Guest Post by Willis Eschenbach
This is an extension of the ideas I laid out as the Thunderstorm Thermostat Hypothesis on WUWT. For those who have not read it, I’ll wait here while you go there and read it … (dum de dum de dum) … (makes himself a cup of coffee) … OK, welcome back. Onwards.
The hypothesis in that paper is that clouds and thunderstorms, particularly in the tropics, control the earth’s temperature. In that paper, I showed that a falsifiable prediction of greater increase in clouds in the Eastern Pacific was supported by the satellite data. I got to thinking a couple of days ago about what other kinds of falsifiable predictions would flow from that hypothesis. I realized that one thing that should be true if my hypothesis were correct is that the climate sensitivity should be very low in the tropics.
I also figured out how I could calculate that sensitivity, by using the change in incoming solar energy (insolation) between summer and winter. The daily average top of atmosphere (TOA) insolation is shown in Figure 1.
Figure 1. Daily TOA insolation by latitude and day of the year. Phi (Φ) is the Latitude, and theta (Θ) is the day of the year expressed as an angle from zero to 360. Insolation is expressed in watts per square metre. SOURCE.
(As a side note, one thing that is not generally recognized is that the poles during summer get the highest daily average insolation of anywhere on earth. This is because, although they don’t get a lot of insolation even during the summer, they are getting it for 24 hours a day. This makes their daily average insolation much higher than other areas. But I digress …)
Now, the “climate sensitivity” is the relationship between an increase in what is called the “forcing” (the energy that heats the earth, in watts per square metre of earth surface) and the temperature of the earth in degrees Celsius. This is generally expressed as the amount of heating that would result from the forcing increase due to a doubling of CO2. A doubling of CO2 is estimated by the IPCC to increase the TOA forcing by 3.7 watts per metre squared (W/m2). The IPCC claims that the climate sensitivity is on the order of 3°C per doubling of CO2, with an error band from 2°C to 4.5°C.
My insight was that I could compare the winter insolation with the summer insolation. From that I could calculate how much the solar forcing increased from winter to summer. Then I could compare that with the change in temperature from winter to summer, and that would give me the climate sensitivity for each latitude band.
My new falsifiable predictions from my Thunderstorm Thermostat Hypothesis were as follows:
1 The climate sensitivity would be less near the equator than near the poles. This is because the almost-daily afternoon emergence of cumulus and thunderstorms is primarily a tropical phenomenon (although it also occurs in some temperate regions).
2 The sensitivity would be less in latitude bands which are mostly ocean. This is for three reasons. The first is because the ocean warms more slowly than the land, so a change in forcing will heat the land more. The second reason is that the presence of water reduces the effect of increasing forcing, due to energy going into evaporation rather than temperature change. Finally, where there is surface water more clouds and thunderstorms can form more easily.
3 Due to the temperature damping effect of the thunderstorms as explained in my Thunderstorm Thermostat Hypothesis, as well as the increase in cloud albedo from increasing temperatures, the climate sensitivity would be much, much lower than the canonical IPCC climate sensitivity of 3°C from a doubling of CO2.
4 Given the stability of the earth’s climate, the sensitivity would be quite small, with a global average not far from zero.
So those were my predictions. Figure 2 shows my results:
Figure 2. Climate sensitivity by latitude, in 20° bands. Blue bars show the sensitivity in each band. Yellow lines show the standard error in the measurement.
Note that all of my predictions based on my hypothesis have been confirmed. The sensitivity is greatest at the poles. The areas with the most ocean have lower sensitivity than the areas with lots of land. The sensitivity is much smaller than the IPCC value. And finally, the global average is not far from zero.
DISCUSSION
While my results are far below the canonical IPCC values, they are not without precedent in the scientific literature. In CO2-induced global warming: a skeptic’s view of potential climate change, Sherwood Idso gives the results of eight “natural experiments”. These are measurements of changes in temperature and corresponding forcing in various areas of the earth’s surface. The results of his experiments was a sensitivity of 0.3°C per doubling. This is still larger than my result of 0.05°C per doubling, but is much smaller than the IPCC results.
Kerr et al. argued that Idso’s results were incorrect because they failed to allow for the time that it takes the ocean to warm, viz:
A major failing, they say, is the omission of the ocean from Idso’s natural experiments, as he calls them. Those experiments extend over only a few months, while the surface layer of the ocean requires 6 to 8 years to respond significantly to a change in radiation.
