Arguments
Levitus et al. Find Global Warming Continues to Heat the Oceans
Posted on 25 April 2012 by dana1981
Levitus et al. had previously published updated ocean heat content (OHC) data on the National Oceanic Data Center (NODC) website, labeled as "Levitus et al. in preparation" (Figure 1).
Figure 1: Global OHC for the upper 2000 meters of oceans (NODC)
Levitus et al. (2012) is now in press, discussing the OHC data published by NODC. Figure 2 below (from Levitus et al.) presents the data in a similar fashion to Figure 1 above, but breaks out the data to show the OHC contribution from the 700 to 2000 meter ocean layer.
Figure 2: Time series for the World Ocean of ocean heat content (1022 J) for the 0-2000m (red) and 700-2000m (black) layers based on running pentadal (five-year) analyses. Reference period is 1955-2006. Each pentadal estimate is plotted at the midpoint of the 5-year period. The vertical bars represent +/- 2 times the standard error of the mean (S.E.) about the pentadal estimate for the 0-2000m estimates and the grey-shaded area represent +/- 2*S.E. about the pentadal estimate for the 700-2000m estimates. The blue bar chart at the bottom represents the percentage of one-degree squares (globally) that have at least four pentadal one-degree square anomaly values used in their computation at 700m depth. Blue line is the same as for the bar chart but for 2000m depth. From Levitus et al. (2012)
Data and Main Findings
Levitus et al. use OHC measurements from ARGO floats corrected for systematic errors, as well as data from expendable bathythermographs (XBT) and mechanical bathythermographs (MBT), with the necessary corrections applied. A bathythermograph is an instrument which has a temperature sensor and is thrown overboard from ships to record pressure and temperature changes as it drops through the water. These were the main instruments used to measure OHC before the ARGO float network was deployed starting about a decade ago to provide more accurate and consistent data.
The authors discuss some of their key results:
"The World Ocean accounts for approximately 93% of the warming of the earth system that has occurred since 1955. The 700-2000m ocean layer accounted for approximately one-third of the warming of the 0-2000m layer of the World Ocean. The thermosteric component of sea level trend was 0.54 ±.05 mm yr-1 for the 0-2000m layer and 0.41 ±.04 mm yr-1 for the 0-700m layer of the World Ocean for 1955-2010."
Significant Ocean Heating Below 700 Meters
This harkens back to some of the discussions between Skeptical Science and Roger Pielke Sr., who rightly noted that we cannot adequately measure global warming by considering surface temperatures alone, as most of the global heating has accumulated in the oceans. However, we noted at the time that Dr. Pielke was only considering the heating of the upper 700 meter ocean layer, which is also an incomplete measure of global warming. Levitus et al. confirm this by noting that in the upper 2000 meters, approximately one-third of the total heating has occurred below 700 meters.
Dr. Pielke had argued based on the 700 meter OHC data that global warming had slowed. Levitus et al. note this leveling off in the upper 700 meters in recent years, but that this recent flattening is much less apparent in the 2000 meter data, meaning that more heat is being stored in the 700-2000 meter layer recently (Figure 2).
"One feature of our results is that the previous multidecadal increase in OHC700 that we have reported [Levitus et al., 2009] (updated estimates available online at www.nodc.noaa.gov) leveled off during the past several years. This leveling is not as pronounced in our OHC2000 estimates indicating that heat is being stored in the 700-2000m layer as we have shown here."
Consistent with Meehl's Hiatus Periods
Meehl et al. (2011) examined what climate models predict will happen during 'hiatus' decades in which surface temperatures plateau for short periods of time. They found that during these hiatus decades, less heat accumulates in the upper layers of the ocean, and more accumulates in the deeper layers (Figure 3).
Figure 3: Left: composite global linear trends for hiatus decades (red bars) and all other decades (green bars) for top of the atmosphere (TOA) net radiation (positive values denote net energy entering the system). Right: global ocean heat-content (HC) decadal trends (1023 J per decade) for the upper ocean (surface to 300 m) and two deeper ocean layers (300–750m and 750 m–bottom), with error bars defined as +/- one standard error x1.86 to be consistent with a 5% significance level from a one-sided Student t-test. From Meehl et al. (2011)
The Meehl model results are exactly what Levitus find is happening. We are in the midst of a hiatus decade where global surface warming has been dampened, the increase of the upper OHC has slowed, but more heat is going into the deeper ocean layers.
Putting Ocean Heating Into Perspective
The amount of global warming which has gone into the oceans over the past 55 years is quite impressive.
