10months.

This shows that the closest match occurs when the CO2 lags the temperature by about 10months. Taken together with the graphs showing a correlation between net emissions and temperature this is very strong evidence that global temperature causes the change in CO2 emissions.

Finding 8: A significant part of the change in the level of CO2 is controlled by global temperature.

Finding 9: The IPCC are wrong to say:
“All these increases [in CO2 from pre-industrial times] are caused by human activity”
or: “The increase in atmospheric CO2 concentration is known to be caused by human activities because the character of CO2 in the atmosphere, in particular the ratio of its heavy to light carbon atoms, has changed in a way that can be attributed to addition of fossil fuel carbon.” (IPCC AR4)
In fact this is impossible as Salby put it.

How can we explain the anomalous relationship between proxy CO2 and temperature and actual CO2 and temperature.

Having shown that the rate of actual net CO2 emissions into the atmosphere correlates with surface conditions, we still have left the anomaly that the level (not net emissions) of CO2 in the proxy record from ice-cores correlates with proxy temperature. There is strong evidence from recent direct measurements that temperature causes the emission of CO2 readings. But, so long as we have the unexplained anomaly, it is possible that the anomalous position is that seen today rather than that seen in the proxy record of CO2 from ice cores.

What Salby proposes is in principle very simple: that ice-cores do not correspond to actual levels of CO2 at the time the snow was formed.

To understand this it is worth recapping how CO2 is trapped in and then obtained from ice. Although snow is frozen water, it is “fluffy” and has lots of air spaces. As more and more snow settles, this air gradually becomes trapped as the snow compacts to form ice. Eventually the air is enclosed by ice and bubbles form in the ice. The assumption is that this air remains completely unchanged for thousands of years in the ice, so that when the ice is drilled to remove a core, the ice can be melted and the air collected. If this air is completely unchanged it will tell us the amount of CO2 in the air at the time the snow fell. In addition the ice can be sampled for the ratio of Oxygen 16 and Oxygen 18 which indicates temperature when the ice was formed.

However, by showing that the relationship between CO2 and temperature is different in the ice core compared to current measurements, Salby shows that this simple assumption may be wrong. That the air in the ice-core may not be a true sample of the air at the time the snow fell.

The hypothesis Salby proposes is that CO2 is lost from these bubbles in some way. Either this is some mechanism that reduces CO2 entering the ice; or that CO2 diffuses slowly through the ice column so that it is slowly lost, or that it is lost at the time of extraction (see below).

Potential Mechanisms that may mean the level of CO2 from ice cores does not represent the level in the atmosphere when the snow fell are given below. They are illustrative & were not in the lecture.

  1. Ice algae: certain types of algae are known to live on snow (Wikipedia: Ice Algae). Their photosynthesis will reduce the level of CO2 in the surface layers causing the entrapped gas to lack CO2. As they float to the surface they bring the carbon with them to the surface. Eventually they are washed out with spring & summer ice-melt depleting the snow of carbon dioxide.

  2. Most materials have micro-pores or defects which permit the movement of gas. Ice that is subject to movement (as most ice sheets are) will be stressed which may allow gas to diffuse away from its original site.

  3. At pressures above 10 atmosphere (about 100m representing about 10,000 years of ice), CO2.size: small;"> will condense out of the gas bubbles to form a liquid. (Wikipedia: CO2.size: small;"> Clathrate)

    The behaviour of ice is complex and even below freezing, the lack of bonds for surface molecules means there is a thin layer of water at the surface. This is what makes ice slippery. The thickness of the liquid-like layer on ice is only about 12 nm at −24 °C and 70 nm at −0.7 °C. But being liquid it will flow, albeit very slowly, from top to bottom causing the bubble of air to rise, leaving the richer, denser CO2.size: small;"> liquid behind. The rate of rise may be insignificant in terms of moving air within the column, but it will still act to separate out the liquid CO2.size: small;"> form the gaseous components. When drilled, because the gases in the ice are under extreme pressure, the sudden release of pressure causes enormous stress and micro-fractures form which allow out-gassing. When the fractures permeate bubbles some, but not all the air escapes, but when they reach the liquid CO2.size: small;">, the higher density means more CO2.size: small;"> escapes. If this happens the sample will be depleted of CO2..

Although Salby did not suggest a specific mechanism, he suggested that the evidence shows that a non-conservative process was causing damping of the proxy CO2 readings. This would explain the anomaly between actual measured CO2 and CO2 in the proxy record. The evidence for this is in the form of the correlation curve as we compare the proxy temperature against the proxy CO2.

