Guest Post by Willis Eschenbach
It has been known for some time that the “Pacific Warm Pool”, the area just northeast of Australia, has a maximum temperature. It never gets much warmer than around 30 – 31°C. This has been borne out by the Argo floats. I discussed this in passing in “Jason and the Argo Notes“, and “Argo Notes Part 2“. I’d like to expand on this a bit. Let me be clear that I am by no means the originator of the claim that there is a thermostat regulating the maximum ocean temperature. See among many others the Central Equatorial Pacific Experiment. I am merely looking at the Argo data with this thermostat in mind.
First, Figure 1 shows the distribution of all of the ~ 700,000 surface temperature measurements taken by Argo floats to date.
Figure 1. A “histogram” shows how many data points fall in each of the 1°C intervals shown along the bottom axis. The maximum is in the interval 28°-29°C.
The number of temperature records peaks around 29°C, and drops quickly for temperatures above 30°C. This clearly establishes the existence of the mechanism limiting the oceanic temperatures.
What else can the Argo data tell us about this phenomenon? Quite a bit, as it turns out.
First, a look at the year by year evolution of the limit, and how it affects the temperatures at different latitudes.
Figure 2. Annual temperature variations measured by all northern hemisphere argo floats that exceeded 30°C. Temperature observations are colored by latitude. Click on image for full-sized graphic.
A couple points of interest. First, the cap clearly affects only the warm parts of the year. Close to the equator, that is most of the year. The further from the equator, the less of the annual cycle is affected.
Second, the majority of the breakthroughs through the ~30° ceiling that do occur are from areas further from the equator, and are short-lived. By and large, nobody exceeds the speed limit, especially those along the equator.
Figure 3 is a closeup of the years since 2005. I chose this starting point because prior to that the numbers are still changing due to limited coverage. To show how the mechanism is cropping the tops of the warmer parts of the year, I have added a Gaussian average (129 point width) in dark gray for each two-degree latitudinal band from 0°-2°N up to 10°-12°N.
Figure 3. Annual temperature variations measured by all northern hemisphere argo floats that exceeded 30°C. Dark lines have been added to highlight the average annual swings of the data by latitude band. Click on image for full-sized graphic.
As you can see, the warm parts of the yearly cycle have their high points cropped off flat, with the amount cropped increasing with increasing average temperatures.
Finally, here is the corresponding plot for the southern hemisphere:
Figure 4. Annual temperature variations measured by all southern hemisphere argo floats that exceeded 30°C. Click on image for full-sized graphic.
Note that there is less of the southern ocean that reaches 30°C, and it is restricted to areas closer to the equator.
Next, where are these areas that are affected by the temperature cap? I had always thought from the descriptions I’d read that the limitation on ocean temperature was only visible in the “Pacific Warm Pool” to the northeast of Australia. Figure 5 shows the areas which have at some point been over 30°C.
Figure 5. Locations in the ocean which are recorded at some time as having reached or exceeded 30°C.
Figure 5a. A commenter requested a Pacific-centered view of the data. We are nothing if not a full-service website.
Clearly this mechanism operates in a wider variety of oceans and seas than I had realized, not just in the Pacific Warm Pool.
Finally, here is another way to consider the effect of the temperature maximum. Here are the average annual temperature changes by latitude band. I have chosen to look at the northern hemisphere area from 160 to 180 East and from the Equator to 45°N (upper right of Figure 5, outlined in cyan), as it has areas that do and do not reach the ~ 30° maximum.
Figure 6. Average annual temperature swings by latitude band. Two years (the average year , shown twice) are shown for clarity.
Note that at say 40°N, we see the kind of peaked summer high temperatures that we would expect from a T^4 radiation loss plus a T^2 or more evaporative loss. It’s hard to get something warm, and when the heat is turned down it cools off fast. This is why the summer high temperature comes to a point, while the winter low is rounded.
But as the temperature starts to rise towards the ocean maximum, you can see how that sharp peak visible at 40°N starts first to round over, then to flatten out at the top. Curiously, the effect is visible even when the temperatures are well below the maximum ocean temperature.
Speculations on the mechanism
I want to highlight something very important that is often overlooked in discussions of this thermostatic mechanism. It is regulated by temperature, and not by forcing. It is insensitive to excess incoming radiation, whether from CO2 or from the sun. During the part of the year when the incoming radiation would be enough to increase the temperature over ~ 30°, the temperature simply stops rising at 30°. It is no longer a function of the forcing.
