How do we know where the carbon is coming from?

by Paul Braterman

CarbonScripps_Institution_of_Oceanography _2011In 1957, Charles Keeling of Scripps Institution of Oceanography began regular measurements of carbon dioxide concentration at Mauna Loa, Hawaii. By 1960, he was already in a position to report a steady increase, together with seasonal variations. In the northern atmosphere, CO2 concentration falls during the spring and summer growing season, but recovers during autumn and winter as vegetable matter decays. This sawtooth pattern is superposed, however, on a steady overall increase.

Above, R: Scripps Institution of Oceanography (Invertzoo via Wikipedia)

The Keeling curve and beyond

Charles Keeling died in 2005, but the work is being continued by his son Ralph. When I visited Scripps in 1995, I saw Charles Keeling's original curve, ink on graph paper, on the wall in the corridor outside his office. That curve has now been designated a National Historic Chemical Landmark, and there are commemorative plaques both at Scripps and at the Mauna Loa Observatory. Charles Keeling's original paper, freely available here, goes into meticulous detail regarding sample collection, calibration, precautions taken to prevent local contamination, and comparisons between the Mauna Loa data and those that numerous other sites, including the Antarctic and samples collected from an aircraft.


L: Atmospheric CO2, 1700 – 2014; NASA via Forbes. Click to enlarge. Note that the zigzags for atmospheric data are not error bars, but annual fluctuations.

By 1985, the record had been extended backwards in time by analysis of air bubbles trapped in ice cores, with dates ranging from the 1980s to the 1600s and earlier. These dates overlap Keeling's data, and take us back to pre-industrial times. Before long, the ice core record had been extended to an 160,000 years, taking us into the Ice Ages, while further work has pushed it back to 800,000 years. We have estimates going back far beyond that, but employing indirect methods and with higher uncertainty.

During the Ice Ages, carbon dioxide played a dual role, as product and as agent. The temperature oscillations at this time were driven primarily by subtle changes in the Earth's motion (so-called Milankovitch cycles). But carbon dioxide is less soluble at higher temperatures (which is why your carbonated drink fizzes inside your mouth). And so in the first place the rise and fall of temperature led to a rise and fall of carbon dioxide in the atmosphere, as the oceans released and reabsorbed the gas. But then, the changes in carbon dioxide concentration amplified the original effect, since more carbon dioxide acting as a greenhouse gas makes the Earth lose heat less efficiently into space.

To summarise the results, current levels of CO2 are the highest they have been for over twenty million years. In the centuries leading up to 1800, levels were steady at 280 parts per million (ppm); a slow but steady increase took place throughout the nineteenth and early twentieth century, so that levels had reached over 310 ppm when Charles Keeling began his studies; this increase has accelerated steadily since then; the present value is over 400 ppm; and the current rate of increase appears to be unprecedented in the geological record.

So where is it all coming from?

There are three great reservoirs of carbon that we need to consider. The first one is what we might call circulating carbon. This includes the carbon dioxide in the atmosphere and that dissolved in the oceans. It also includes the carbon in living things. In photosynthesis, plants convert the carbon in atmospheric carbon dioxide into organic material, releasing oxygen. But sooner or later, nearly all this organic material gets re-converted to carbon dioxide, as the plants themselves or the organisms that feed on them use this organic material as food. Next, there is the carbon in non-fossil minerals, mainly as calcium carbonate in chalk and limestone. Finally, there is buried biologically processed carbon. This includes all deposits of coal, oil, and natural gas. In addition, it includes enormous amounts of buried organic matter finely dispersed in sediments. Truly enormous amounts. All the oxygen in the Earth's atmosphere was produced by photosynthesis from carbon dioxide, meaning that over the course of the Earth's history, an equivalent amount of carbon has been buried in the sediments.

Over very long timescales, there are links between these three reservoirs. When living things die, most of their carbon is converted by scavengers or bacteria to carbon dioxide, but some gets added to the pool of buried organic material. Carbon dioxide is being slowly removed from the atmosphere by the weathering of rocks; to put it simply, silicate-containing rocks react with carbon dioxide to produce carbonate-containing rocks and silicon dioxide (the silica in sand). Sediments containing buried fossil carbon are slowly dragged back into the Earth's mantle by the conveyor belt of plate tectonics, while atmospheric carbon dioxide is continually replenished from the mantle by volcanoes.

