03 March 2010

Why is the ocean cold?

Folks reading headlines about record warm oceans might be surprised by this question. But it's a real question, if perhaps from a different viewpoint than you might think.

If we look at the surface of the ocean, we see that most of the ocean is warm. The presentation at the link over-emphasizes the polar regions -- they're actually much less of the earth's area. Even so, over half the ocean -- surface -- is warmer than 20 C.

So you might figure that the volume of the ocean would also be some moderately warm figure, maybe a bit colder since cold water sinks, but still fairly warm. Surely over 10? In fact, the volume of the ocean -- average up every blob of water there is -- averages 3.5 C. Go back to the surface map and take a look at how much of the ocean is that cold. Answer: Not much. Even less when you allow for the fact that the map is exaggerating how big the polar regions are (I get about 14% of the ocean surface was at least that cold on February 26th). The importance of those small areas of cold water is that there is no refrigerator in the ocean. Once water leaves the surface of the ocean (except for one even smaller exception I'll get to), there's no way to make the water any colder. So if you see water that's -1.0 C in the ocean, you know it came from somewhere that had water at least as cold as -1.0 C. It could have been even colder -- the original cold blob might have mixed with a warmer blob of water.

Some of you might have objected up there when I mentioned that the average ocean temperature is 3.5. There's a fairly popular error that says the deep ocean has to be 4 C. It runs this way: Water is densest at 4 C, so as you cool a body of water, once it reaches 4 C all this cold water sinks to the deeps. As you cool the surface further, the water is less dense, so it quits sinking. That leaves you with 4 C water in the deep.
The thing that is wrong with that, and it's only one error, is that the statements are only true for fresh water. If you're looking at ocean water -- where the density and freezing point also depend on how much salt you have, things are different. For saltwater with salinity greater than 22, the water gets denser all the way down to the freezing point. Typical ocean water has a salinity of about 35, and extremely little of the ocean is as fresh as 22. So you can get as cold as about -1.9 C in the ocean before freezing, and water as cold as that sinks to the bottom. (Or at least it tries.)

One part of our story, then, is that the deep ocean is cold, very cold, because cold water is denser than warm water and salt water gets denser all the way to its freezing point.  We can look at the maps of surface temperature and see that the only areas that get cold enough to supply most of the volume of the ocean are the Antarctic, the North Pacific and Sea of Okhotsk, the Labrador Sea (east coast of Canada), and the Nordic Seas (between Greenland, Iceland, and Norway).  If we look at the bottom of the ocean, we see even colder temperatures than the 3.5 C average.  Makes sense -- the bottom would have the coldest, densest water of all.  It's down to 2 C or even -1.  That limits the possible source regions even farther.  We find that the bottom waters (our creative name for the waters on the bottom of the ocean) come from either the Antarctic (Antarctic Bottom Water) or North Atlantic (the Nordic Seas, supplying North Atlantic Deep Water) -- and not the Pacific.

That points us to a different issue.  Temperature is not the only thing that affects the density of water.  How much salt is present does as well.  The North Pacific is fresher than the North Atlantic and Antarctic.  So even if we cool it to the freezing point, the North Pacific water can't get dense enough to sink to the ocean bottom.

But ... is there a place that has really salty water?  Can it get salty enough to make up for being warmer?  The answer is yes, and yes/no.  The yes is, predominantly, the Mediterranean Sea.  Water flowing out from the Mediterranean sea has a salinity of about 38, versus the about 35 of average ocean water.  That makes it quite dense.

The yes/no is a matter of history.  The answer today (and for a long time) is no.  Which is a little surprising.  The thing is, if you took the temperature and salinity of Mediterranean Outflow Water, and of Antarctic Bottom Water, and computed their density at the surface of the ocean, the Mediterranean water is denser.  So you'd expect it to be in the bottom of the ocean.  But, if you compute the density at depth -- where the water would be trying to get to if it wants to be bottom water -- you find that below a certain depth, the Antarctic water is denser.  Because Antarctic (and North Atlantic) water does fill the ocean below that depth, the Mediterranean water can't sink through to the bottom.  But if Mediterranean water were already filling the deep ocean, then it is the Antarctic and North Atlantic water that would be blocked from reaching the deep ocean.  An ocean structured like this would be extremely different from the one we do have today.

