From the Massachusetts Institute of Technology:
While the Arctic Ocean is largely a carbon sink, researchers find parts are also a source of atmospheric carbon dioxide
For the past three decades, as the climate has warmed, the massive plates of sea ice in the Arctic Ocean have shrunk: In 2007, scientists observed nearly 50 percent less summer ice than had been seen in 1980.
Dramatic changes in ice cover have, in turn, altered the Arctic ecosystem — particularly in summer months, when ice recedes and sunlight penetrates surface waters, spurring life to grow. Satellite images have captured large blooms of phytoplankton in Arctic regions that were once relatively unproductive. When these organisms die, a small portion of their carbon sinks to the deep ocean, creating a sink, or reservoir, of carbon.
Now researchers at MIT have found that with the loss of sea ice, the Arctic Ocean is becoming more of a carbon sink. The team modeled changes in Arctic sea ice, temperatures, currents, and flow of carbon from 1996 to 2007, and found that the amount of carbon taken up by the Arctic increased by 1 megaton each year.
But the group also observed a somewhat paradoxical effect: A few Arctic regions where waters were warmest were actually less able to store carbon. Instead, these regions — such as the Barents Sea, near Greenland — were a carbon source, emitting carbon dioxide to the atmosphere.
While the Arctic Ocean as a whole remains a carbon sink, MIT principal research scientist Stephanie Dutkiewicz says places like the Barents Sea paint a more complex picture of how the Arctic is changing with global warming.
“People have suggested that the Arctic is having higher productivity, and therefore higher uptake of carbon,” Dutkiewicz says. “What’s nice about this study is, it says that’s not the whole story. We’ve begun to pull apart the actual bits and pieces that are going on.”
A paper by Dutkiewicz and co-authors Mick Follows and Christopher Hill of MIT, Manfredi Manizza of the Scripps Institute of Oceanography, and Dimitris Menemenlis of NASA’s Jet Propulsion Laboratory is published in the journal Global Biogeochemical Cycles.
The ocean’s carbon cycle
The cycling of carbon in the oceans is relatively straightforward: As organisms like phytoplankton grow in surface waters, they absorb sunlight and carbon dioxide from the atmosphere. Through photosynthesis, carbon dioxide builds cell walls and other structures; when organisms die, some portion of the plankton sink as organic carbon to the deep ocean. Over time, bacteria eat away at the detritus, converting it back into carbon dioxide that, when stirred up by ocean currents, can escape into the atmosphere.
The MIT group developed a model to trace the flow of carbon in the Arctic, looking at conditions in which carbon was either stored or released from the ocean. To do this, the researchers combined three models: a physical model that integrates temperature and salinity data, along with the direction of currents in a region; a sea ice model that estimates ice growth and shrinkage from year to year; and a biogeochemistry model, which simulates the flow of nutrients and carbon, given the parameters of the other two models.
The researchers modeled the changing Arctic between 1996 and 2007 and found that the ocean stored, on average, about 58 megatons of carbon each year — a figure that increased by an average of 1 megaton annually over this time period.
These numbers, Dutkiewicz says, are not surprising, as the Arctic has long been known to be a carbon sink. The group’s results confirm a widely held theory: With less sea ice, more organisms grow, eventually creating a bigger carbon sink.
A new counterbalance
However, one finding from the group muddies this seemingly linear relationship. Manizza found a discrepancy between 2005 and 2007, the most severe periods of sea ice shrinkage. While the Arctic lost more ice cover in 2007 than in 2005, less carbon was taken up by the ocean in 2007 — an unexpected finding, in light of the theory that less sea ice leads to more carbon stored.
Manizza traced the discrepancy to the Greenland and Barents seas, regions of the Arctic Ocean that take in warmer waters from the Atlantic. (In warmer environments, carbon is less soluble in seawater.) Manizza observed this scenario in the Barents Sea in 2007, when warmer temperatures caused more carbon dioxide to be released than stored.
The results point to a subtle balance: An ocean’s carbon flow depends on both water temperature and biological activity. In warmer waters, carbon is more likely to be expelled into the atmosphere; in waters with more biological growth — for example, due to less sea ice — carbon is more likely to be stored in ocean organisms.