Fig 13: Correlation between CO2 and temperature in the proxy record (blue) and theoretical model with damping (red)

What we would expect in the above graph if the proxy CO2 level were only related to the temperature at the time the ice in the ice-core formed, is that the graph would be very narrow (too narrow to show). This is because the curves would only match when we are comparing the temperature at a particular date to the CO2 at that same date. In contrast, the measured curve (blue) in figure 13 is very broad with correlation of tens of thousands of years. This shows that the proxy CO2 and temperature match over a very broad range of offsets. It is as if we are comparing a blurred image with the original scene. The blurring means we are not quite sure where it fits so there is a range of positions where they could match.

A non-conservative “damping” process like diffusion would cause this to happen. We can model this, and a match in their behaviour is strong evidence for such a damping process (although we do not know exactly what it is). To test this, Salby created a theoretical model and used this to calculate what the correlation would look like if a non-conservative damped mechanism were present. (red curve). The two curves are a very close match showing that some form of non-conservative process is a very good explanation for the anomaly. The reason for the anomaly is because the proxy record of CO2 is not an accurate record of past CO2. Also note how the actual curve is lopsided toward a positive lag. This is further proof that temperature leads CO2 in the proxy record.

Finding 10: The anomaly between proxy CO2 and actual CO2 can be explained by a non-conservative damping mechanism such as diffusion within the ice core.

If there is a non-conservative process causing damping in the signal, this process would cause the level of ancient CO2 to appear much lower in the ice core than that in the atmosphere when the ice formed. So, the evidence for this damping process is evidence the ancient CO2 levels have been underestimated. As such it is not possible to say whether the present rise in CO2 is unprecedented.

Finding 11: Because there is evidence the CO2 in the ice core does not represent actual CO2 , it is not possible to say whether the recent 100ppm rise in CO2 is unprecedented.

Can the temperature induced CO2 emissions explain all recent changes?

Whilst the presentation more than ably demonstrated that temperature induced emissions of CO2 must be responsible for part of the increase in CO2 seen in recent times, there remains the question of how much could still be due to human emission. Salby did not address this question directly but instead showed that all the recent rise in CO2 could be explained by emissions due to surface conditions.

Fig 14: Modelled CO2 net emissions (blue) from surface conditions compared to actual CO2 (green).

Figure 14 compares the net CO2 emissions (blue) calculated by integrating the sum of net emissions derived only from surface conditions (surface temperature and soil moisture). This tracks actual CO2 closely showing that this model can explain the entire recent history of CO2 (except the yearly swings which were not modelled).

Fig 15: Modelled (blue) versus actual (mauve) net emissions of methane. Modelled from surface properties


And as figure 15 shows, surface conditions also closely predict the net methane (CH4.) emissions. As methane is not produced during combustion of fossil fuel, this is further strong evidence that surface conditions are responsible for these emissions. However, the above reconstructions only use satellite temperature data which is not available much before the 1980s. To show that this model could explain all the recent history of CO2 since records began in the mid 1800s, Salby used the less reliable surface thermometer network to produce the following plot, this time without including soil moisture.

Fig 16: Plot of modelled CO2 in the atmosphere using satellite (mauve) and surface thermometers (red) against CO2 from direct measurements (light green) and ice-core proxies (red). Due to the uncertainty in surface temperature measurements there is a corresponding predicted range of modelled CO2 (pink shading)

Figure 16 (above) shows that the modelled level of CO2 from temperature alone using satellite (mauve) and surface thermometers (red) closely matches the measured CO2 (light green) as well as the less robust level of CO2 from the ice core proxies. It is all well within the range of uncertainty due to the uncertainty in surface temperature data.

Finding 12: All the recent history of CO2 can be explained entirely from surface conditions alone.

Finding 13: Man-made CO2 emissions are not necessary to explain recent changes to CO2 levels in the atmosphere.


Anon Carbon dioxide clathrate - Wikipedia, the free encyclopedia [Internet]. Available from: <http://en.wikipedia.org/wiki/Carbon_dioxide_clathrate> [Accessed 12 November 2013a].

Anon Ice algae - Simple English Wikipedia, the free encyclopedia [Internet]. Available from: <http://simple.wikipedia.org/wiki/Ice_algae> [Accessed 12 November 2013b].

EDGAR EUROPA - EDGAR - Carbon dioxide (CO2) [Internet]. Available from: <http://edgar.jrc.ec.europa.eu/part_CO2.php> [Accessed 12 November 2013].

IPCCFAQ 7.1 - AR4 WGI Chapter 7: Couplings Between Changes in the Climate System and Biogeochemistry. Available from: <http://www.ipcc.ch/publications_and_data/ar4/wg1/en/faq-7-1.html> [Accessed 12 November 2013a].

IPCC Figure 7.3 - AR4 WGI Chapter 7: Couplings Between Changes in the Climate System and Biogeochemistry [Internet]. Available from: <http://www.ipcc.ch/publications_and_data/ar4/wg1/en/figure-7-3.html> [Accessed 13 November 2013b].