Let's go back to the one tiny exception to my 'no subsurface refrigerators' rule above.  It involves another subtlety in how water behaves.  Namely, the freezing point of water, like all materials, depends on the pressure.  The higher the pressure, the lower the freezing point.  Now think about the Antarctic Ice Shelves, the big ones -- the Filchner-Ronne in the Weddell Sea, and the Ross Ice Shelf in the Ross Sea.  They each have an area of about 500,000 km^2 (200,000 square miles), about the size of France or Texas.  Important to our concern here, they are very thick, from 300 m at the front edge, up to 1000 meters back where they meet up with the Antarctic ice sheet.  Water that was near the freezing point -- at the surface -- can circulate under these ice shelves and cool even farther, to the freezing point of water under pressure.  It's a small effect, about 0.075 C per 100 meters of ice shelf thickness.  But it isn't zero, and turns out to be important for the 'Ice Shelf Water' method of making Antarctic Bottom Water.  So you can see water at -2.3 C thanks to this particular refrigerator. (We talk about water mass 'formation'; it's really a matter of figuring out how things get cooled, saltier, or mixed together.)  Still, at a total of 1 million km^2, versus 360 million km^2 of the ocean, it's a pretty small exception to the rule.


carrot eater said...

Are you saying that in general, any water in the ocean depths cannot be colder than the coldest surface waters?

Is that necessarily so? Say we start with today's ocean, and I crank up the sun so it's giving off a higher intensity of radiation. The ocean surface everywhere will warm up first, and thus it should be possible for all surface points to be warmer than the coldest part of the depths. Given the density effects which would inhibit mixing, it could remain this way for some time.

Am I going wrong here, or am I misreading the original concept?

Robert Grumbine said...

Let's say that you missed my unwritten 'barring magic wands or rapid changes in the surface climate'. You're right as far as you go. The more precisely correct statement is that if you see water of some temperature in the deep ocean, then the surface it came from at the time it was last in contact with the surface was at least that cold.

If you show me a global ocean that everywhere has a surface warmer than +3, and I find deep ocean water at 0, I know that the ocean surface used to be colder than we see today.

For any ocean circulation more or less in balance, you won't see such a mismatch. On the other hand, the current deep ocean cycling time is around 500 years, so some modest imbalances can be expected.

If we did wave our magic wand and prevent the surface ocean anywhere from getting colder than 3 C, however, the deep ocean would warm up. The mismatch between the surface and deep has a limited life span. Deep water is warmed by the geothermal heat flux (50 milliwatts per square meter -- pretty small, but a lot of area) and by diffusion from warmer waters above. Also, at some point in time, the deep ocean will warm enough that Mediterranean water will start sinking to the deep, replacing the -1 to +3 water with water around +12. At that point, we'll be back to normal -- able to trace water temperatures back to then-current surface areas.

GFW said...

So, 500 years is roughly the time it takes for water to go from the equatorial surface to the equatorial deeps, via the North Atlantic? Or is that a misunderstanding of "deep ocean cycling time"?

Bigger question: we know that the loss of arctic sea ice gives a positive albedo feedback. But it's also increasing the area available in the Nordic seas for surface waters to give up heat and become North Atlantic Deep water. Is this a possible negative feedback for the planet as a whole? Could it cause the "conveyor belt" to speed up, rather than (as is postulated for large meltwater pulses) slow it down? Is there any modelling or other research on this?

jyyh said...

GFW: The ice and especially the snow on top of it is quite an insulator, so I doubt much speeding is possible (maybe only in autumn after the equinox?), but I think our host may give a more trustworthy answer.

Thomas Palm said...

I've been aware of the salty water in the Mediterranean sinking to intermediate depths because it isn't quite heavy enough to sink all the way to the bottom. Now it just struck that if it became heavier not only would it sink to the bottom, but it would lock away a lot of salt in the process. As larger portions of the deep ocean become filled with extremely salty water the concentration near the surface has to decrease, making it harder and harder to produce new water dense enough to reach the bottom. This seems like a pretty stable configuration, so how can it be broken? Slow geothermal heating of the deep water until the density drops?

Robert Grumbine said...

500 years is roughly the time for a blob of water to sink in the North Atlantic and cycle around the ocean until it rises in the North Pacific. Give or take some.