"The global linear trend of OHC2000 is 0.43x1022 J yr-1 for 1955-2010 which corresponds to a total increase in heat content of 24.0±1.9x1022 J"
This is an immense amount of energy being added to the oceans which Levitus et al. put into perspective (emphasis added):
"We have estimated an increase of 24x1022 J representing a volume mean warming of 0.09°C of the 0-2000m layer of the World Ocean. If this heat were instantly transferred to the lower 10 km of the global atmosphere it would result in a volume mean warming of this atmospheric layer by approximately 36°C (65°F)."
Levitus et al. note that of course this heat won't be instantly transferred to the atmosphere (fortunately!), and that this comparison is simply intended to illustrate the immense amount of energy being stored by the oceans.
This heating amounts to 136 trillion Joules per second (Watts), which as Glenn Tramblyn noted in a previous post, is the equivalent of more than two Hiroshima "Little Boy" atomic bomb detonations per second, every second over a 55-year period. And Levitus et al. note that this immense ocean heating has not slowed in recent years - more of it has simply gone into the deeper ocean layers.
Spoiler Alert
Coincidentally, a team led by several Skeptical Science contributors recently submitted a paper for review which uses this NODC OHC data from Levitus et al. and comes to many of the same conclusions regarding continued global warming. With any luck, the paper will be published in a few months and we'll have more to say on the subject at that time. In the meantime, Levitus et al. have once again reminded us that although the surface warming may have been dampened in recent years, global warming hasn't magically vanished, and that heat stored in the oceans will eventually come back to haunt us.
dvaytw - Note that the experimentally measured parameter that I discussed above is the thermal gradient of the ocean skin layer and its changes WRT downward atmospheric IR, averaged over multiple days. The gradient is the result of all acting processes, including IR, ocean turbulence, winds, evaporation and condensation, passing fish, etc - indicating that these other processes (while undoubtedly affecting the gradient to some degree) do not override or overwhelm the IR related changes.
Any increase in evaporation in particular (one of the more common off-the-cuff objections to this information), results in its own increase in GHG absorption and re-radiation near the surface - increasing the IR input to the skin layer, and again acting to reduce the thermal gradient and ocean heat loss.
Thanks for all the info, guys.
The dude is also trying to say that the ocean heat content is just climatologists making AGW theory unfalsifiable by changing their story after the fact. I recall there was a study quite a while back predicting oceans would take up most of the heat. I know Hansen had something about this as early as 1997, but I remember there was something even earlier. Does anyone know what this was?
dvaytw @52.
Oceans acting as the damper on any rising temperatures is not too difficult to establish. So a reference could well be found way back in time. But before 1997? Anything post 1990 will be preceeded by the IPCC FAR. A quick look at the Summary yields.
IPCC WG1 1990 Policymakers' Summary p xxvi
Info box - EQUILIBRIUM AND REALISED CLIMATE CHANGE
"When the radiative forcing on the earth-atmosphere system is changed, for example by increasing greenhouse gas concentrations, the atmosphere will try to respond (by warming) immediately But the atmosphere is closely coupled to the oceans, so in order tor the air to be warmed by the greenhouse effect, the oceans also have to be warmed, because of their thermal capacity this takes decades or centuries This exchange of heat between atmosphere and ocean will act to slow down the temperature rise forced by the gieenhouse effect."
IPCC WG1 Policymakers Summary p.xxix
"Global-mean temperature alone is an inadequate indicator of greenhouse-gas-induced climatic change Identifying the causes of any global-mean temperatuie change requires examination ot other aspects ol the changing climate, particularly its spatial and temproal characterisitcs (and?) the man-made climate change signal."
Guys thanks again for all input. I was wondering how you'd respond to this point from the Eschenbach article:
"The good news is that we’re measuring ocean heat content (OHC), so it’s very different from temperature. We can simply subtract the changes in the 700 metre level OHC from the 2000 metre level OHC changes, and what is left is the change in heat content for the layer from 700 metres down to 2000 metres. Can’t do that with temperature. Figure 3 shows the same OHC data as in Figure 2, except split out into distinct and separate layers, at the same scale. as Figure 2...
I was quite surprised by this result. Once I split the information up so that I could see the changes in each of the layers separately, much of the apparent change post-2001 disappeared. In Figure 2 there’s not a lot of change in 2001."
dvaytw @54.
Coinsidently, I think that's been covered up-thread @38. Still it is perhaps wothy of repeating being such a forehead-smackingly stupid piece of analysis (although you need not look far on Wattsupia to find similar stuff).