So which of these pools is responsible for the added CO2 in the atmosphere, or is it, perhaps, more than one? The obvious culprit is the burning of fossil fuels, coal, oil, or natural gas, and the amount of carbon dioxide that we have produced in this way over the past two centuries is more than enough to account for the increase in the atmosphere. Indeed, the only reason that atmospheric CO2 is not already way above 500 ppm, is that roughly half this added carbon dioxide has ended up in the oceans. This is making the oceans measurably more acidic (or less alkaline). Since the added CO2 tends to dissolve calcium carbonate,1 this is going to make life more difficult for all the organisms, from corals to cuttlefish, that incorporate calcium carbonate in their structure.

So the arithmetic strongly implicates fossil fuels. But no scientist would be happy to let so momentous a conclusion rest on one single line of evidence. How can we really be sure that there is nothing else going on, such as a change within the circulating pool, or increased weathering of carbonate rocks, or something else we haven't even thought about?

Three kinds of carbon atom

We can find out more by looking at the abundances of the three isotopes of carbon that occur in nature. Isotope: Greek isos, same, topos, place; same place in the periodic table. There is carbon-12 (12C for short), the most common form, generated in stars by the fusion together of three helium-4 nuclei in red giant stars; this makes up just under 99% of the carbon here on Earth. There is carbon-13 (13C), formed late in the life of such stars by addition of a neutron to carbon-12. And finally there is carbon-14 (14C), the isotope used in radiocarbon dating. All carbon atoms contain six protons in their nucleus, and six electrons outside it, and these electrons dominate the chemistry. 12C has six neutrons in its nucleus; 13C, seven; and 14C, eight. 12C and 13C on Earth are stable, while 14C is radioactive, with a half-life of 5730 years.2

Real bombs, fake wines, and radiocarbon dating

Carbon-14 is unique among the Earth's radioactive isotopes. The others all come from radioactive materials dating back to the formation of the solar system itself. Any such ancient 14C would have long since disappeared. Fresh 14C, however, is being generated on this planet all the time by cosmic ray bombardment of nitrogen-14 in the upper atmosphere. That 14C, in the form of carbon dioxide, then mixes with the lower atmosphere and the oceans, where is taken up by plants through photosynthesis and hence by animals, only to be breathed out again. So the 14C in the reservoir of circulating carbon dioxide is continually being replenished. At the same time, it is being removed by radioactive decay, and so it builds up a steady concentration, where the amount being produced is equal to the amount being removed. As long as an organism is alive, its carbon is part of the circulating pool. But once it dies, the carbon is withdrawn from the pool, and starts to decay, with a half-life, as I've mentioned, of 5,730 years. This is what is used in radiocarbon dating. There are details and refinements which might form part of a later post. However, all we need to notice here is that if carbon contains no more than trace amounts3 of 14C, it must been out of circulation for tens of thousands of years, and perhaps much much longer.

Over the past decades, there have been major disturbances to the 14C budget. One of these was aboveground nuclear testing. This added large quantities to the atmosphere in the decade leading up to the 1963 partial nuclear test ban treaty. France and China continued aboveground testing after that date, but the much more extensive US and USSR atmospheric tests peaked around 1960 before coming to an end in 1962. The test ban treaty was not popular in all quarters. In 1952, because of his agitation in favour of the ban, the State Department refused the chemist Linus Pauling a passport. However, they did restore it in 1954, in time for him to go to Stockholm to collect a Nobel Prize for his contributions to chemistry. In 1960, to universal approval, Pauling was to collect a second Nobel Prize, this time the Peace Prize, for his political activities.

R: You can't judge a wine by its bottleCarbonWinebottle-50573_960_720

If you want to study the year-on-year variation of 14C effects, you need recent samples with well-attested dates. There was much good-humoured teasing when Murdoch Baxter, my colleague at Glasgow University, started collecting samples of vintage malt whiskies and French wines from the bottlers. But there was good reason to do so. The alcohol in these came from grapes and barley of known date, a sample of that year's distribution of carbon isotopes in the atmosphere. The results were striking; there was a steady increase from 1955 through 1965, when atmospheric 14C reached almost twice its pre-test level, followed by a decline. (One odd surprise, though; a much-prized wine labelled 1918 proved to date from 1963.)