The Nordic Seas -- bounded by Greenland, Iceland, Norway, and Svalbard -- don't really have much ice cover in the present day, even in winter. Enough that you'd want to pay attention to ice charts! But most of the ice hugs the Greenland coast already, and the water there is fresher (so less dense) than in the rest of the area. So I wouldn't expect even complete elimination of the ice cover there to change the deep water formation picture much.

I think you answered your own question. You can only lock away so much salt in the deep ocean before you become unable to make deep water at all. This translates, at least over the long term, to the salty bottom water formation rate being self-limiting. The steady state, if there is one, balances the burial of salt in the deep sea by sinking processes with the freshening of that salty water by diffusion of that salt into the less salty water above, and the diffusion of heat from the warmer water above (I'm presuming that there _is_ warmer water above, which might not be true, but that would lead to a different very interesting subtlety of water's behavior) down to the deep ocean. Both processes will make the deep ocean water less dense, therefore capable (eventually) of resurfacing and maintaining a global deep ocean circulation.

One feature of all the salty bottom water ocean scenarios (that I've seen, which is no longer complete) is that the circulation is much slower than our current ocean. The outflow of pure Mediterranean Outflow Water is only about 1 Sv. Allowing for mixing, the resulting bottom water (if it did become bottom water) is, say 4 Sv. (1 Sv = 1 Sverdrup = 1 million cubic meters per second of water flow). Those 4 Sv compare with the 40 Sv coming, now, from the North Atlantic and Antarctic combined. (In all cases, treat figures as very loose, easy to remember, values.) The deep ocean circulation time would become more like 5,000 years, to the current something like 500. That extra time gives diffusion of salt and heat time to be much more significant than in the present ocean.

Hank Roberts said...

Asked at RC then I realized this is likely a better place--are there circulation maps that, for example, tell us the temperature of the water circulating over the areas currently leaking methane on the Siberian continental shelf, and the origins of that water? I realize it may be coming from some distance away and not reflect the local surface or the current temperature. If we had that plus info on how long water temperature takes to propagate into the sediment, we might be able to know more about methane emission over time.

It occurred to me to wonder if the gas/oil exploration companies, as well as the submarine navies, would have temperature and current records from their work, that could be data-mined for mapping ocean circulation by correlating temperature events over time if they were merged, somehow without losing them their proprietary advantage.

Hank Roberts said...

Another couple of asides

-- Are there any other gases besides methane that make clathrates? Particularly any of the sulfur compounds? And are there any clathrate-forming gases that would dissolve quickly and get carried down into the deep water if they did outgas, instead of bubbling to the surface? Pure speculation on my part, I don't know anything about this at all.

-- I think I recall mention in Peter Ward's "Under a Green Sky" that some possible changes in the North Atlantic (fresh meltwater, water, maybe also changes in something to do with the catabatic wind off of Greenland affecting mixing and heat transfer?) that would stop Arctic water forming large masses dense enough to sink, and when that happens water would instead sink in areas warmer, less oxygenated, further south.

I don't have the book handy and may well have mixed this up. Is there anything you can speculate on about change in location, rather than just the much discussed possible slowdown?

I think I recall that Peter Ward went off on an icebreaker last fall sometime, along with Eric Steig, to revisit the outcrop on some Antarctic island that led to that book, commenting that whatever they found could take a year of lab work once he got back; I've seen nothing more, but keep wondering.

Lou Grinzo said...

Hank: Best I know of are the maps here:


SST and anomaly maps for the whole planet, updated daily.

Hank Roberts said...

Lou, what I'm asking about is not sea surface temps alone, but the temperature of the circulation below the surface layer. I know the navies and marine biologists keep track of this. That's what's heating the sediments.

Aside: in the 1960s when I was an undergrad, I learned in a marine biology class that British, French and US marine biologists met regularly to exchange data, because each nation's scientists had been provided sonar gear suitable for tracking the movements of plankton and fish that follow temperature layers up and down over time -- and each instrument had cutouts at the particular frequency range that would detect that nation's submarines. Each group of biologists could sometimes spot the _other_ two nations' submarines hiding in temperature/depth layers suitable to masking their own hardware.

An early tribute to data-sharing.

Robert Grumbine said...