The fool was only "quite surprised" to find that a number as big as 205 could be converted into three numbers as small as 45+60+100. But only being "quite surprised" didn't stop him yet again making a fool of himself. He was after all a fool so he couldn't wait to tell the rest of the class. 'Look,' he said 'Look how the wiggles on the graph stay so close to the line at the bottom when I split the numbers up. How fantastic this arithmateratic is - numbers made small by cutting them up into seperate pieces!'
Ohhh... that comment went over my head the first time. Thanks!
dvaytw:
Apart from MA Rodger's comment, perhaps it helps to know how heat content and temperature are related. The graphs above are actually changes in heat content, relative to a baseline. This has to be the case, as the oceans did not have negative heat content in the 1960s.
Now, to get the change in heat content, you simply take the change in temperature, multiply it by the heat capacity (in J/kg/degree or J/m3/degree, whichever you prefer), and then multiply that by the number of kg (or m3, whichever you used for heat capacity), and magically you have now converted temperature changes (in degrees) to heat content changes (in Joules). To a first approximation, you could consider heat capactiy of water to be a constant for all the ocean, and then the temperature-->heat content conversion and averaging becomes one of weighting the layer temperatures by the volume or mass in each layer. It's alittle more complex when you start to take the slight variations in heat capacity due to temperature, denisty, etc., into account, but it doesn't change the big picture.
Granted, my PhD thesis (partly) involved developing a numerical model for heat transfer, so I may know a bit more about this than most, but does anyone here really think that a person that who thinks Ocean Heat Content is "very different from temperature" should be considered a reliable source? Maybe it's a surprisingly different thing to a Blog Scientist, but it's high school physics. It doesn't take much to conclude that Eschenbach is in way over his head.
Bob Loblaw thanks for that!
And sorry to be a pest, but I was wondering if anyone could give me a good analogy to explain why my opponent is wrong on the issue of evaporation. Something along the lines of, "If what you're saying was true, it would mean..."
In general terms I get what you guys are saying, but I don't get it well enough to analogize, and a nice analogy always makes a point stick better.
dvaytw - Evaporation is controlled by both temperatures and by local relative humidity. Relative humidity worldwide has changed less than 0.6% worldwide over the last 30 years (Dai 2005), so this continues to act as a limiting factor - the atmosphere can only hold so much moisture at any temperature.
Even more importantly, though, is that water vapor is a powerful greenhouse gas. Additional water vapor (absolute humidity) given by the same relative humidity at higher temperatures means more IR trapped, a warming feedback rather than a cooling one (Atmospheric Infrared Sounder data).
Last but most certainly not least, the data regarding IR driven skin layer gradients controlling heat loss from the oceans shows, over a long enough period for any evaporative feedback, that evaporative effects do not override the gradient effects. In other words, such claims are flatly contradicted by the data.
In general, as several people have commented, the claims that "evaporation will stop warming" are just another "single-cause" argument, not considering other contributing factors, much as in discussions of CO2 without considering other forcings. And as such, they are incorrect.
KR,
Are you sure of this number?
0.6% is pretty low. According to another SkS thread, it is expected that each C increase should increase humidity levels by 6 - 7.5% (H2O is most important GHG). If temps have risen by ~ 0.5 C over the past 30 years, humidity should have risen by 3 3.75% over that time, not 0.6%.
Tell me if I have mixed up percentage increases and percent point increases, please.
franklefkin - Keep in mind the difference between relative humidity (the percentage of how much water air at a particular temperature can hold without saturation) and absolute humidity (the mass of water in an air volume) - these are very different numbers.
Relative humidity, including over the oceans, has changed very little globally, meaning that the 'evaporation thermostat' hypothesis is not supported by the data. Evaporation/precipitation drives any heat loss, and while regional effects have been strong, they to a large extent cancel out globally, with only a small and difficult to detect mean global increase (Zhang 2007).
Absolute humidity, on the other hand, has been increasing - and with each 1C increase in temperature trapping an additional 2 W/m2, meaning that increased water vapor with temperature has an overall positive feedback, not a negative/cancelling one.
KR,
Thanks, its been a long day and I knew the answer was something simple like that.
Hey guys I'm sorry to keep pushing here, but I'm trying to communicate these issues to people who don't understand the science well - which is difficult, when I don't either! I read the "Real Climate" article on GHG ocean heating, but this paragraph in particular I found hard (and obviously it is key):
"The figure below shows just the signal we are seeking. There is a clear dependence of the skin temperature difference on the net infrared forcing. The net forcing is negative as the effective temperature of the clear and cloudy sky is less than the ocean skin temperature, and it approaches values closer to zero when the sky is cloudy. This corresponds to increased greenhouse gas emission reaching the sea surface."
Can someone translate this into plainer English?
