Guest Post by Willis Eschenbach [See updated graph]
Inspired by some comments on another thread, I decided to see what I could find in the way of actual measurements of the amount of CO2 in the surface layer of the ocean. I found the following data on the Scripps Institute web site. What they did was drive around the ocean on four different cruises, measuring both the atmospheric CO2 levels and at the same time, the amount of CO2 in the surface seawater. Figure 1 shows those results:
Figure 1. All air-ocean simultaneous measurements from four Scripps cruises are shown as blue dots. The horizontal axis shows sea surface temperature. The vertical axis shows the difference between the CO2 in the overlying air, and the CO2 in the water. The red line is a lowess curve through the data. The paper describing the Scripps data and methods is here.
Now, I have to say that those results were a big surprise to me.
The first surprise was that I was under the impression that there was some kind of close relationship between the atmospheric CO2, and the CO2 in the surface seawater. I expected their values to be within maybe 5 ppmv of each other. But in fact, many parts of the ocean are 50 ppmv lower than the CO2 concentration of the overlying air, and many other parts of the ocean have 50 ppmv or more of CO2 than the CO2 in the air above.
The second surprise was the change in not only the size but even in the sign of the trendline connecting temperature and CO2 (red line in Figure 1). Compared to the CO2 level in the air, below about 17°C the seawater CO2 decreases with increasing temperature, at a rate of about -2 ppmv per °C.
Above about 17°C, however, the seawater CO2 content relative to the air increases fairly rapidly with temperature, at about +4 ppmv per °C.
To describe the situation in another way, when the water is cool, it contains less CO2 than the overlying air … but when the water is warm, it has more CO2 than the overlying air.
Say what? I gotta confess, I have little in the way of explanations or comprehension of the reason for that pattern … all suggestions welcome.
w.
[UPDATE] By popular request, here is the same data, but in absolute rather than relative units and without the lowess curve.
Figure 2. As in Figure 1, but showing the CO2 content of the surface seawater directly. Atmospheric CO2 varied very little during the time of the measurements.
My main question in all of this is, how does the CO2 content of the seawater get to be up to 100 ppmv above the CO2 content of the overlying air? It seems to me that the driver must be biology … but I was born yesterday.
Regards,
w.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Solubility of CO2 in water is not the only factor to consider. In sea water there are two significant processes that affect the amount of dissolved CO2. One is biological (phytoplankton blumes etc.) and the other is inorganic formation of carbonates and bicarbonates (both soluble and insoluble). It appears that below 17 degrees the biological processes dominate while above 17 degrees the inorganic processes are in control. Also,land based biological activity has a strong effect on the atmospheric concentration of CO2 over the oceans. Another thing to consider is that frozen water does not absorb CO2.
Geographical, seasonal, and time-of-the-day splits could bring some order into this mess, but even under the assumption of gas solute equilibrium (which I believe is achieved within minutes under all these conditions), there will be a lot of variation that is difficult or impossible to control for.
If life has any effect on this, the effect will not be simple. There are multiple producers and consumers of CO2, O2, and N2 in ocean water, they interact with each other, and even in isolation, each kind can behave in interesting ways. For example, it can appear and disappear:
http://www.lmd.ens.fr/geomix/Lectures/Adrian_Martin/Lecture_2/Lecture-2-HTML-Explorer/patchy1_fichiers/slide0024_image016.gif
Technologically, measurements taken 50 years ago would be the same as now — no better, no worse; probably a bit more expensive than now because data logging is now automatic.
I’m wondering; what was the barometric pressure when each reading was made? Should that be another axis? And… how close together were the readings? Did they drive to an area, park and take multiple readings? What is the variance in the same location at roughly the same time? Hmmm… Let’s go look……………………………….. OK… I’m back…
I only had time to read briefly over on the Scripps site, but when I see something like this quote;
“The atmospheric air stream on the LUSIAD cruise displayed a diurnal cycle in measured CO2 mixing ratio [Waterman et al, 1996, p. 20-21], likely caused by degassing of the plastic airline when exposed to sunlight. Only nighttime air data was considered free of this effect and thus acceptable.”
then, I wonder if all the data on Willis’ plot is all apples or if there might be more than a few oranges, lemons, and cacti mixed in on the plot? What data was thrown out from the other cruises?