In short, while the Arctic Ocean as a whole seems to be storing more carbon than in previous years, the increase in the carbon sink may not be as large as scientists had previously thought.
“The Arctic is special in that it’s certainly a place where we see changes happening faster than anywhere else,” Dutkiewicz says. “Because of that, there are bigger changes in the sea ice and biology, and therefore possibly to the carbon sink.”
Manizza adds that while the remoteness of the Arctic makes it difficult for scientists to obtain accurate measurements, more data from this region “can both inform us about the change in the polar area and make our models highly reliable for policymaking decisions.”
This research was supported by the National Science Foundation and the National Oceanic and Atmospheric Administration.
Written by Jennifer Chu, MIT News Office
Right On, Bart. Well considered points.
The only part we don’t know enough about at this point is which temperature modulated processes are important in the mass balance. There are so many possibilities it boggles the mind … it could be something as “simple” as ocean-surface-evapo-transpiration and/or storm-wind-wave-action or as complicated as the mosaics of primary productivity found in algal blooms and rainforest. All of the above sounds good too, rofl.
BioBob says:
December 5, 2013 at 10:27 am
“There are so many possibilities it boggles the mind…”
Agreed. I suspect deep ocean upwelling because it has been so inadequately characterized, and has potential to be enormous. But, that does not at all preclude other possibilities.
Bart says:
December 5, 2013 at 10:03 am
and
BioBob says:
December 5, 2013 at 10:27 am
To repeat the obvious:
There are only a few natural sinks/sources which may react fast enough on the year by year addition by humans: the oceans and the biosphere.
The biosphere is a proven sink for CO2, based on the oxygen balance:
http://www.bowdoin.edu/~mbattle/papers_posters_and_talks/BenderGBC2005.pdf
The oceans are not the cause of the increase in the atmosphere, as any substantial release of CO2 from the oceans or a substantial increase of circulation ocean-atnosphere-ocean would increase the 13C/12C ratio of the atmosphere, but we see a firm decrease in ratio with human emissions, both in the atmosphere as in the ocean surface layer:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/sponges.gif
Other sources/sinks are either too smal or too slow.
Further, the short term variations follow the (sea surface) variations, in particular the ENSO (El Niño) events. The largest increase in the atmosphere (or the smallest sink capacity) is during an El Niño event, when less deep water is upwelling and thus less CO2 is released from the tropical oceans. The extra release is from the tropical forests which give – temorarely – more decay and less uptake partly by higher temperatures, party by changed rain patterns. See Fig. 7 in the first link.
This is contrary to the long term trend where an increase in temperature causes more release from the oceans and more uptake by vegetation.
Thus the short term variability and the longer term trend are from different processes and one can’t deduce the cause of the trend from the short term variability by assuming that only one process is at work…
Biobob, have a look at Feely e.a. at:
http://www.pmel.noaa.gov/pubs/outstand/feel2331/background.shtml
and following sections. It gives a good overview of what is already known of the carbon cycle in the oceans-atmosphere segment. Still a lot to discover and the coverage still is sparse, but the main lines are quite known. E.g. the 13C/12C ratio of the oceans (deep and surface) is much higher than of the atmosphere. Even including the isotopic fractionation at the sea-air border (and back) would increase the 13C/12C ratio of the atmosphere with a substantial release from the oceans…
Ferdinand Engelbeen says:
December 5, 2013 at 1:58 pm
The data show we do not even have a handle on the normal CO2 cycle, much less all the factors influencing isotope and oxygen ratios. Sorry. The relationship is solid. As I stated, you will see soon, if we continue along with non-increasing temperatures and increasing emissions. That will be my final word on this thread.
So Sorry, Ferdinand Engelbeen …let me know when somebody has order of magnitude data / understanding of 4 season data on CO2 fluxes over large portions of the globe (that is why we would use massively replicated random sampling). Something like a global network of satellites targeted for this purpose with randomized local sampling for calibration would do the trick. I will not hold my breath. As it is now, all we have are estimates pulled from somebody’s posterior, at best.
And that’s all I have to say about that.