If you want near real time observations of polar currents, there isn't much, particularly not subsurface. If you want model output, there also isn't much. But something should be showing up at http://polar.ncep.noaa.gov/ofs/ in the next year as the current regional (Atlantic ocean) model gets expanded to the globe.

I really don't know the chemistry to know if other things form clathrates or hydrates. The candidate with sulfur would be, I guess, H2S. As I recall it, though, the bacteria that produce that need anoxic conditions, and the polar oceans are well-oxygenated.

I tend to think that meltwater, at least from melting sea ice, is unlikely to drive any major change in the North Atlantic branch of the ocean circulation. A sufficiently large, rapid, melt of Greenland -- that managed to send its meltwater far off the coast of Greenland -- might do it. Moving the main sinking region south of the Greenland-Iceland-Faroes ridge might mean that the water there didn't sink to great depths, and instead the source became the Labrador Sea. Labrador water already sinks fairly deep. Without the densest North Atlantic water, it could sink deeper. At least if nothing changed over the Labrador Sea. Since that area also is experiencing changes ... it gets hard to predict. In general, if something gets to be hard to predict, I figure it's unlikely to work out in my favor.

Robert Grumbine said...

Oops. Stopped too soon. If historical observations will do for you, then the polar observations from the Soviet Union, and then Russia, and the US were made available as part of the 1990s 'Gore-Chernomyrdin' accord. The US copy of the data collection is at the National Snow and Ice Data Center. I don't know of any oil company data being released, but I'm sure they (the companies, that is) have some.

Hank Roberts said...

> Gore-Chernomyrdin
Searching didn't turn up the files, but that was the key search term for Scholar, thanks. Lots of publications have come out referencing those data, e.g.

And, duh, as soon as I looked, I found quite a few gases form clathrates, often different ones in different layers in the same sediments.

Also interesting that the N. Pacific isn't an area where cold water sinks, instead it's an area of upwelling -- but what's coming up is trouble from the deep:

alias Ernest Major said...

Fide Wikipedia, CO2 is one of the gases that forms clathrate hydrates. One would have to dig deeper to find out under what conditions it does so.

I do wonder whether this can be exploited for CO2 sequestration.

GFW said...

Robert, a couple more questions about the meridional overturning circulation.

First, what fraction of the ocean is participating in the flows that we call the MOC at any given time? That question is motivated by thinking of the atmospheric jet streams which only contain a small fraction of the air at any given time. Is the MOC like that, or is it more that all the currents in the ocean add up to the MOC?

Second, if the cycling time is 500 years, then a good fraction of the deep water currently in the MOC was formed during the little ice age. Would that deep water be any colder than deep water formed today (or during the MWP for that matter)? If so, that could create a mechanism for multi-centenary internal variability. I.e. one has a cold or warm period in the northern Atlantic, and roughly 500 years later that would affect the northern Pacific (with some effect on the southern ocean in between, I'm guessing). Is that reasonable? Already well known? Already considered and dismissed?

A third - is there a difference in definition, opinion, or otherwise between your statement of 500 years for the cycling time and this from Wikipedia: "While the bulk of it upwells in the Southern Ocean, the oldest waters (with a transit time of around 1600 years) upwell in the North Pacific (Primeau, 2005)."

Hank Roberts said...

So there's some news:

Has anyone heard anything from Peter Ward lately? I thought of his Medea Hypothesis when I read this:

"... in my short time at AGU, I discovered four scientists who are already creating some form of survival retreat for their family, and they told me there are many more. But they are all too scared of being ostracized in the scientific community if they speak of it. ..."

Robert Grumbine said...

I'll have to follow up your article. Before I shifted to my current line of work, I was looking at deep ocean circulation. Temperature was one part. But the other was to consider the carbon budget in the deep ocean. A warming deep ocean is going to be carrying less carbon out of the atmosphere.

The Medea Hypothesis is not exactly new. The name, perhaps, and particular examples. But ages back, when I was in graduate school, we raised some of these same examples to James Lovelock when he visited our department. I think it's probably going too far to assign 'suicidal' to the planetary system, for much the same reasons as I think it's too far to assign 'stable' to it. But I'll have to look up Ward's work and see what he had to say in detail.

Darn. Sorry I hadn't gotten back to your questions. They're good, and I'll probably take them up in their own post, with link from here.