Thanks Willis. This is interesting. I’m going to have a longer look-a-see on the Scripps site over the holiday weekend.
Me too.
I have read other sources saying, quite reasonably in my view, that oceanic pH measurements are almost useless without salinity measurement and adjustments. CO2 measurements are, of course, sensitive to pH.
@Willis and Retired Engineer and lenbilen
There are some interesting hits and a lot of blanks in here today.
Learn from Bob Tisdale: upwelling ocean currents bring things that went under 800 or 1000 years ago.
As this a really ‘macro’ look at subject, here are two macro observations:
1. The warmer oceans with the higher CO2 levels are rainier places than the colder and dryer oceans. Rain carries a great deal of CO2 into the ocean – much more than the ocean can sustain because it a) can only retain a certain level of solvated molecular CO2 whereas fresh water is much better at this ‘skill’ and b) there are so many CO2-eating activities going on in sea water (making carbonic acid out of only some 1% of it – don’t believe the guff).
lenbilen: >The rain forests and the Gobi desert has the highest concentrations of CO2 on land and parts of Siberia the least.
Parked as I am on the north edge of the Gobi I will toss in that water droplet clouds blowing in from any direction evaporate in the air over the Gobi most of the time releasing much more CO2 than lakes and streams which are buffered (a lot) by rocks and sand.
2. The upwelling of CO2-rich cold water that is heated as it arrives at the surface is a very strong emitter so one would expect it to be higher than the air above it. The chart of course does not have upwelling/downwelling attached to each dot.
So can generalize about the rainfall stripping the atmosphere in the wet tropics (as per global CO2 charts) and maintaining it in the warm oceans, even without upwelling. After entry, the salt water environment tries to expel some of it
The combination of being up/downwelling and the rainy /dry areas provide the large amount of scatter. In a still sunny ocean, Henry’s law governs the sea water concentration. But it also governs the concentration in fresh water droplets (rain). As soon as it rains, the sea water is ‘overloaded’. That explains why the top can be warm and higher. I presume everyone has noted that the level of CO2 in sea water is quite different from CO2 in fresh water.
Have some fun:
http://www.thuisexperimenteren.nl/science/carbonaatkinetiek/Carbondioxide%20in%20water%20equilibrium.doc
20 deg water holds a lot less CO2 that 5 deg water: “At 20 C the CO2 content of [fresh] water is 0.029 mol/litre (1.28 g/litre)” (ibid) but 3 g at 5 C. That is a very big difference.
The only way to get an increase is to have brought it from below (old), or poured in from above where it disperses either down or up and out (new).
Is it now time to demote “Henry’s law” to “Henry’s suggestion” ?
Willis: Did you honour the data quality flag which the paper mentions as there was apparently significant diurnal contamination noted along with other data quality issues?
“Say what? I gotta confess, I have little in the way of explanations or comprehension of the reason for that pattern … all suggestions welcome.”
A meaningless scatter diagram from which one should draw no conclusions even were it some type of controlled, structured or random sample, which it is not. Beating such data with a statistical hammer is a waste of time.
michael hart says:
Not measurements themselves — not in a wide enough range of pH around neutral — but ions in water can affect gas solubility.
Typical pCO2 measurements *are* pH, measured behind a gas-permeable membrane in contact with the target medium and calibrated in partial pressure units using ambient air, calibration gas, or some calibration solution magic that provides a buffered source of dissolved CO2.
