The Carbon Cycle

Robert Rohde, who is lead scientist for Berkeley Earth, has created a really nice illustration of the carbon cycle. It shows how the CO2 cycles between the different carbon reservoirs, and how our emissions have perturbed the carbon cycle so that atmospheric CO2 concentrations have risen from around 280ppm in the mid-1800s to over 410 ppm today.

If we were to cease emitting CO2 into the atmosphere, then CO2 would continue to be taken up by the natural sinks, and the atmospheric concentration would actually drop. However, as I discuss in this post, there is a limit to how much can be taken up by the natural sinks. As a consequence, between 20% and 30% of our emissions will remain in the atmosphere for thousands of years.

As Robert Rohde pointed out on Twitter, the process that ultimately draws down atmospheric CO2 is sedimentation, which is very slow. It will probably take more than 100000 years for atmospheric CO2 to return to pre-industrial levels. We’ve essentially perturbed the carbon cycle so that atmospheric CO2 concentrations will remain elevated for a very long time.

Credit: Robert Rohde

This entry was posted in Climate change, Research, Science and tagged , , , , , . Bookmark the permalink.

40 Responses to The Carbon Cycle

  1. Uli says:

    A very nice video to show the flow of carbon.
    Of course there are some simplification. While the ocean carbon cycle is in detail the biosphere is summarized with soil carbon.
    The fossil fuel deposits seems to show reserves only. Totally, including resources, the carbon stored there is about ten times the showed value. Which is one of the main reasons to worry about future emissions. So this underestimates the risk due burning all fossil fuels (including resources) by a factor of ten.

  2. One word: Awesome!

  3. David B. Benson says:

    However, we could hasten the drawdown of carbon dioxide. One way which would certainly work is to plant trillions of trees.

    As the Sahara desert was once green and indeed currently contains the vast Siwa Oasis we know that it largely suffices to merely add water; for example acacias are nitrogen fixers.
    So consider “IrrigatedAfforestation of the Sahara desert and the Australian outback to …” by Ornstein et al. The pdf is freely available via the abstract.

  4. Uli says:

    In addition:
    My first comment looks too negative, because I’ve concentrated on the weaknesses.
    The video is really great! It’s a wonderful visualization of the carbon cycle.

  5. Great video, I suspect I will me making use of it!

  6. Everett F Sargent says:

    Data source(s)? I checked teh twit so if it is there then I missed it.

    From the last frame I get (2017 – PI nets in GtC)…
    424 (FF) = 276 (A) + 165 (O) – 18 (L) = 423 GtC (so close enough for its purpose)

    That appears to give a mass airborne fraction of 276/424 = 65% (on a mass basis). I’m a little slow, so does this mass ratio comport with the volume ratio (eg 1 ppmv = x.xx GtC).

    I was thinking that the airborne fraction was something like 40-50%.

    OK, so 1 ppmv CO2 ~ 2.13 GtC

    So 65%/2.13 = 31% (which now appears to be somewhat low)

    Otherwise, a very nice animation of the carbon cycle.

  7. “As Robert Rohde pointed out on Twitter, the process that ultimately draws down atmospheric CO2 is sedimentation, which is very slow. “

    Sedimentation in the ocean, right? Or more accurately, the fat tail of diffusion of CO2 into deeper and deeper layers of ocean water where it is effectively sequestered very slowly over the span of thousands of years.

  8. looks right. We should make a website about this!

  9. Steven Mosher says:


  10. sbm, What kind of web site are you thinking of? Would you like interactive capabilities?

  11. Dave_Geologist says:

    It’s the weathering that’s slow Paul. Once in the rivers, it will reach the sea in years or less. Once it’s in the sea it’s (largely) gone that day. Although, as I pointed out in the last thread, there is a trade-off between silicate weathering taking up CO2, and enhanced erosion due to a supercharged hydrological cycle exposing fossil and subfossil carbon to oxidation. Palaeocene–Eocene Thermal Maximum prolonged by fossil carbon oxidation. Ideally we’d have a bit of enhanced weathering due to stronger carbonic acid in the rain, and a bit more hydrologic activity, but not too much or it could actually “contribut(ed) to the delayed recovery of the climate system for many thousands of years”.