Much of the 14C from bomb blasts would have been carried up into the stratosphere, taking a few years to mix with the atmosphere at ground level, hence the time lag from 1960 to 1965. As for the decline, it is much more rapid than the decline due to radioactive decay, and by 2010 atmospheric 14C was back to within 10% or so of its pre-bomb levels. There are two sources for this decline. One is circulation between atmosphere and oceans, and to a smaller extent between atmosphere and living things, within what I have called the circulating reservoir, and this was the main effect until about 1990. The other is the addition of carbon dioxide to the atmosphere, which is decreasing the concentration of 14C as surely as adding water to whisky decreases the concentration of alcohol. And the rate of decrease shows that the added CO2 itself is itself free from 14C, since otherwise the decline would be much slower. This tells us that the added CO2 has not come from a redistribution within the circulating pool. But that still isn't enough to tell us whether it came from fossil fuels, or from the mineral pool. For that, we must turn to the testimony of the other two isotopes.

Lies your teacher told you

IMG_6843L: The carbon in this rosebush is measurably different from the carbon in the atmosphere. Image by author

You will probably told at school that different isotopes of an element have different physical properties (an atom of the heavier isotope weighs more), but identical chemical properties. Like so many of the things you were told at school, this is not quite true. Consider 12C and 13C; they both have the same number of protons (6) in the nucleus, and that same number of electrons (6 again) outside the nucleus, and these are the electrons involved in chemical bonding. The only difference is that 12C has 6 neutrons in its nucleus, while 13C has 7. But as a result of quantum mechanics,4 this does give rise to a very small difference in their chemical behaviour. 13C tends to settle down in situations where it is most tightly bonded (in carbon dioxide, for example), and to react slightly more slowly. As a result, biological carbon is slightly depleted in 13C. The difference is not great, and the shifts in relative abundance are around 20 to 25 parts per thousand,5 but modern mass spectrometers can readily measure these. Importantly, fossil fuels including coal and petroleum inherit this difference. So isotopic analysis of atmospheric carbon dioxide should tell us about its origins.

Keeling's original data, now verified many times, provide a check on our reasoning. Remember that atmospheric carbon dioxide decreases, because of plant uptake, during the growing season. The carbon dioxide that has disappeared to produce biomass will be relatively depleted in 13C, so that what remains should be relatively enriched, and the 13C/12C ratio in the atmosphere should increase. During the winter, when more organic matter decays than is produced, the trend should be reversed. This is exactly what is found.

The culprit identified

We can now distinguish between the three possible sources of added CO2. We can immediately excludes the circulating pool, because the added CO2 contains no 14C. Of the remaining two possible sources, carbon dioxide from fossil fuels will be depleted in 13C relative to a mineral standard, while carbon dioxide from mineral sources will not. So the question is, has atmospheric carbon dioxide become more depleted in 13C over time, as its amount has risen?

Unambiguously yes. We have been following the process in real time since the late 1950s. We have extended the record back a thousand years using air trapped in ice cores, and have verified the change by examining the carbon isotopes in tree rings, which we can date by direct counting. (Tree ring carbon is of course depleted in 13C relative to atmospheric CO2, but it is depleted by a constant amount, so changes between rings match changes in the atmosphere). What we find is that over the past two hundred years, after having been nearing constant for centuries or more, atmospheric CO2 has become progressively more depleted. And the degree of depletion is exactly what we would expect if the added CO2 comes from an organic source. But remember that we have already excluded any living source, because of the absence of added 14C. That leaves fossil fuels as the only possibility.

One last piece of evidence

One final check, using a completely independent method developed around 1990 by Ralph Keeling, Charles's son, as his Ph.D. project.