My first thought is that PPMV is a poor measure because of its different meanings in non-homogeneous fluids versus roughly homogeneous (but chemically heterogeneous) gases. My second thought is that there’s something seriously wrong with this data, or there’s an elephant in the room. Assuming that the only relevant correlative variable is temperature, the variance and the poor correlation coefficient here suggests that the data is very dirty and you can’t really see the relationship. Whatever relationship there is here is overwhelmed by some other factor(s). Salinity, maybe, I’m guessing? Temperatures taken at day versus night (consider effects of penetration of sunlight into shallow water regions)? How about more obvious things such as wind patterns and water turbulence due to storm, tide or current activity — both of which affect mixing of “deep” and “shallow” layers of both substrates, which may significantly affect the data?
Also, I wonder what “sea surface temperature” is, and whether it is an appropriate variable here? Is it a water temperature, or an atmospheric one? How is it taken? Does humidity play a role? What would happen if you took both water and air temperatures and plotten their difference against the CO2 difference?
To many “wait a minute, what if”s here for me to believe this data is saying anything significant.
There are multiple patterns in this. It is inappropriate to put one trend in (though that is what you first do). I’ll bet that there are separate patterns based on position/basin/region and time.
This information is important for two questions/observations I made recently (to me): that the seasonal variation in CO2 content as measured at Mauna Loa is uni-modal in a world that is bi-modal (by temperature and time of maximum organic activity FOR two hemispheres).
Intuitively I would expected two peaks and troughs reflecting increased/decreased absorption of CO2 in sea water during cold/warm periods, and increased/decreased production/absorption of CO2 by plant matter in the growing, stability and decomposition phases. I would not have expected a nice balancing of these factors, as the hemispheres are not equal for either temperature changes, percentage of oceanic waters (of the same temperature variations) or plant growth. I would expect such a simplicity if there was one overwhelming factor that determined CO2 variation.
The overwhelming factor I see is a REGIONAL surface sea water temperature. The times of peak CO2 is beginning May, the low, mid-October. That is a 4 1/2 month split – in a 12 month year. But sea surface temperature can also create plankton blooms and die-offs, so, as shown by Mauna Loa “adjustments” and cleaning, the asymmetric, unimodal style looks to me to reflect a region of seas that
1) come to peak warming in May and peak cooling in October – releasing CO2 in May and absorbing it in October, OR
2) come to a low biogenic activity in May and peak activity in October – allowing decomposition to dominate in May, releasing CO2, and pulling it out of the atmosphere in October when photosynthesis is at its height.
Both scenarios would involve, obviously, a planetary level smoothing of all factors. However, the dominance theory says we search for the ocean area that “counts” – which is probably the Indian/West-Central Pacific. The rest of the planet is noise; only that region provides the signal.
I found this paper to be a good read on the subject
http://www.ldeo.columbia.edu/~csweeney/papers/taka2002.pdf
from the Abstract:
A zone between 40 and 60 latitudes in both the northern and southern hemispheres is found to be a major sink for atmospheric CO2. In these areas, poleward-flowing warm waters meet and mix with the cold subpolar waters rich in nutrients. The pCO2 in the surface water is decreased by the cooling effect on warm waters and by the biological drawdown of pCO2 in subpolar waters. High windspeeds over these low pCO2 waters increase the CO2 uptake rate by the ocean waters.
Air-water gas exchange is not efficient without mixing. In the ocean wave action is needed. Under what temperature conditions would there be more wave action? These ranges might be when the air/water partial pressures are more comparable.
Let me be more explicit. The amount of free CO2 is affected by CO2/bicarbonate/carbonate equilibration [and the rate, since this is not at equilibrium], which is dependent on pH.
michael hart says:
True, that, although this dependence is bi-directional. Incidentally, this is just the system that is used in the most common type of pCO2 probe. It has a carbonate/bicarbonate buffer inside that is monitored with a regular pH probe.
To make things more interesting, marine photosynthetics have carbonic anhydrase, which allows them to uptake bicarbonate. They don’t care what form of inorganic carbon is there, so there is more than one way they can shift equilibrium. They respire, too.
Willis, the figure bothers/puzzles me too. The red line is meaningless, for one thing. For another, there is clearly evident structure, multiple bands at a statistically significant level, across the entire figure but especially in the range from 17 C to 25 C. Warmer than 25 C there may be structure as well, but the bands appear to have merged and smeared.