Dvawtw:
"There is a clear dependence of the skin temperature difference on the net infrared forcing." The Ocean surface emits IR radiation upward and absorbs IR from the atmosphere (backradiation). The temperature of the ocean surface depends on the difference between how much it emits and how much it absorbs (the net forcing).
"The net forcing is negative as the effective temperature of the clear and cloudy sky is less than the ocean skin temperature, and it approaches values closer to zero when the sky is cloudy. " The ocean is usually warmer than the atmosphere so the ocean loses energy to the atmosphere. On clear days the atmosphere is cooler and the ocean radiates energy into the atmosphere. On cloudy days the net forcing is small and the ocean loses little energy.
" This corresponds to increased greenhouse gas emission reaching the sea surface." On cloudy days more IR radiation from the atmosphere hits the ocean surface so the net forcing is small.
Is that any better?
The ocean is warmer than the atmosphere so the ocean emits IR to cool off. They measure a change in the ocean surface temperature that depends on the difference between how much the ocean emits upward and how much it abosrbs from Greenhouse gases emitting downward. When GHG increases, the downward emissions increase and that warms the ocean.
Dvawtw... You might also want to challenge your skeptic friends on their eyeball rolls related to models. (Sorry, I know it's off topic here. We can move elsewhere if there is more to discuss.) Your friends use models every day, even though they may not realize it. Models are not wrong. They're "boundary condition" experiments, like the model you use in your head (or planner) for planning your own activities through the day or week. They are not exact depictions of weather, which are "initial condition" experiments. Nor is your mental or planner model expected to exactly depict your activities.
More on that issue here.
dvaytw - The ocean skin layer, constrained by surface tension, is thicker than IR can penetrate and again with surface tension not convecting. Only thermal conduction can remove energy from deeper penetrating visible sunlight.
Downward IR warms the top of the skin layer, which decreases the thermal gradient across it, and less energy moves to the atmosphere.
Analogy: Take a metal rod, heat one end relative to the other. If the difference in temperature is, say, 10C, a certain amount of energy (heat) will flow from one end to another, at a rate determined by the thermal conductivity of the rod. If the difference between ends is only 5C (say with IR warming of the cool end), then less energy will be conducted through the rod, as the gradient is lower. That is equivalent to the thermal conduction of the ocean skin layer - a warmer surface means less energy flowing through the skin layer.
Got it guys, thanks!
In this thread I posted a link to Schmitt et al 2005 and stated that "The bottom line is we don't really know how much 'missing' atmospheric heat has wound up in the oceans." Dave123 replied that "the paper makes the case that mixing was faster than anticipated" and quoted form the paper here. He suggested checking cites so I did, although not all 71 of them. Searching within the cites for GCM was dry, but searching for ARGO brought up this paper: website: scientiamarina.revistas.csic.es path: /index.php/scientiamarina/article/viewFile/1384/1488 that suggests that the Schmitt results were localized that were not present in the rest of their Atlantic ocean cross section. I didn't pursue the cites further.
But more apropos is the Levitus papers themselves such as this one: World ocean heat content and thermosteric sea level change (0–2000 m), 1955–2010 In order to fill in missing data for the 700-2000m depth they have to model how much heat flows down from the 0-700 depth which has good coverage. The relatively small temperature changes that SASM asks about in post 28 is answered by Tom two posts later as being 0.1C over the period. However the annual energy change is roughly 1022 Joules so the temperature change for 6 x 1023 cubic cm of seawater with 4 J/g/C is 0.002C per year.
The answer to KR's reply a few posts later that the large number of sample points reduces the error: there are only 4 data points in each one degree grid square in the model in Levitus. The mixing shown in the Schmitt paper and the other paper linked above requires simulation at the 1 degree cell resolution to simulate the mixing processes.
The latter paper also notes that the accuracy of temperature measurement was 0.002C so measuring a 0.002C change is problematic. Also the ARGO network has about 3 degree spacing according to their website which makes it basically impossible to simulate the mixing, so it must be parameterized.
Eric (skeptic) - There are, quite frankly, a number of misconceptions present in your last post.
"The latter paper also notes that the accuracy of temperature measurement was 0.002C so measuring a 0.002C change is problematic." - Incorrect. You seem not to understand how the margin of error changes with multiple measurements, the law of large numbers. Measurement error decreases with the number of samples, with the rough proportion of 1/sqrt(n). The uncertainty of a single measurement is a vast overstatement of the uncertainty of multple measurements.
WRT to sampling grid size, regardless of whether or not the sample spacing is fine enough to resolve individual 'salt fingers', the mass effect of heat transfer is quite measureable. And that holds whether or not the phenomena driving the mass transfer of energy is below the scale requiring parameterization in a model - if it has a significant effect, that will show up in the larger scale measurements.