If we are generating CO2 from mineral sources such as limestone, this should not affect atmospheric oxygen, since carbon is already fully bonded to oxygen. But if we are generating it by burning organic material, either rapidly as fuel, or more slowly through metabolism, then for every added CO2 in the atmosphere there should be one less oxygen (O2) molecule. The trouble is, the atmosphere is something like 21% O2, but the kind of change we are talking about is only a few parts per million. So we would need a way of measuring O2 concentration to very high accuracy.

The method used is interferometry, a technique that measures extremely small changes. This was the method used by Michelson and Morley in their famous attempt to measure the speed at which the Earth was moving through the light-carrying ether (answer; no such thing). More recently it was in the news for its use in detecting gravitational waves by way of the extremely small distortion that they caused in distances. Here the change measured relative to a fixed standard depends on the difference in refractive index (the extent to which the speed of light is modified) between oxygen and nitrogen, which differ in the fifth decimal place. The results show the expected variation. An increase in carbon dioxide corresponds to a decrease in oxygen, and this is true both the seasonal variations, and of the underlying trend. Scripps is now host to a research group dedicated to monitoring such oxygen fluctuations at a number of sites throughout the world.


O2 and CO2 at Mauna Loa, as published by Scripps O2 program

Why do we need to monitor at several different sites, and why do we need to monitor both oxygen and carbon dioxide? Because plants grow, winds blow, and ocean currents flow. Because oxygen is released, and carbon dioxide absorbed, by plankton in the oceans. Because circulation mixes the atmosphere in different places, and at different levels, on a timescale of years. Because there is exchange on the hundred year timescale between deep ocean currents and surface water, and because the surface water exchanges both oxygen and carbon dioxide with the atmosphere. Because this last exchange, as already mentioned, increases the acidity of the oceans at a rate unprecedented for at least many millions of years, with serious possible implications for everything that lives there. Because we need to understand more fully what we are doing to the planet in order to develop sensible policies for controlling it. And because every major industrial or industrialising nation on Earth, including every member of the G20, recognises the need for such policies.

With one exception.

For more of Paul Braterman's writing, see Primate's Progress.

This blog post goes into more detail than the skeptoid podcast on the subject, which triggered it. Yet more detail and references to this specific topic at realclimate. An excellent general source for information about carbon dioxide and the evidence for human-caused climate change, with detailed analysis of claims to the contrary, is Skeptical Science.

I thank Eric Steig of the University of Washington, Dana Nuccitelli of Skeptical Science, and Brian Dunning of Skeptoid, for useful discussions.

1] We can write chemical equation for how excess carbon dioxide tends to dissolve calcium carbonate:

CaCO3(s) + H2O + CO2 = Ca2+ + 2 HCO3-

Or, in words,

Solid calcium carbonate + water + carbon dioxide gives soluble calcium bicarbonate

Actually, many organisms (including cuttlefish and coral) build their structures out of aragonite, a tougher but slightly more soluble form of calcium carbonate than the calcite found in limestone, which makes things worse.

2] So after one half-life we have one half the original amount of material; after two half-lives, one quarter; after three, one eighth and so on, down to roughly one thousandth after ten half-lives, which takes us to the limits of detection. But all that matters for our purposes is that the amount of 14C, in all material that has spent many tens of thousands of years outside the circulating atmosphere-biosphere-ocean pool, is negligible.

3] It has been known for a long time that coal, even several hundred million years old, contains very small amounts of 14C. This is thought to arise from radioactive bombardment of nitrogen present in the coal. As we would then expect, the amount of 14C depends on the amount of nitrogen, being greater for soft coal than for anthracite, and undetectable in graphite formed alongside the coal.

4] This is a result of what is called zero point energy, energy trapped in molecular vibrations even at absolute zero of temperature. Higher mass, same forces, lower vibrational frequency, lower zero point energy. In CO2, carbon makes very stiff bonds to oxygen, which are more difficult to disrupt for the heavier isotope.

5] I.e. around 20 parts per thousand of the 1.1% 13C, or roughly 22 parts per hundred thousand of the total. Some smaller variations depending on the kind of plant, but these are not enough to affect the present argument. There is also slightly diminished uptake of 14C into plants, relative to the atmosphere, but radiocarbon dating takes this into account by using organic material, not the atmosphere, as its reference point.