I suspect that part of the structure reflects the effects of ocean currents. Ocean currents can take warm water with an overpressure of CO_2 and transport it to colder waters where it mixes but takes time to re-equilibrate. Sampling within currents (e.g. the Gulf Stream) would then show banded structure like that seen.
The other thing that I suspect is sampling error. You say they drove around on four cruises, sampling along the way. In the temperate 17-25C range, there appear to be four bars. A coincidence? I doubt it. The ocean’s surface CO_2 content is evidently inhomogeneous (that much is correct beyond any doubt) but it also has a strong spatial autocorrelation, again probably due to surface transport and upwelling. Note also the tremendous sparseness of the data at colder temperatures compared to the data at warmer temperatures. In part this reflects (no doubt) the Jacobean and shape of the continents and oceans — there is a lot more water at tropical latitudes than there is at the poles — but I’d bet my sweet bippy that it also reflects the selection bias of researchers to prefer ocean cruises in the warm, sunny tropics with lots of interesting places to stop and things to see relative to cruising around the Cape of Good Hope or Tierra del Fuego or knocking around Iceland or the Bering Straits — presuming one can get in through the ice and so on. There are hardly any samples from 4C water, and none from water at the saltwater freezing point. Nobody from California wants to spend their time wearing mittens and trying not to touch metal because it will freeze your skin by sticking to it.
With sparse, biased samples, it is difficult to draw any conclusions from this data.
I’m surprised that they didn’t put CO_2 sampling into the ARGO floats. What the hell were they thinking? You’re going to spend a small fortune per float anyway, and the marginal cost of additional testing hardware is tiny compared to the cost of funding junkets for researchers to badly sample a biased trajectory through the ocean.
If I were going to sample the ocean, I would take the globe and crank out random coordinates on its surface (intelligently, not treating lat/long as scaled uniform deviates). Then I would use a computer to solve the problem of efficiently visiting the sites, and either arrange cruises to systematically work through the list (ideally starting from multiple locations, not just California) or devise e.g. a computer/radio controlled, unmanned solar powered “ship” that would work its way through the list without human intervention. Or I’d go and stick CO_2 sensors on board ARGO floats the next time they are serviced and learn about the spatiotemporal CO_2 profile of the entire ocean to depth.
The really interesting, tantalizing question that the data above is insufficient to answer is — in the warmest waters there is a really substantial, quite possibly robust, CO_2 overpressure in the water relative to the air. One has to wonder — is this CO_2 moving INTO the air (predominantly) or OUT of the air? The usual assertion is that cold water holds more CO_2, but this graph suggests otherwise in the only place that really matters, at the surface. The other thing that bothers me is that the graph doesn’t present the actual atmospheric CO_2 associated with each dot, so we cannot answer some enormously interesting questions, such as what the mean atmospheric concentration of CO_2 in ppm at sea level actually was, averaging over all four cruises. Did the data reveal that CO_2 is indeed “well-mixed” so that the averages more or less match those observed at Mauna Loa? Or did they get some other answer? At places where oceanic CO_2 was higher than atmospheric CO_2, was atmospheric CO_2 higher than the Mauna Loa mean or lower? Is there evidence, in other words, that the CO_2 is moving in or moving out in the warmest waters!
Maybe all of this is revealed in other data taken during the voyages, but now you’re gonna have to winkle it out too — the figure above isn’t enough to draw any conclusion beyond “four voyages were made and one can easily guess the spatiotemporal pattern of the sampling from the appearance of irrelevant banding in the data”. And the fact that four samples — not the hundreds that the graph APPEARS to show but FOUR for FOUR trajectories — is not a lot to draw conclusions from with the smear and structure already apparent in the four trajectories.
rgb
How does time of day of the sample change this graph?
We know that CO2 levels above corn fields vary dramatically during the day, bottoming out around noon-time. Plants suck half the CO2 out of the air in roughly the first 6 hours of the day.