  12. Everett – I suspect you need to include cumulative land use change emissions in the denominator.

  13. no, I was just kidding around and wanting to sign on to read comments by my peers and betters. But, if we could make an interactive website that would suck CO2 out of room air in response to fingers hitting keystrokes, I think we could fix this thing. I need to patent that idea

  14. izen says:

    It is an excellent representation of the amounts and flux rates out of the sources, and sinks in the Carbon cycle.

    But at the risk of being a cynic, or curmudgeon, please consider what impact this would have on anyone who has little or no knowledge of the carbon cycle.
    Who is ideologically resistant to ‘scientists’ with their complex diagrams and explanations.
    It conveys information to an audience that either already has a good background knowledge of the subject, or is willing to study and educate themselves about the meaning of the various boxes and all the little moving dots.

    I make no claims for the accuracy or correct representation of the complexity of the system in this graphic below.
    Just that this is a visual metaphor that because of its shortcomings is simple enough to communicate the underlying basic concept.

  15. two great animations ;o)

  16. Dave said:

    “It’s the weathering that’s slow Paul. “

    I generated this animation this morning.

    The random walk diffusion is downward into deeper ocean water. The depth at which the front proceeds goes according to Fick’s Law as square root of time, which means that at first the front proceeds quickly but then progressively slows down. Note that some CO2 molecules appear to disappear at the bottom but then pop back up. That’s the “fat-tail” nature of random walk and describes the sequestering behavior of CO2 into the ocean.

    Note: This is 1-dimensional random walk so you can keep track of individual molecules more easily.

  17. izen says:


    I keep encountering the hypothesis that the sequestration of the high CO2 levels after the PETM was at least helped along by Azolla. That the growth of this fern on freshwater lakes and its subsequent sinking into a sediment layer was a key mode of rapid removal. Apparently its ancient role as a rice paddy fertiliser, and possible future method of CO2 atmospheric removal are under investigation.
    As this biological process would have preferentially used C12 over heavier isotopes, would this have had an impact on the isotope ratio ?

  18. Keith McClary says:

    @Paul Pukite
    Marine sedimentation is thought to be mostly phytoplankton shells and feces (plus ~10% terrestrial organic matter) sinking to the seafloor.

  19. Keith McClary says:

    I hope he will provide sequels showing the carbon cycle up to (say) 2100 based on BAU and FF reduction scenarios.

  20. Dave_Geologist says:

    Paul and Keith: yes, more fully, simply washing quartz grains into the ocean doesn’t sequester carbon. Carbonic acid reacting with more complex silicates, the products washing into the ocean with that carbon fixed in inorganic form, then carbonate-producing organisms making shells, tests or skeletons which sink to the seabed and are buried, is the full sequestration mechanism. The enhanced hydrological cycle speeds that up, as can other unrelated things like elevation of mountain belts in suitable climatic zones. That may or may not fit a simple diffusion model, but the physics and chemistry are quite different. I tend to go back to my physical chemistry teaching that all processes which lack memory generate similar-looking first-order kinetics. The map may look like the territory, but the map is flat and made of paper and ink, whereas the territory is 3D and made of rock, soil, plants, cities etc. To me, the territory is much more interesting than the map.

    Some other things to consider would be the impact of flooding or un-flooding continental shelves as sea level rises and falls as a result of temperature change. You’d be replacing forests or grassland with shallow seas or vice versa, with concomitant impact in the balance between photosynthesis and respiration, carbonate deposition or solution (karstification and leaching), burial of dead plants as peat vs. washing around in the ocean, exposure of fossil peat to oxidation, etc. I doubt if that level of detail is covered in global carbon-cycle models, at least areally as opposed to in planetary boxes. There are interesting differences between the isotopic anomalies and occurrence of euxina between pre-land-plant and post-land-plant glaciations. In the Ordovician glaciation, the exposed shelves were barren and there was a loss of primary productivity, With land plants present, the situation is more nuanced, and subject to a balance of competing and interacting factors.

  21. Dave_Geologist says:

    izen, from a quick search (Google Scholar Azolla PTEM) it looks like the Azolla blooms were later than the end of the PETM, and speculation is more about them driving long-term cooling since the Middle Eocene. And they seem to have required very specific continental and oceanic configurations, which probably don’t apply now. As a form of geoengineering, I’m not sure the neighbours would fancy having the world’s lakes turned into algal soup. I would see the end PETM C isotope change as a reversion to normal following a perturbation. Whatever drew down the pulse of organic carbon would reverse the initial isotopic anomaly.