I have always found this argument to be specious, for several reasons:
1 The only part of the ocean that is interacting with the atmosphere is the surface skin layer. The temperature of the lower layers is immaterial, as the evaporation, conduction and radiation from the ocean to the atmosphere are solely dependent on the skin layer.
2 The skin layer of the ocean, as well as the top ten metres or so of the ocean, responds quite quickly to increased forcing. It is much warmer in the summer than in the winter. More significantly, it is much warmer in the day than in the night, and in the afternoon than in the morning. It can heat and cool quite rapidly.
3 Heat does not mix downwards in the ocean very well. Warmer water rises to the surface, and cooler water sinks into the depths until it reaches a layer of equal temperature. As a result, waiting a while will not increase the warmth in the lower levels by much.
As a result, I would say that the difference between a year-long experiment such as the one I have done, and a six-year experiment, would be small. Perhaps it might as much as double my climate sensitivity values for the areas that are mostly ocean, or even triple them … but that makes no difference. Even tripled, the average global climate sensitivity would still be only on the order of 0.15°C per CO2 doubling, which is very, very small.
So, those are my results. I hold that they are derivable from my hypothesis that clouds and thunderstorms keep the earth’s temperature within a very narrow level. And I say that these results strongly support my hypothesis. Clouds, thunderstorms, and likely other as-yet unrecognized mechanisms hold the climate sensitivity to a value very near zero. And a corollary of that is that a doubling of CO2 would make a change in global temperature that is so small as to be unmeasurable.
In the Northern Hemisphere, for example, the hemispheric average temperature change winter to summer is about 5°C. This five degree change in temperature results from a winter to summer forcing change of no less than 155 watts/metre squared … and we’re supposed to worry about a forcing change of 3.7 W/m2 from a doubling of CO2???
The Southern Hemisphere shows the IPCC claim to be even more ridiculous. There, a winter to summer change in forcing of 182 W/m2 leads to a 2°C change in temperature … and we’re supposed to believe that a 3.7 W/m2 change in forcing will cause a 3° change in temperature? Even if my results were off by a factor of three, that’s still a cruel joke.
Steve Goddard (12:12:38)
I repeat that I am dealing with average temperatures, not peak temperatures. If the desert sun stayed at the average intensity of the whole dawn to dusk period for 30 hours, the average dawn to dusk temperature would get slightly warmer, but not much.
Ron Broberg (13:39:20)
I don’t understand. What are you calling “RF”? Radiative forcing? If so, I don’t understand the question. The solar constant is ~ 1,366 W/m2 at TOA. The difference if I were to assume 1,360 or 1,370 W/m2 is trivial in this analysis. What’s the point here?
Willis Eschenbach;
I agree entirely, with one caveat – the “sweet spot” (usually called the “atmospheric window”) only exists when the sky is clear. This is because clouds are essentially black bodies for all longwave frequencies. Since cloud cover on the earth is on the order of 60%, this is far from a trivial factor.>>
So everything I know about clouds is in your last sentence… but might I hazard some guesses?
1. I expect that clouds predominate in warmer climes, less so in cool ones (ie below +15)?
2. At only 40% clear sky, the atmospheric window would still be capable of limiting temperature increases.
3. Clouds don’t stand still. Any warming they cause at surface would be like compressing a coiled spring… provided the surface temp doesn’t go over that +15 top end (not likely in the arctic regions) and would uncoil (P=CT^4) when the cloud cover moves off. So the 40% window would move around over time. Lag yes, but the heat still escapes?
4. What temperature is the cloud at? I assume that if LW from earth surface at (for sake of argument) 25 degrees C hits a cloud on the way up and gets absorbed, then what temperature does the cloud re- emit at? Is it in the atmospheric window? If yes, then some gets emitted down and some up (photons have a remarkable lack of sense of direction) and the cloud is now leveraging the atmospheric window?
5. Anecdotal thought here… I’ve seen way too many 40 below mornings for my liking, or 30 below for that matter. I don’t remember any clouds those days. Clear blue sky in January in Winnipeg = did anyone plug the block heater in last night? No? Back to bed the car won’t start anyway.