BioBob says:
December 5, 2013 at 3:35 pm
Biobob, one doesn’t need any local or regional flux of CO2 to know the main lines in the carbon cycle.
We have the mass balance (Bart, wait a minute until the end…) which shows a deficit in natural inflows compared to natural outflows:
increase in the atmosphere = natural inflows – natural outflows + human emissions
Human emissions are known with reasonable accuracy (sales taxes * burning efficiency) and atmospheric increase is known with high accuracy. That gives:
4.5 GtC/yr = natural inflows – natural outflows + Y + 9 GtC/yr
or
natural inflows – natural outflows = -4.5 GtC/yr
That is the mass balance: nature is a net sink for CO2 over the past 50+ years:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/dco2_em2.jpg
We have the 13C/12C decline, caused by fossil fuel burning. The decrease is about 1/3rd of what can be expected from the human addition, because “thinned” by the (deep) ocean exchanges which have a higher 13C/12C ratio. That can be used to calculate the ocean-atmosphere exchanges:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/deep_ocean_air_zero.jpg
We have the pre-bomb 14C/12C decline, caused by 14C-free fossil fuel burning, which made it necessary to correct the radiocarbon dating after ~1870.
We have the oxygen balance, which shows that vegetation is a net sink for CO2, with preference for 12CO2, leaving relative more 13CO2 in the atmosphere and thus not the cause of the 13C/12C decline.
We have the equilibrium ratio over 800,000 years, which is quite linear 8 ppmv/K, not the over 100 ppmv/K we see in the past 50+ years, which is impossible from temperature alone: Henry’s law does give maximum 17 ppmv increase in the atmosphere for 1 K increase in global seawater temperature.
Thus all available observations are consistent with human emissions. None refutes that cause.
————
Now Bart (and Salby) have an alternative theory:
Temperature is the only cause of the increase of CO2 in the atmosphere.
Theoretically that is possible, if some natural exchanges in and out the atmosphere increased at the same ratio and in the same timeframe as human emissions combined with extreme short decay rates for CO2 to equilibrium. But that violates about all known observations:
– the residence time of CO2 in the atmosphere should be reduced a theefold over the past 50 years, while recent estimates show an increase
– the biosphere can’t be the source, as there is no 3-fold increase in seasonal amplitude, not in CO2, not in 13C/12C ratio.
– the 13C/12C ratio decline would go opposite to what is seen in the atmosphere and ocean surface waters if the deep ocean – atmosphere exchanges increased over time:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/deep_ocean_air_increase_290.jpg
– the 14C bomb spike would show an increasing faster decay rate over the past 50 years, but that is not what is measured.
– the observed e-fold decay rate is 210/4.15 GtC/GtC/yr = ~50 years
210 GtC is the excess amount of CO2 above the equilibrium for the current temperature
4.15 GtC/yr is the current sink rate.
The decay rate is fast enough to follow the relative small temperature changes over the centuries but not fast enough to remove all human emissions on short term.
Even the nicest theory is disproven with one inconsistency. Here we have only inconsistencies, except for a nice fit of the two variables temperature and CO2 levels over short term variations…
Ferdinand Engelbeen says:
December 6, 2013 at 3:39 am
“Thus all available observations are consistent with human emissions. None refutes that cause.”
This refutes it. It is a solid fact. Your efforts to explain it away are unphysical.
“Even the nicest theory is disproven with one inconsistency.”
Yes, if it is a solid one. As the above.
“Here we have only inconsistencies, except for a nice fit of the two variables temperature and CO2 levels over short term variations…”
Inconsistency with a narrative is not determinative. Inconsistency with solid fact is.
Bart says:
December 6, 2013 at 9:24 am
This refutes it. It is a solid fact. Your efforts to explain it away are unphysical.
Bart, that plot is a mix of two processes: a fast reaction of the sea surface and vegetation on fast temperature changes, but limited in capacity and a slower reaction on human emissions by the deep oceans and more permanent storage in vegetation in direct response to the increase in the atmosphere above equilibrium:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/dco2_em4.jpg
Ferdinand Engelbeen says:
December 6, 2013 at 1:30 pm
“…that plot is a mix of two processes…”
You may believe that if you like. It is immaterial in any case.