    I should have mentioned, above, that direct sequestration of newly formed organic carbon can also contribute to CO2 drawdown. That’s another trade-off which will depend on topography and climate at the time. Lyons et al. found that erosion of fossil carbon, and exposure long enough to oxidise some of it, outweighed any enhanced washing of fresh carbon into deep-water sediments. The latter can happen though. Thirty years ago I was involved in a possible revitalisation project for the Bombay High oilfields. We walked away for engineering and commercial reasons – too much old, poorly maintained kit and poor commercial terms. One thing we did look into was the petroleum source system, to assess the untapped potential. It’s an oddity in that it produces waxy crude, but the source rocks are the usual prodelta shales. Waxy crude typically has a terrestrial source, because land and very shallow marine or lake plants protect themselves from dehydration and UV with a waxy cuticle (some of the UV protection is really cool, they use iridescence like beetle wings, but for UV!). The source was land-plant debris washed into the sea and rapidly buried, presumably due to the combination of big, fast-flowing rivers in a mountainous, monsoon climate with a forested hinterland. In that specific setting, the Indus is (or at least was) a carbon sequestration mechanism.

  22. Dave_Geologist says:

    Contingency comes into play in examples like the Indus Delta (h/t Stephen Jay Gould!). The Ganges Plain contains several Gondwanan coal basins. Had they been caught up in the Himalayas, it’s likely that Lyons-style erosion of fossil carbon would have won out and the system become a carbon source rather than a carbon sink.

  23. Dave said:

    ” That may or may not fit a simple diffusion model, but the physics and chemistry are quite different. I tend to go back to my physical chemistry teaching that all processes which lack memory generate similar-looking first-order kinetics. “

    Lots of these words are missing the salient points of Rohde’s simulation, and my own contribution above.

    First, I don’t know if this was his intent, but Rohde’s simulation illustrates the role of higher-order compartment models, which if looked into detail result in fatter tails than the simplistic first-order kinetics that you refer to. Each compartment showing a flow of CO2 qualitatively adds an extra damped exponential to what is referred to as the Berne model for CO2 sequestering.

    Second, diffusion is always, and by definition, higher-order kinetics. Any simulation of diffusion involves the creation of a large number of slabs that adds an extra order o flow for each slab. This automatically generates a fat-tail response to an impulse of CO2 injected into the environment. That’s what I am showing with my simulation.

    It should be a common understanding that the Berne model provides a simple fat-tailed heuristic for a multi-compartment model that also features diffusion (or reaction-diffusion) kinetics within the compartments.

    And this can be extended to thermal diffusion and explains why the first-order models of people such as Nic Lewis fail so miserably at producing useful climate sensitivity values.

    Murray Gell-Mann died yesterday. He was fond of pushing the idea that most phenomena can be explained by existing math — ” They follow from the fundamental theory. They are what we call emergent properties. You don’t need — you don’t need something more to get something more. That’s what emergence means.” He really should be considered alongside Feynman in the way scientists think about solving problems. “The fundamental law is such that the different skins of the onion resemble one another, and therefore the math for one skin allows you to express beautifully and simply the phenomenon of the next skin. “

  24. Dave_Geologist says:

    I think we’re talking at cross-purposes Paul.

    Sedimentation in the ocean, right?


    Or more accurately, the fat tail of diffusion of CO2 into deeper and deeper layers of ocean water where it is effectively sequestered very slowly over the span of thousands of years.

    Wrong and not accurate at all (that is, it may be an accurate description of ocean-atmosphere interaction in the Bern model, but is an inaccurate description of atmosphere-hydrosphere-rock-plankton-seabed-sediment interaction). Weathered rock dissolves in rivers. Some eroded material directly carries carbon into the sea as oxycarbonates, where it sinks to the seabed. Most is carried as bicarbonate ions. They are biologically converted into calcium carbonate, mostly by plankton in the shallow photic zone. The dead skeletons are denser than water and sink to the sea floor, where they become buried. Sequestration is in sea-bed sediments, not in deep ocean water. As limestone, not as dissolved CO2. None of the carbon has been in a CO2 molecule since it left the atmosphere.