6. On the cooling side of the cycle… clouds would be less of a factor. As surface temperatures drop below the atmospheric window, water vapour would start absorbing again and re-emitting before clouds could get to it. Presuming the clouds have a net warming effect, this would augment the thermostat limiting temperature decreases. If they have a net cooling effect, the GHG would still be able to limit the temperature decrease, it would just take longer because they have to fight off the clouds 60% of the time. (Unless 1. Above is correct and then there wouldn’t be much cloud in the areas acting as the thermostat)?
7. This ought to be fairly easy to test. If one had a long term surface temperature set broken down by latitude (that you could trust), and an outgoing LW dataset at TOA broken down by latitude (that you could trust), I would think that you could correlate efficiency of LW emission to space by temperature band, clouds or no clouds.
Eschenbach: I don’t understand. What are you calling “RF”? Radiative forcing? If so, I don’t understand the question. The solar constant is ~ 1,366 W/m2 at TOA. The difference if I were to assume 1,360 or 1,370 W/m2 is trivial in this analysis.
Thanks. Yes, RF = radiative forcing
Parameterized solar constant at 1,366 W/m^2
What’s the point?
Just trying to understand your methodology. Of course, modeling the solar constant with a fixed parameter is justified here. Using measured data would only adjust it by one part in several hundred – and then only if you were measuring over a span of years.
Willis, you wrote:
The problem with your example is that unlike with the climate, energy is added during both the positive and negative swings of the current. In addition, because of the very high frequency of the cycles, there is almost no temperature change in the filament with time. Neither of these are true in the current situation.
If you look at power rather than current you will see that there is a sinusoidal input just like the insolation in your example. I should have been more careful in my description but I thought it would be obvious that I would comparing watts in with watts out.
The frequency in my example is high but the load is very much smaller. I am not saying that this makes it an exact analogy but I am trying to allert you to a potential flaw in your logic. This is the nub of the problem. I believe that the relatively small variation in average global temperatures is due to the thermal inertia of the earths climate system and not to low sensitivity. Others have made the same point in different ways. I could be wrong but it needs a more detailed analysis to prove it one way or another.
cal (08:03:47) :
>In the case of a light bulb one sees little fluctuation in temperature yet the input energy is varying by 100% over the cycle.
This is of course true but the fluctuation in heat and therefore the light output will lag the 100Hz/120Hz input cycle by close to 90 degrees. What that means is you’re far into the stopband of a first order lowpass filter.
If temperatures were lagging by three months then I wouldn’t believe the results were accurate. Two months, I’m fine. one month and I’m happy. Four months is nightmare time. Because now you’re past 90 degrees and therefore into a second order system.
Alexander Harvey (13:20:29) : edit
Alex, thanks much for your detailed analysis. I didn’t understand the bottom line, however. Are you saying that we can’t estimate the answer? Or are you saying that we can estimate the answer, but you haven’t done so? If the latter, could you give us an estimate of the answer?
Many thanks for your work in fighting my ignorance, and I mean that seriously, not as a snark of any kind, this is science at its best,
w.
Rienk (15:40:52)
The lag is 4-6 weeks.
w.
cal says:
How does the much larger heat capacity of the oceans vs the atmosphere affect any claims that increasing CO2 is going fry us all?
Masive amounts of GCM software exists to solve these problems. You can’t make a back of the envelope qualitative calculation of climate sensitivity.
Willis Eschenbach would you please tell me what is the physical quantity “forcing” and how is it measured. I have some idea about the others but “forcing” really puzles me. And if you have time would you tell me what is “down welling” or “up welling” radiation.
Bill Illis (08:47:27)
Bill, what is the source of that information?
w.
Steve Goddard (16:12:49) : edit
If you believe that the GCMs are anything more than tinkertoy models, you haven’t looked at the problems. These include a near-total lack of discussion by the modelers of the following issues:
1. Propagation of errors in the models. In any iterative model, this is a very important issue.
2. A listing of all of the “parameters” of each model, with justification for their values and an estimate of what an error in that parameter would lead to in the final result.
3. The effect of the simulation of viscosity dissipation in the models.
4. The approximations of the Navier-Stokes equations, and any discussion of whether they converge.
5. Verification and Validation, which is a part of software engineering that is routinely done on all mission-critical software.