Bart says:
December 7, 2013 at 11:43 am
You may believe that if you like. It is immaterial in any case.
Not a matter of “believe”, just a matter of looking at the history of CO2/T levels and knowing physical laws like that of good old Henry’s…
Here a plot of dCO2/dt and dT/dt per month as observed, emissions/12 and the integral of dT/dt * 0.7:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/dco2_em5.jpg
The integral of dT/dt has zero slope and a small offset of 0.017 ppmv/month (not even visible in the plot) which is the extra increase/month of 10 ppmv over 50 years, according to Henry’s law. The slope of dCO2/dt is entirely from human emissions and the variability around the slope is entirely from the short term temperature variation…
Ferdinand Engelbeen says:
December 7, 2013 at 12:27 pm
I did too much work by recalculating the integral of dT/dt, as it is already incorporated in Wood for Trees…
As said before: most of the offset and the trend (0.1-0.2 ppmv/month) is caused by human emissions and the contribution of temperature to the CO2 increase is average not more than 0.03 ppmv/month in the period 1979-2000 and zero thereafter. That caused a slight reduction in the increase rate after 2000.
The trend in increase rate of temperature over the full period 1979-2013 is slightly negative, so is the trend in contribution of temperature to the increase rate of CO2 in the atmosphere…
As there is no factor needed to match the slopes between dT/dt and dCO2/dt, the amplitude is not changed, whatever the relative slopes of T and dCO2/dt.
Ferdinand Engelbeen says:
December 8, 2013 at 12:52 am
The integral of dT/dt is T, and it is 90 deg out of phase with CO2, i.e., this is not the relationship which is evident in the record. There is no way to make it so, just because that is how you want it to be, and your exercise is meaningless.
Bart says:
December 8, 2013 at 11:30 am
The integral of dT/dt is T, and it is 90 deg out of phase with CO2, i.e., this is not the relationship which is evident in the record. There is no way to make it so, just because that is how you want it to be, and your exercise is meaningless.
CO2 lags T with 90 deg for the short term variability as direct response of CO2 changes to T variability per Henry’s law. By taking the derivative of both, you shift both back with 90 deg, still with a 90 deg lag of dCO2/dt after dT/dt. Which makes that T and dCO2/dt match in phase. That is the only reason, nothing to do with any physical process.
For both T, CO2 and dT/dt, dCO2/dt integration of T or dT/dt has the same effect on CO2 or dCO2/dt by shifting it 90 deg, thus giving a perfect match of the timing.
About the scales:
– the increase in temperature 1960-2013 was 0.6 K or 0.011 K/year or 0.0009 K/month, not visible in the variability or trend of dT/dt
– integrating dT/dt indeed gives the 0.6 K back, but also its effect on CO2, which is maximum 10 ppmv per Henry’s law over the same period or 0.18 ppmv/year or 0.016 ppmv/month, again not visible in the variability or trend of dCO2/dt
That makes that practically all the variability in CO2 and dCO2/dt is caused by the variability of T and thus dT/dt but that very little of the trend in CO2 and none of the trend in dCO2/dt is caused by T and dT/dt
No matter if the trend in dCO2/dt is caused by a natural cause or human emissions, there are two separate processes at work, where the variability and the trend of dCO2/dt are from different origin.
Anyway, you can’t say that one must integrate T to match the increase of CO2, as that is a non-unique solution and – again – doesn’t match the amplitude of the variability as that depends of the relative slopes of T and dCO2/dt, while direct integration of dT/dt always does, because there is no slope or hardly one (even slightly negative).
Ferdinand Engelbeen says:
December 8, 2013 at 12:20 pm
“That is the only reason, nothing to do with any physical process.”
You’ve given no “reason”. You have merely made an assertion. An assertion which, as I have repeatedly explained to you, is fundamentally, physically impossible.
“Anyway, you can’t say that one must integrate T to match the increase of CO2, as that is a non-unique solution…”
The outcome of an integration is unique.
“…while direct integration of dT/dt always does…”
… always does produce a series for which the variation is 90 deg out of phase with the variation of CO2.