This is very important because of the oft-repeated AGW claim that surface temperature is a linear function of forcing, and that when forcing increases (say from CO2) the temperature also has to increase. The ocean proves that this is not true. There is a hard limit on ocean temperature that just doesn’t get exceeded no matter how much the sun shines.
As to the mechanism, to me that is a simple question of the crossing lines. As temperature rises, clouds and thunderstorms increase. This cuts down the incoming energy, as well as cooling the surface in a variety of ways. Next, this same process moves an increasing amount of excess energy polewards. In addition, as temperature rises, parasitic losses (latent and sensible energy transfers from the surface to the atmosphere) also go up.
So … as the amount of total radiation (solar + greenhouse) that is warming any location rises, more and more of the incoming solar radiation is reflected, there are more and more parasitic losses, more cold water and air move from aloft to the surface as cold wind and rain, and a greater and greater percentage of the incoming energy is simply exported out of the area. At some point, those curves have to cross. At some point, losses have to match gains.
When they do cross, all extra incoming energy above that point is simply transferred to the upper atmosphere and thence to the poles. About 30°C is where the curves cross, it is as hot as this particular natural system can get, given the physics of wind, water, and wave.
I make no overarching claims for this mechanism. It is just one more part of the many interlocking threshold-based thermostatic mechanisms that operate at all temporal and spatial scales, from minutes to millennia and kilometres to planet-wide. The mechanisms include things like the decadal oscillations (PDO, AMO, etc), the several-year Nino/Nina swings, the seasonally opposing effects of clouds (warming the winters and cooling the summers), and the hourly changes in clouds and thunderstorms.
All of these work together to maintain the earth within a fairly narrow temperature band, with a temperature drift on the order of ± 0.2% per century. It is the stability of the earth’s climate system which is impressive, not the slight rise over the last century. Until we understand the reasons for the amazing planetary temperature stability, we have no hope of understanding the slight variations in that stability.
My regards to you all,
w.
UPDATE (by Anthony):
Dr. Roger Pielke Sr. has some praise for this essay here:
A temperature limit for ocean water appears to me to imply a similar temperature limit for dew point in surface level troposphere. I largely agree with this.
I have heard of surface level dew point exceeding 30 C in central, east-central, south-central and central-southeast USA. In my experience of hearing about these occurrences, all 3 of these factors apply:
1) The land is grassland, weedland or agricultural land with high moisture, sometimes by irrigation.
2) A heatwave associated with a high pressure system is in progress.
3) Dewpoint exceeds 30 C only before convection dilutes the humid surface-level air with drier air from above the cumulus cloud base level. That level is usually less than 1 km aloft when cumulus clouds start forming. The cumulus cloud base level usually increases to around 1.5 km, sometimes closer to 2 km as convection mixes drier air from aloft with the humid surface layer. The dewpoint typically drops after 10 AM to 1 PM or so in summer heatwave days.
Even then, dewpoints exceeding 28 C (82 F) are quite rare.
Stories of temperatures in degrees F and RH simultaneously in the 90’s in USA hot areas are nearly enough entirely from people remembering RH at wake-up time or breakfast time or early in morning commute time, plus usually lag of .25 to 1.5 hours from measurement to reporting, and *falsely assuming* that RH does not have a major decrease as temperature increases during a hot sunny day.
Roger Sowell says on February 12, 2012 at 12:00 pm
Roger, I advise you not to get your science from Hollywood.
He was not referring to spurious data points measured electronically, stored, transmitted, etc. etc. Spurious data has many “causes” and is “ever present” in large measurement data bases. GK
Willis Eschenbach said in part in a debate with Richard Verney, and I hope I got that right:
February 12, 2012 at 11:38 am
“Nonsense. Yours is the trap that the AGW folks have fallen into, of hypothesizing in advance of the data.”
I don’t think it’s a big sin to hypothesize in advance of the data. The data can be used to test the hypothesis. I thought the “scientific method” involved getting data after hypothesis to increase or decrease support for the hypothesis, as proposed by experiments designed to test the hypothesis.
My biggest problems with “warmist side” are what appear to me to be favoring “fudged data”, and “fudged usage of the data”. A few more years of good data should humiliate them into significantly decreasing their favored projections of global temperature and sea level rise. IPCC may start doing so significantly as early as AR5.
Willis, you wrote in the legend, “Two years shown for clarity.” Is that one year (i.e. all points plotted vs day in the year) double-plotted?
I don’t understand the point of selecting for study those stations that at some time reported a temperature over 30.
Note that at say 40°N, we see the kind of peaked summer high temperatures that we would expect from a T^4 radiation loss plus a T^2 or more evaporative loss.