  25. It’s my understanding that the 20% – 30% that will remain in the atmosphere is after what some call ocean invasion and is essentially once all the CO2 that can be dissolved in the ocean has been dissolved. This is limited by the Revelle factor which essentially says that the perturbation of atmospheric CO2 will be about 10 times greater than the pertubation in dissolved inorganic carbon in the ocean. As Dave then highlights, this enhancement is then drawn down on the timescale of about 100000 years through sedimentation (or, weathering) which involves carbon in the form of (I think) calcium carbonate sinking to the ocean floor and then into the lithosphere through subduction.

    I’m not an expert at this, so may not have described it as well as it could have been described.

  26. “Wrong and not accurate at all”

    Oh boy Dave, that’s such a small fraction of the overall dynamics. CO2 does not permanently precipitate out that easily, and that’s why it has an adjustment time of thousands of years. That’s also why CO2 is considered a non-condensing GHG. Instead it is held in a non-condensed limbo as a random-walk dispersive state that Rohde’s simulation is showing. This is essentially the abstract dynamic sequester that the Berne model impulse response function describes.

    “None of the carbon has been in a CO2 molecule since it left the atmosphere.”


  27. Paul,

    CO2 does not permanently precipitate out that easily

    I don’t think Dave suggested it was easy. I think he’s trying to distinguish between it mixing throughout the ocean, and it sedimenting to the sea floor.

  28. Dave_Geologist says:

    Good description ATTP. “Sedimentation” made me assume that was what the conversation was about. However, subduction actually recycles the CO2 back into the atmosphere. It’s sequestrated perfectly fine on the sea bed. Even if it’s thrust up and weathered later, it takes CO2 from the atmosphere to release the carbon from the rock as bicarbonate ions. Subduction drives the CO2 out of the rocks as they are heated, and it comes up through volcanoes (although a lot of the sediments are scraped of into an accretionary prism and plastered onto the continental margin). The CO2 can also be released by rifting and volcanic activity. Some of the CO2 which plagues some Southern North Sea and South China Sea gasfields comes from thermal metamorphism of limestones. Essentially the same process as cement manufacture, converting clays and limestone into calc-silicate minerals.

  29. Dave_Geologist says:

    Read again Paul, starting with “Weathered rock dissolves in rivers”. It’s solution then ultimately biological fixation and burial beneath the seabed. I didn’t say it precipitates easily. It’s mediated by biological templates which I’ll call catalysts, although that’s probably the wrong word. It’s why seashells are intricately patterned mixes of organic lamellae and calcite or aragonite crystals. Even then it takes tens of thousands to millions of years. But the fixation is not the time-limiting part. That’s obvious from the speed with which plankton blooms form, die, then sink to the seabed. It’s the supply of raw material, in the case of carbon by rock weathering, in the case of micronutrients from various sources. That is on such a long time scale that I presume it’s not in most carbon-cycle climate models. It’s why I say “don’t be complacent about ECS – it’s not actually equilibrium”. And why ECS and ESS are different. For reasons of human relevance and computational tractability., most climate models aren’t run out to a million years. Only geological ones, modelling Snowball Earths or trying to match the onset and waning of glaciations.

    Some carbon is directly sequestered in the organic matter that falls with the tests, but that’s a relatively small contributor as most of that gets recycled by ocean life, and what’s left only survives in anoxic bottom sediments.

  30. izen says:

    ” It’s mediated by biological templates which I’ll call catalysts, although that’s probably the wrong word. ”

    I am not sure the exact process is fully understood. (at least by me)
    There are cellular pumps that increase the concentration of Ca+ and decrease the concentration of H+.
    Then there are polysacceride membranes or fibres or a matrix that have macromolecules on the surface that have a pattern of carboxy and sulphate groups at the nano scale that promote nucleation and the precipitation of amorphous calcium carbonate and pattern the further crystallisation into specific patterns of calcite or aragonite crystals. Further biologically driven processes can convert that to hydroxyapatite. As with teeth.
    It is an evolutionary conserved process with small variations that also are the basis of bone formation in larger animals, including us. It consumes cell energy resources to overcome, and vastly exceed the abiotic chemistry that would otherwise inhibit or slow carbonate precipitation.

  31. Dave_Geologist says:

    Thanks izen. I was aware of the template for nucleating crystals. Hence their alignment, most spectacularly in the compound eyes of some trilobites, where each lens is a single calcite crystal. The ionic pumps make sense because that’s what cells do for all sorts of purposes.