What you are claiming is mathematical gibberish, Ferdinand. You have to match the phase. The derivative of your CO2 model output must match the derivative of the actual CO2. Otherwise, your model is incorrect. There is no alternative. You cannot just mix and match your preferred drivers as you please.
Bart says:
December 8, 2013 at 2:12 pm
Bart,
– the variation of CO2 follows the variation in T with 90 deg.
– integrating the variability of T gives the variability of CO2 with the right phase, with a small linear slope, caused by the small linear slope of T.
– the derivatives of both shifts both 90 deg back.
– which makes that dCO2/dt still lags dT/dt with 90 deg.
– which also makes that T and dCO2 match in phase (which has no physical meaning).
– integrating the derivative of T gives the derivative of CO2 with the right phase, with near zero slope and a very small offset.
– integrating the derivative of T also gives T with a small linear slope and the right phase.
– double integrating the derivative of T gives the CO2 increase with a small linear slope and the right phase.
The rest of the curvature of CO2 and the slope of dCO2/dt is from a different process than what causes the variability in CO2 and dCO2/dt.
[mods: missed an end /i after derivative of]
[That “derivative of” .. or the other one? Mod]
Ferdinand Engelbeen says:
December 8, 2013 at 3:10 pm
Ferdinand, this is just a jumble of words, with no particular meaning. T is affinely related to dCO2/dt, which means CO2 is an affine function of the integral of T. That’s it. You have to integrate T to get CO2. You have to integrate every component of T to get CO2. There is no way around it.
“…which has no physical meaning…”
You keep asserting this, but it is nonsense. It has the physical meaning that CO2 is an affine function of the integral of T.
“…integrating the derivative of T gives the derivative of CO2 with the right phase, with near zero slope and a very small offset.”
Integrating the derivative of T gives T. They are inverse operations, in a unique 1-1 correspondence. If you are integrating the derivative of T and getting anything other than precisely what you started with, then you are doing it wrong.
Bart says:
December 9, 2013 at 2:13 am
You have to integrate T to get CO2. You have to integrate every component of T to get CO2. There is no way around it.
If you integrate T you will find the full variability of CO2 and a small linear slope of CO2 caused by T and the right phase shift. The latter is not more than 10 ppmv for the 0.6 K temperature increase over the past 53 years, according to Henry’s law. That is all. The rest of the slightly non-linear slope of CO2 is from a different process, NOT from the integral of T.
If you integrate dT/dt you will find the full variability of T and a small linear slope of 0.6 K over the 53 years. dT/dt itself is near flat with a very small offset of 0.011 K/year or 0.0009 K/month.
If you integrate dT/dt you will also find the influence of dT/dt on dCO2/dt: the full variability with the right phase shift and zero slope with a small offset of 0.18 ppmv/year or 0.016 ppmv/month. That is all for the direct influence of temperature on the CO2 increase rate. The rest of the slope and offset in dCO2/dt are from a different process, NOT from the integral of dT/dt.
What you (and Salby) did is not integrating T itself for its direct influence on CO2 (slope and variability), but assuming that the full slope of CO2 is caused by T, which anyway is not the case.
I have been away.
Ferdinand Engelbeen says:
December 9, 2013 at 3:43 am
“If you integrate T you will find the full variability of CO2 and a small linear slope of CO2 caused by T and the right phase shift.”
If you integrate it relative to the proper baseline, you will reproduce the entire CO2 record.
“The rest of the slope and offset in dCO2/dt are from a different process, NOT from the integral of dT/dt. “
It matches perfectly, or as near perfect as you can expect to find.
“What you (and Salby) did is not integrating T itself for its direct influence on CO2 (slope and variability), but assuming that the full slope of CO2 is caused by T, which anyway is not the case.”
It is caused by the fact that T is offset from its equilibrium value. We know this because regardless of the chosen offset relative to the equilibrium temperature, which integrates to the trend in CO2, the slope in T integrates to the proper quadratic curvature when T is scaled to match the variability.
This does not mean that T is driving CO2, but that a T dependent process is. As human inputs are not T dependent, they are evidently being rapidly sequestered, and having little effect.