That requires more explication. I think that you might mean “limited summer peak temperatures”, but how exactly the T^4 and T^2 produce that exact limit needs explication. (with total downward radiation as it is, with absorption rate as is, then there is a limit on how high the surface and air temp can be raised, but what is it exactly?) Especially if you are going to claim later that an increase in CO2 can not possibly raise the limit a little by raising total downwelling radiation. I am not saying that isn’t possible, just that it eventually need to be worked out.
good work, as usual.
I wonder how these tropical water temperatures compare with land temperatures?
Data from Singapore Changi Airport that is close to sea level and the equator shows annual mean temperatures in the range 26.7-28.4 degrees C, maximums 30.1-32.4, and minimums 24.0-25.1 over the period 1982-2011. http://www.tutiempo.net/en/Climate/Singapore_Changi_Airport/486980.htm
So sea temperatures are warmer than the land?
richard verney says:
February 12, 2012 at 9:22 am
The top micron layer of the ocean is cooler than the layer immediately below it. This is because evaporation (and convenction) takes place from the very top micron layer. It therefore is cooled by the latent heat involved.
There is no significant salinity profile that I am aware of in the first metre or so of the ocean. Are you aware of such a salinity prodike, if so please link it.
You may not have understood my point. It may be that due to the laent heat etc involved in evaporation that the top surface layer cannot heat to above 30degC. However since evaporation does not take place from the water which is at 5 cm or 10 or 20 cm below the surface, it may be that water at that specific and particular depths can be heated to above 30degC.
You seem to believe that the Argo buoys can measure the temperature of the “top micron layer”. They can’t.
And no, there isn’t any significant salinity profile in the top metre of the ocean. Which is exactly the point I was making.
I appreciate your effort to find a resting place for Trenberths missing heat but this idea won’t work. It would be better to postulate that the North Atlantic Deep Water and/or the Antarctic Bottom Water has become slightly warmer and simultaneously saltier (and thus equally dense as before). This is physically possible and has the further merit that it takes place in areas where there are no Argo Buoys, and practically no other measurements either, and so can neither be proved or disproved.
Thanks everyone for the replies!
If you check the link from
“DocMartyn says:”,
Page 369, section 8, and corresponding figure 12 on the following page. The math is beyond me, but I’m wondering how it checks out. 30 deg C just feels like an arbitrary number (why not 25? 34? etc).
The paper says, “For seawater, no formulae appear to be available for the change of latent heat with salinity and temperature.” (The paper then comes up with a few equations.)
If the incoming energy/heat from the sun stays relatively constant, is it a simple product of being just the right amount of energy/heat once the water has warmed up to ~30 deg C (at the composition of equatorial seawater)? It feels too easy to be algebraic: input variables (composition of seawater, amount of energy required to evaporate said water, amount of energy incoming to water), solve for limit T(emperature) of water, at which point the cloud formation, convection, and everything else takes over to enforce the limit.
Would this not also help explain why earth’s temperature managed to stay the same over millions of years? If the salinity of the oceans has also stayed relatively the same over time (the last 2.5 billion years or so I think). It would be interesting to see if the “cap” differs in any significant or observable way, however small, depending on salinity – say, the Red Sea, Persian Gulf, or Sargasso Sea versus the Black Sea, Baltic Sea, etc., at comparable latitudes when available.
Willis Eschenbach says: February 12, 2012 at 11:38 am
No, it is not necessary to theorize before one has data, it is a mistake to do so. It is what I described as being the “trap that the AGW folks have fallen into, of hypothesizing in advance of the data.”
_______________
Agree. First look at the data with an open mind.
I believe there is now, or soon will be, enough good-quality satellite data for us to sort out the main causes and effects of this complex subject.
What we may are missing in the current satellite data is and inflexion point, such as occurred in ~1945 and ~1975, between a warming cycle and a cooling cycle (or vice-versa).
I suspect we may be seeing one such inflexion now, or we will soon, and that should tell us a great deal more about the relationship between atmospheric CO2 concentrations, air and ocean temperatures, and the impact of the Sun and planets. In another decade or so, we should have it – maybe less.
For years, I have suggested to some of the finer minds in this debate that they are focused on the minutiae, and the scientific truth of the big picture in climate science may be right in front of them. It is a very interesting time.
