Reposted from Dr. Roy Spencer’s site
January 15th, 2020 by Roy W. Spencer, Ph. D.
The increasing global ocean heat content (OHC) is often pointed to as the most quantitative way to monitor long-term changes in the global energy balance, which is believed to have been altered by anthropogenic greenhouse gas emissions. The challenge is that long-term temperature changes in the ocean below the top hundred meters or so become exceedingly small and difficult to measure. The newer network of Argo floats since the early 2000s has improved global coverage dramatically.
A new Cheng et al. (2020) paper describing record warm ocean temperatures in 2019 has been discussed by Willis Eschenbach who correctly reminds us that such “record setting” changes in the 0-2000 m ocean heat content (reported in Zettajoules, which is 10^^21 Joules) amount to exceedingly small temperature changes. I calculate from their data that 2019 was only 0.004 0.009 deg. C warmer than 2018.
Over the years I have frequently pointed out that the global energy imbalance (less than 1 W/m2) corresponding to such small rates of warming is much smaller than the accuracy with which we know the natural energy flows (1 part in 300 or so), which means Mother Nature could be responsible for the warming and we wouldn’t even know it.
The Cheng (2017) dataset of 0-2000m ocean heat content changes extends the OHC record back to 1940 (with little global coverage) and now up through 2019. The methodology of that dataset uses optimum interpolation techniques to intelligently extend the geographic coverage of limited data. I’m not going to critique that methodology here, and I agree with those who argue creating data where it does not exist is not the same as having real data. Instead I want to answer the question:
If we take the 1940-2019 global OHC data (as well as observed sea surface temperature data) at face value, and assume all of the warming trend was human-caused, what does it imply regarding equilibrium climate sensitivity (ECS)?
Let’s assume ALL of the warming of the deep oceans since 1940 has been human-caused, and that the Cheng dataset accurately captures that. Furthermore, let’s assume that the HadSST sea surface temperature dataset covering the same period of time is also accurate, and that the RCP radiative forcing scenario used by the CMIP5 climate models also represents reality.
I updated my 1D model of ocean temperature with the Cheng data so that I could match its warming trend over the 80-year period 1940-2019. That model also includes El Nino and La Nina (ENSO) variability to capture year-to-year temperature changes. The resulting fit I get with an assumed equilibrium climate sensitivity of 1.85 deg. C is shown in the following figure.
Fig. 1. Deep-ocean temperature variations 1940-2019 explained with a 2-layer energy budget model forced with RCP6 radiative forcing scenario and a model climate sensitivity of 1.85 deg. C. The model also matches the 1940-2019 and 1979-2019 observed sea surface temperature trends to about 0.01 C/decade. If ENSO effects are not included in the model, the ECS is reduced to 1.7 deg. C.
Thus, based upon basic energy budget considerations in a 2-layer ocean model, we can explain the IPCC-sanctioned global temperature datasets with a climate sensitivity of only 1.85 deg. C. And even that assumes that ALL of the warming is due to humans which, as I mentioned before, is not known since the global energy imbalance involved is much smaller than the accuracy with which we know natural energy flows.
If I turn off the ENSO forcing I have in the model, then after readjusting the model free parameters to once again match the observed temperature trends, I get about 1.7 deg. C climate ECS. In that case, there are only 3 model adjustable parameters (ECS, the ocean top layer thickness [18 m], and the assumed rate or energy exchange between the top layer and the rest of the 0-2000m layer, [2.1 W/m2 per deg C difference in layer temperatures away from energy equilibrium]). Otherwise, there are 7 model adjustable parameters in the model with ENSO effects turned on.
For those who claim my model is akin to John von Neumann’s famous claim that with 5 variables he can fit an elephant and make its trunk wiggle, I should point out that none of the model’s adjustable parameters (mostly scaling factors) vary in time. They apply equally to each monthly time step from 1765 through 2019. The long-term behavior of the model in terms of trends is mainly governed by (1) the assumed radiative forcing history (RCP6), (2) the assumed rate of heat storage (or extraction) in the deep ocean as the surface warms (or cools), and (3) the assumed climate sensitivity, all within an energy budget model with physical units.
My conclusion is that the observed trends in both surface and deep-layer temperature in the global oceans correspond to low climate sensitivity, only about 50% of what IPCC climate models produce. This is the same conclusion as Lewis & Curry made using similar energy budget considerations, but applied to two different averaging periods about 100 years apart rather than (as I have done) in a time-dependent forcing-feedback model.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
I think we should stick with looking at SST as the development of SST will tell us which way the wind will be blowing.
I am puzzled to find that SST in the NH is much faster than in in the SH.
http://www.woodfortrees.org/plot/hadsst3gl/from:1964/plot/hadsst3nh/from:1964/plot/hadsst3sh/from:1964/plot/hadsst3gl/from:1964/trend
I do have an answer from Julian Flood on this, which would make sense to me. However, looking at it from 1850 makes me think that it must be the sun?
see here:
http://www.woodfortrees.org/plot/hadsst3gl/from:1850/plot/hadsst3nh/from:1850/plot/hadsst3sh/from:1850/plot/hadsst3gl/from:1850/trend
I am puzzled to find that SST in the NH is much faster than in in the SH.
should be
I am puzzled to find that SST in the NH is rising much faster than in in the SH.