Photosynthetic organisms in the water could do the same thing to a greater or lesser degree.
see: http://jonova.s3.amazonaws.com/graphs/co2-absorption/chapman-corn-field-fig-1-co2.gif
Unfrozen Caveman, Dodgy Geezer and dccowboy on escaping the fate of Venus,
The Sun is understood to have increased in brightness over geological time. This might be pertinent.
@Willis: “Say what? I gotta confess, I have little in the way of explanations or comprehension of the reason for that pattern … all suggestions welcome.”
Easy, ceteris paribus is only valid when the ceteris is paribus. The lab and life are not always the same, or even in the same area code: Unless you start accounting for the confounds.
All the mentions of confounds so far seem reasonable to some degree or another. But I’d wager the majority effect here is one of aerobic life getting its groove on. Not as catchy or as sane to put forward and a cap and trade idea on carbon based life though…
Would you have a similar graph with respect to oxygen? I am wondering if there are more animals in the tropical regions that breathe more oxygen and emit more CO2.
I’m delighted to hear you say that, Gene. I occasionally point out the same thing, but it gets roundly ignored. In fact all life forms make extensive use of carbonic anhydrase, not just photosynthetic organisms. How long the activity persists in dead vegetative matter is another interesting question.
Plants (phytoplancton etc) respire in dark – give out CO2 and photosynthesise in light- take in CO2
The biggest variation in yearly CO2 is at northern most monitoring stations. The CO2 variation is therefore likely to originate around these.
Decomposition requires warm temperatures for the bacteria to work.At the end of summer temperatures are falling rapidly so decomposition will not release CO2 rapidly and certainly would not continue into December. This suggests that spring growth and autumn decomposition would not cause the CO2 dip.
CO2 creating carbonic acid in the sea would not release O2 nor absorb O2
this plot shows that the change in O2 is the inverse of the change in CO2 – Just what you would get with 24 hour daylight vs 24 hour dark. The yearly variation is simply the plants breathing!!
You will note that there is a general decline in O2 over the years. Burning plant growth are suitable causes dissolved CO2 emerging from warmed oceans is not the cause.
http://bit.ly/Ipcj4R
CO2(aq) + H2O H+ + HCO3-
i.e. there is no absorbed O2 nor released O2 as sea water takes up/releases CO2
Henry’s Law is valid for standard conditions, meaning pressure is a variable that also has to be normalized (controlled) to make sense of the equation. At higher pressure, more CO2 dissolves in surface water. The Scripps Oceanographic vessels gather data about atmospheric pressure also, along with the ones mentioned here. It would be simple to normalize the data, making pressure a constant. One must isolate all of the ancillary variables if the relationships among the key study variables are to be understood properly.
Robert
It’s hard to see what is going on unless the raw data for both data series is provided. But as you say the role of both temperature and pressure cannot be properly resolved without “normalising” the pressure.
The other thing that makes this a difficult issue is that one would also need to assess the amount of physical mixing and the composition of the water with respect to the “bicarbonate system”. Furthermore, I suspect the mere presence of biological activity would have a role to play.
Thank you, I’m certain it is.
Although Earth and Venus have certain physical similarities, they are quite distinct and were destined to evolve into two distinct planetary environments:
a) Venus is a very dry planet, and the Venus we see now is nothing like what once existed. The reasons for the loss of its water are still in dispute, please see:
http://www.vanderbilt.edu/AnS/physics/astrocourses/AST101/readings/water_on_venus.html
b) Besides carbon dioxide, Venus’ atmosphere contains vast amounts of sulfuric acid, which actually [accounts] for the lovely clouds we see in photographs. The combined molecular weight of the atmosphere on Venus is far greater than Earth, which is one reason for the hellish conditions on the surface (higher pressure at the surface, temperature etc.).
c) It seems obvious to say this, but Venus is much closer to the sun, and the inverse-square law dictates that Earth receives proportionally much less energy from the sun by the square of the distance.
Comparisons of Earth to Venus by jokers like Hansen drive me nuts! This one is characteristically bad: http://www.climatevictory.org/venus.html