Further data to illustrate the point of the mythical 30 degrees C maximum sea surface temperature, this from NOAA’s National Data Buoy Center at
http://www.ndbc.noaa.gov/view_climplot.php?station=42002&meas=st
This particular buoy is 42002, and measures sea temperature at one meter below the surface. The average temperature for the hottest month, July, is 30 degrees C with maximum of a bit more than 34 degrees C.
Buoy 42002 is “W GULF 207 NM East of Brownsville, TX,” translated as in the Western Gulf of Mexico, 207 nautical miles east of Brownsville, Texas, USA. Brownsville is the southern-most city in Texas, on the coast and on the border with Mexico.
Again, there exists no such limit as 30 degrees C for sea surface temperatures. If there were a limit, then the above data would not exist. I have not strolled through the entire available data on sea surface temperatures, but there are other ocean/bay/gulf locations that are hotter than buoy 42002.
As skeptics are fond of saying, let’s look at the data, and not cherry-pick one data set and draw universal conclusions.
Looking at the lower temperatures at the 30 degree latitude, it dropped to +10 in 2002 and down to almost 0 in 2007. This would seem to be an incredible range (10 degrees) in a world where climate science is trying to come up with a global average temperature. It is also interesting that the big mid-latitude temperature drop (2007) was occurring at the same time the summer Arctic ice was reaching a minimum. I wonder if there is a correlation between Arctic ice extent and minimum ocean temperatures at mid-latitude in the Northern Hemisphere?
The coldest equatorial part of our world’s oceans, according to your “Figure 5”, seems to be just where El Nińo lives – i.e. around the west coast of South America. I wonder why – –
More hot sea surface temperatures from NOAA’s buoy data base: at Naples, Florida, and Key West, Florida, the average temperatures for July and August are 87 degrees F (30.6 degrees C); thus it is highly likely that the excursions around that average are greater than 31 or 32 degrees C.
http://www.nodc.noaa.gov/dsdt/cwtg/egof.html
Those are not the hottest bodies of ocean around, either, merely ones easily found in a few moments’ searching.
“Richard M says:
February 12, 2012 at 5:54 am
It does appear that the limit only shows up in the open ocean. The coastal areas and seas have no problem going over 30°. It may be informative to look only at these areas and see if they have another limiting factor”
Salinity. As the evaporation rate increases, salinity increase, and then the rate of evaporation drops. With a fixed heat input the higher the salt content, the higher the temperature and the lower the rate of evaporation.
Thus, salinity should plot with Tmax.
…or at about 30 to 31 C any marginal increase is heat flux is carried back up by evaporation.
How about this –
The AGW proponents have it totally wrong – the Sun’s radiation warming the Earths’ surface is not the measly 170 W/sq m 24/7 as deceptively stated but wayyyy more during the day.
The atmosphere and oceans REDUCE the surface temperature during the day to much lower than it would otherwise be – evidence: the day temperature of the Moon with no atmosphere shows the Sun’s radiation could fry us during the day.
I can never understand the dismissal of this important fact – without an atmosphere and water we’d all be burnt to a crisp by the Sun’s radiation.
To me it seems likely that CO2 released from warmer ocean waters is the most significant source of increasing atmospheric CO2 and may be a “safety valve” mechanism for a global thermostat providing extra radiative transport for heat to space and actually providing no radiative forcing – not a positive one anyway.
Wouldn’t that be funny – increased atmospheric CO2 cools the Earth ?
Thanks Willis for an outstanding article. It is great illustration of the Earth’s built-in thermostat. It has great implications for the global warming theory. Evaporation of water drives increasing convective cooling and energy is transported to the poles and radiated to space. Insolation is limited by increasing cloudiness. There is a large negative feedback in the climate and this is part of it. It all overcomes peak insolation in the summers limiting the maximum temperature of the ocean.
The mild effect of CO2 is a bit player in this action.
Figure 3(a) of this paper (Sea Surface Temperature and it’s Variability in the Indonesian Region, 2005) also show many near-coastal areas with sea surface temperatures greater than 30 degrees C. The mean monthly temperature for August is shown in Figure 3(a), with the highest temperatures at 30.5 to 31 degrees C. When variability is considered, temperatures of 32 to 34 are quite likely – although the paper does not show the raw data.
http://www.tos.org/oceanography/archive/18-4_qu.pdf
The paper also states that surface winds are the key factor in sea surface temperatures.
I note in passing, reference the “Hollywood science” from a commenter above, that none of the paper’s authors are from Hollywood.
G. Karst says:
February 12, 2012 at 8:59 am
Yes there certainly IS a mechanism to limit sea surface temperature. It’s called the sun. The blackbody temperature is the theoretical maximum. There needs to be more energy than the sun currently provides to exceed it. The earth of course is a gray body but that can only lower the S-B blackbody temperature not raise it.