Ian,
“Given the large uncertainties associated with the OHC measurements prior to ARGO and bearing in mind the power of the central limit theorem,”
The central limit theory is totally misused for climate measurements. Taking a thousand different measurements from a thousand different measurement devices at a thousand different times simply invalidates all the rules that make the central limit theory work. It’s one of the reasons why the “global average temperature” is really a monumental joke.
If you took a thousand measurements from *ONE* measurement device all taken at the same time (or nearly so) *then* you could legitimately use the central limit theory to determine a more accurate measurement. But that is not what is being done with the climate.
I am puzzled to find that SST in the NH is much faster than in in the SH.
should be
I am puzzled to find that SST in the NH is rising much faster than in in the SH.
In Charles Rotter’s post, is the following sentence. ‘That model also includes El Nino and La Nina (ENSO) variability to capture year-to-year temperature changes.’
Willis Eschenbach, in a post a few months ago (sorry, I don’t have the reference readily available) showed that heat transfer from ENSO to the global atmosphere is the result of two different mechanisms, ‘convection’ and ‘advection’. In this case, convection is the transfer of heat through evaporation of water, from the ocean surface, and transported through out the atmosphere by convection. Advection on the other hand is the transport of the warm (or cold) ENSO waters around the oceans through ocean currents.
I doubt very much, that the ‘El Nino and La Nina (ENSO) variability’ referred to above includes the ‘advection’ component, which operates on a time scale that is an order of magnitude longer than the ‘convection’ mechanism for ENSO. However, as Eschenbach showed in his paper, the ‘Advection’ mechanism is as important as is the ‘convection’ mechanism.
If the ‘Advection’ mechanism were to be included in the analysis, I doubt that there would be much room left for the effect of green-house gasses.
Has there been a calculation of the rate of atmospheric CO2 due just to the increase in SST? Curious as to how much of the 2 ppm/ yr increase is explainable by SST rise vs emissions. If CO2 lags and not leads temperature perhaps the (unknown) root cause of SST increase is responsible (in part) for increasing CO2.
Interesting study below showing complexity of ocean contribution to the carbon cycle.
https://earthobservatory.nasa.gov/features/OceanCarbon
Click on my name to read my report that CO2 is not a factor in warming.
is this the same Dr Spencer whose 2013 chart on global warming which “disproved” warming had to be thrown out and replaced because it was so wrong?
hmm.
Tim,
The Central Limit Theorem does apply in this case. It all depends on how you define “the sample”.
The sample can be the individual measurements of OHC, which vary in both time and space. Each of these measurements has an error that gets progressively worse with time prior to the start of the use of ARGO buoys.
Equally, the sample could be defined as the total OHC (i.e. the sum of the individual measurements) at any given time. This is a repeated measurement of a quantity (n = sample size = 1) that has its own associated uncertainty. Granted, you can argue about whether or not this quantity has any useful meaning in the real world but climate scientists believe that it is representative of the state of the overall climate system.
https://www.khanacademy.org/math/ap-statistics/sampling-distribution-ap/sampling-distribution-mean/v/central-limit-theorem
Of course, the larger the sample size (n), the closer distribution points about the drifting long term means will approach that of a bell-curve. This means that we can use the upper and lower envelop of the observations to crudely gauge the drift of the 1st moment of d(OHC)/dt.
Nothing becomes true merely by asserting it to be so.
So, if this is true…prove it.
Nicholas,
Take a smoothly varying discrete function that is monotonically increasing or decreasing and place it in a spread-sheet. Add some random noise to the value of each of the individual points in the discrete function. Make sure that the noise level is large but not so large as to overpower the absolute magnitude of the increase or decrease in the function over the range involved. See if you can discern the form of the function by taking the mid-points between the maxima and the minima of the plotted data points.
Here are two functions that you might want to try. Can you make a crude guess as to the general form of the function that is hidden in the noise?
When perihelion nearly coincides with the NH solstice AND earth’s obliquity is closer to a minimum – then OHC will begin to decline below present day values.
Perihelion is presently very close to being coincident with the SH solstice.
Ocean surface area is much more extensive in the SH.
Oceans buffer solar energy – land not so much.
Minimum obliquity results in less solar energy for NH & SH winter ice at high latitudes to diminish.
The ice will entually be back notwithstanding CO2 levels.
Do Cheng et al. try to explain why the slope of their heat content graph (rate of ocean warming) seems to increase around the year 1995? Does that coincide with an increase in the number of Argos buoys, or have earlier data been “adjusted”?
It’s also strange that Cheng et al’s data show ocean warming (although at a slower rate than recently) from 1940 to 1965, when air temperatures were cooling. If air temperatures were cooling during that period, wouldn’t that generate a stronger temperature gradient between the ocean and atmosphere, with increased heat transfer from the ocean to the air, tending to cool the ocean?
“I calculate from their data that 2019 was only 0.004 0.009 deg. C warmer than 2018.”
Seems extraordinarily unlikely that such a change could be outside the various sources of measurement/estimation error. Does anyone really claim to be able to measure the entire ocean with that kind of precision?
I’ve said this 1,000,000 times, but it is worth repeating. CO2 and LWIR between 13 and 18 microns won’t warm the oceans. The oceans control the global climate. A warming ocean is evidence CO2 isn’t causing global warming. A warming ocean is evidence that more visible radiation is reaching the oceans…which is happening. Simply look at the lower cloud layer data and you will see that fewer clouds result in warmer oceans, CO2 has nothing to do with it.