The core of your post, however, appears to be a claim that there are inaccuracies not accounted for in Levitus et al and the literature in general, inaccuracies sufficient to invalidate the presented data. I would strongly suggest you actually read Levitus et al 2012, in particular the appendix labeled "Error Estimates of Objectively Analyzed Oceanographic Data", and if you have any serious objections to their methods point them out. Otherwise, you are simply making vague and unsupported insinuations of inaccuracies without evidence - and such statements without evidence can in turn be dismissed without evidence.
Eric- So what you've said is that after I pointed out some gaps, you went back and tried to fix it. That's good. If you truly aspire to be a compelling sceptic, you need to do this first, not as an afterthought. And I'd suggest you should assume that the people writing those papers have done so. Peer reviewers tend to be a little reactive when they see a paper than hasn't properly acknowledged prior work in a field. It's one of the easiest things to 'ding' a paper for.
Let me add one other point about accuracy that you seem to have missed and KR didn't discuss. We are interested in the change of temperature in the deep oceans. Thus the standard procedure of computing the temperature anamoly for each device and working range applies. A given Argo Float measures a hypothetical temperature of 8.023 C at a depth of 1500 meters. The measurement is repeated on a regular basis of 5 years. You can run a regression line through the temperatures, which does much the same thing as taking the average of all the temperatures and substracting that average from each data point. Now you have the anamolies, and those are independent of whether the "true" temperature was 8.021 or 8.025. If a given thermocouple reads a little high, it always reads a little high. For a narrow (or not so narrow) range of temperatures you can treat that error as a constant offset. Looking at the anomoly also allows us to pool data across multiple floats and thus create a statistic that represents the total ocean heat content change.
Add this to the law of large numbers and you get a statistically sound measurement of an increase in heat going into the lower depths.
KR and Dave123, thanks for the replies. I had read a website a few months ago that explained the way Levitus et all assimilated the ARGO data into their climate model. Basically the model would predict temperatures, they would compare those to the real temperature measurements, then adjust the model for a better fit in an iterative process. The key point is that the model provided the energy change results not the measurements directly. When I find that website again, I will read through it and post what I find here.
I read through the appendix of Levitus et al 2012 and saw a process for estimating the error of a data smoothing algorithm. I'm not sure how I can evaluate that yet.
The crux of the issue is determining how much heat has percolated into the deeper ocean. KR, do you mean that the "mass effect" allows the integration of many estimates of heat transfer to determine an aggregate heat transfer over large regions? That's true as long as the factors and effects are somewhat uniform. I'm not sure that is true in cases where both heat transfer and heat accumulation vary over relatively small regions. Nonuniformity seems to be the case looking at figures S5 (which is averaged longitudinally) and S6 (which is not).
And Dave123, a linear regression through data points will only work if there is a somewhat linear rise (or fall) in temperature. That's true for some basins to some extent and true for the world, but is only true sometimes for individual floats. A higher variance in any particular float means a higher standard deviation and uncertainty.
But as I said above, I owe you a better answer once I find the web site (it was not a paper, but a NOAA website describing the assimilation process).
Eric (skeptic) - Again, I would refer you to the actual literature, Levitus et al 2012, and the appendix on error calculation.
They describe in considerable detail their use of irregularly sampled temperature measurements and resampling down to 1 degree fields - and in their analysis they discuss the standard deviation of temperatures over distance:
Your concern would be valid only if there was sufficient short range variation in temperature anomaly, sufficient non-uniformity, to make the measurements unusable - and as Levitus et al have noted, that is not the case. Temperature anomaly data is sufficiently homogenous for irregular and relatively sparse sampling to track, and the uncertainties involved are part and parcel of their error bars.
Keep in mind that, like in atmospheric temperature anomalies (Hansen & Lebedeff 1987, while temperatures may vary over short spatial scales oceanic temperature anomalies have a far stronger correlation over distance. Lacking evidence contradicting the Levitus error calculations (and while you have expressed concerns, you have not presented either data or references supporting such), I would consider their data valid.
Probably worth also have a scan of Von Schuckmann and La Traon 2011, "How well can we derive Global Ocean Indicators from Argo data?"
Thanks scaddenp. I quickly scanned it and will address their conclusion of relatively low uncertainty (20% or less). But the deeper ocean has more uncertainty which includes model uncertainty from the assimilation method that I mentioned above. That will be my main focus.
Dont forget in your analysis that deep ocean heat transport is still constrained by steric sealevel rise.