Greenhouse gases work in a way that can be equated with albedo change because in the end they cause the gray body to be a little less gray and a little more black. Average albedo of the earth is between 30 and 40 percent so there’s a lot of room to make it more black and thus raise the surface temperature. But make no mistake there is a maximum and that is the S-B blackbody (not grayt body) temperature.
There’ s a fair amount of confusion because the infamous 255K temperature of the earth sans greenhouse gases is a gray body number not a black body number. Anything that can effectively lower the albedo will raise the maximum possible average temperature.
“”””” February 12, 2012 at 12:32 am
Try this for size. You can not heat water from above because of surface tension. The only energy entering the ocean is via the sun’s rays which penetrate the surface tension no problem.The sun is relatively stable and so is the temperature. “””””
Surface tension has nothing to do with it. The bulk of the solar spectrum (energy) goes deeper than the surface waters (several metres) so it doesn’t directly warm the surface. But radiation in the 1 to 5 micron region which people think of as “heat” (it isn’t) is strongly absorbed in water, particularly at 3 microns, where 10 microns of water absorbs over 99% of the radiant energy, and converts it into “heat” (waste).
It is the Temperature of the very surface layer of the water that determines if evaporation takes place. The Temperature of the atmosphere has very little to do with evaporation; but it does determine the amount of water vapor the atmosphere can contain.
A molecule of H2O near the top few molecular layers of the surface, has no knowledge whatsoever of the atmospheric Temperature, or even that the atmosphere exists. It does know the mean binding energy to be overcome to escape from the surface; and only after escape, does the H2O molecule become aware of the atmosphere and its Temperature. Scientists are still researching the question of exactly where the brains of the H2O molecule reside.
Likewise, evaporation is one of the key elements of the Temperature brick wall that Willis is talking about (Nice presentation Willis). Evaporation not only cools the water surface (look at hurricane tracks) but it transports a lot of latent heat to the immediate atmosphere (590 calories per gram or so). Winds and storms carry off the H2O so they assist the evapration by stopping the accumulation of H2O above the surface.
As every process chemist knows, a properly designed chemical process, has to remove the reaction products from the interface, to keep the process going, otherwise the bidirectional reaction comes to a screeching halt.
Actually I live on the shore of a deep lake at 30 degrees north (sub-tropics). Max surface temperature record that I know of is just under 32C and average summer max is probably about 30C. It cools off pretty fast with depth because the clarity is far lower than open ocean, there’s little mixing from waves, so the sunlight gets completely absorbed closer to the surface and doesn’t mix downward well.
I expect there’s something pretty close to a law for deep bodies of water with sufficient mixing so that there’s little diurnal temperature variation where maximum possible temperature is then purely a function of latitude.
Is there a way to look at the diurnal temp data by season and lat/long?
The 30C SST limit means the sun (TSI) has no more effect on the Earth’s climate than GHG changes.
Assuming the 30C limit is caused by the hydrological cycle, and I can’t see another mechanism, the only things which could cause climate change are things that affect the phase changes of water – aerosols, and perhaps GCRs and UV changes.
Outstanding – and a pleasure to read.
(In all the stuff written on climate, there’s a language fingerprint to look out for – snottiness and arrogance. If it’s there, I usually stop reading.)
Commenting on ferd berple et al comments on the cyclone sideline :
“As I recall from my sailing days, 28C is the ocean surface temperature at which you are at risk for cyclone formation. The ocean surface temperature will not go much above this, because the energy goes into storm formation.”
About right. Here (NE Australia) cyclones tend to form in the South Solomon Sea (eg the hot spot on Willis’ graphs) when there is a thunderstorm and the sea temperature exceeds 24°C. They then tend to track south and west (except when they do something different of course). If the track is into waters below 24°C, the cyclone tends to fizzle.
“Cyclones (huricane/typhoon) are not distributed evenly around the globe. They tend to affect the east coast of continents more than the west coast. ”
Not here – north-west coast has higher frequency and strength.
The reason behind this is a well known fact, and are one condition to form hurricanes:
“…waters of this temperature (26.5C) cause the overlying atmosphere to be unstable enough to sustain convection and thunderstorms”
http://en.wikipedia.org/wiki/Tropical_cyclone
Until we understand the reasons for the amazing planetary temperature stability, we have no hope of understanding the slight variations in that stability
Then, the question is raised. Though it seems that the answer would imply a change of paradigm.