  32. izen says:

    I have been led to believe that one reason we do not have a CO2 atmosphere like Venus is that early on in the Earths history bacteria first developed carbonate precipitation leaving stromatilites, then with the Cambrian explosion Eukaryotas forming calcium carbonates tipped the VERY long term carbon cycle in favour of sequestration of CO2 into carbonate rocks.
    As a result the large amounts of CO2 in the atmosphere of Venus are locked on Earth semi-permanently and preferentially into rock (limestone, chalk etc), which contains at least an two orders of magnitude more Carbon than the active air-land-sea cycle we observe.

  33. Dave_Geologist says:

    Shhhh… when coupled with the Dim Young Sun, which made the early high-CO2 atmosphere compatible with liquid water and life, you’ll have someone claiming a benevolent deity encouraged the evolution of skeletonisers to keep the planet habitable for us.

    Except there are indications that the early oceans were very hot, like Yellowstone springs, and I have an each-way bet that the couple of early Proterozoic ice ages (not the c. 700 Ma ones, the older ones) were abortive attempts at going from partial to full oxygenation, which failed because the bacteria drew down CO2 too far and precipitated a Snowball Earth. Bit careless, that!

  34. Thanks for writing about this! Most fascinating video I’ve seen in a while. Couldn’t resist passing it on. “Fundamentals – Earth’s Carbon Cycle By The Numbers – R.Rohde” Also at WUWT.

  35. Dave (2:12pm), nice summary. Have any suggested reading tips for a student.

  36. Dave_Geologist says:

    Precambrian supercontinents, glaciations, atmospheric oxygenation, metazoan evolution and an impact that may have changed the second half of Earth history.

    The author’s favoured model is that the formation of supercontinents led to enhanced weathering which drew down CO2 (but wouldn’t the continental interiors be dry?), then Snowball Earth ice cover stopped weathering so CO2 could build up again (but he favours a slushball, without permanently frozen oceans). He also argues for rifting to cause sag and breakup of the supercontinent. As with the Cryogenian glaciations, ice extended to low latitudes and there was a sharp return to tropical conditions with cap carbonates, suggestive of a bifurcation between bistable states. The association with banded iron formations indicates either large fluctuations in ocean oxidation state, or mixing of waters with very different oxidation states.

    He argues for an impact to explain the failure of metazoan evolution to take off following the Great Oxidation Event. I wonder if there could be a bistability in the ability of the Earth to support abundant photosynthetic life – that there was a window where insolation was such that severe drawdown of CO2 precipitated a snowball or slushball, shutting down much photosynthesis and allowing CO2 levels to recover.

  37. It’s mind blowing the information they have been able to tease out of zircons and other minerals. Thanks for that link, some fascinating stuff.

  38. Dave, finished it, fascinating watching the resolution keep increasing, I knew the rough outlines, now I’ve got a glimmer of new details, plus learning about windows other than zircons to add to my appreciation of Earth’s deep past. Have I mentioned I’m an Earth Centrist, every little bit helps. 😉

    Thanks for taking the time to share it.

  39. Dave_Geologist says:

    Thanks for the thanks. It’s a while since I read it but it’s worth looking at the references (why I chose a recent one) to get other perspectives. Young was originally a bit of an outlier, not accepting that there were cap carbonates and favouring a Uranus-style rotation axis in the plane of the ecliptic, which gives extreme Milankovitch cycles with positions where you get equatorial rather than polar ice. That was back when there was much less data and Snowball Earth bifurcations were in their computational infancy. More data has come in, and as with Keynes, his opinions changed when the facts changed. Things are still very fragmentary though, due to poor preservation and the absence of macrofossils.

    I’m surprised no-one has tested the idea of extreme carbon-cycle excursions as photosynthesis expanded in a GCM. Just think how fast a plankton bloom can form, then imagine one with a whole new environment to exploit, with undepleted nutrients. Maybe they have, I’ve not actively searched. As well as being interesting for its own sake, it’s worth testing models beyond realistic modern boundary conditions, because it may highlight weaknesses in the model which are not normally apparent, and reassure you that they happen a long way from our zone of interest – or not. It’s one of the justifications people give for modelling planets and exoplanets.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.