6. Software Quality Assurance, which again is a crucial part of software engineering for any important software. These last two are routinely used on software for everything important from elevators to jet airplanes to space missions. In WG1 they don’t even rate a mention.
All of these issues have been raised, time after time, by people concerned about the models. Yet they are totally untouched in the IPCC WG1. Shows how serious the modelers are about actually seeing if their models are worth more than a bucket of warm spit …
George Steiner (17:48:01) : edit
George, good questions. Climate science, like any other specialized field, has its jargon.
A “forcing” is anything that changes the amount of energy absorbed or emitted by a part of the climate system. Solar insolation is a forcing, for example, as is greenhouse radiation. People also speak of e.g. “cloud forcing”, meaning the change in radiation due to clouds.
Downwelling and upwelling refer to the direction of radiation with respect to the surface. Sunshine on the surface is downwelling, longwave (infrared) radiation from the earth to space is upwelling. Longwave radiation is often referred to as DLR or ULR for down (up) welling longwave radiation. The atmosphere and clouds produce both DLR and ULR.
The only stupid question is the one you don’t ask …
Thanks, Willis, for an elegant back-of-the-envelope analysis.
Willis,
I’m sure you know that I share the same criticisms of climate models as what you listed. I’ve made most of those complaints here myself.
Nevertheless, the mathematical approach they take is more or less the correct one for the difficult problem they are attempting to solve. Will they ever be any value? I doubt it. But I don’t think any other approach will do much better.
Climate is inherently unpredictable for the same reasons as weather more than about three days out.
Gary Hladik (18:58:48) : edit
Thanks for the vote, Gary. However, in my mind the jury is still out as to whether my analysis is valid or not. I think that the admittance issue does not change my results significantly, but I’ve been wrong before, and Alex makes a good case. I’m still waiting for a numeric result from his analysis.
Which is what I love about WUWT. Rather that posing as a site like RC that claims that the science is settled and they’re going to spoon feed it to us, this site is the epitome of real science. I make claims, and people see if they can shoot them down.
Onwards …
Willis Eschenbach (14:04:40) :
With regard to the transport of heat between hemispheres, the primary north/south transport mechanism in the tropic is the Hadley circulation, which is on average centered near the equator(http://en.wikipedia.org/wiki/Atmospheric_circulation). Note however that the equatorial convergence zone of the Hadley circulation is not really centered on the equator, but instead tracks the solar angle:
“Though the Hadley cell is described as lying on the equator, it is more accurate to describe it as following the sun’s zenith point, or what is termed the “thermal equator,” which undergoes a semiannual north-south migration.”
Which is essentially saying that the center of flux of heat poleward changes with season form ~23 degrees north in the northern summer to ~23 degrees south in the northern winter. This means that heat from one geographical hemisphere (the hemisphere in summer) is indeed transported in large quantity to the other geographical hemisphere (the hemisphere in winter) by the Hadley circulation, because the Hadley cell does not lie on the geographical equator, but rather follows the north-south season change in the position of the sun (or the point of highest solar input).
With regard to ocean heat accumulation/release with seasonal progression (especially outside the tropics), as you said, the ocean heat does indeed lag the solar thermal input through all seasons, making the average temperature warmer that it would otherwise have been through fall and winter, and cooler than it would otherwise have been through spring and summer. However, the seasonal change in ocean surface temperature is much smaller at each latitude over ocean than over land, which (I think) can only be explained as a strong influence of the uptake and release of heat by the ocean. Finally, major ocean currents (http://upload.wikimedia.org/wikipedia/commons/0/06/Corrientes-oceanicas.gif) carry a huge quantity of heat poleward year round, further warming high latitudes in the winter, especially in the northern hemisphere. Without the influence of the oceans, seasonal temperature swings at higher latitudes would be much higher than they are.
As I said earlier, I just don’t think you can generate an accurate estimate of overall global climate sensitivity if you ignore the contributions of atmospheric heat transfer from the tropics poleward, and ignore the substantial tempering effect of the ocean. The temperature in polar regions in winter (with essentially zero solar heat input) would fall to extremely low temperatures due to radiative heat loss (the blackbody temperature of space is ~3K) if they were thermally isolated from the rest of the globe. But they are not isolated; they receive substantial heat from lower latitudes throughout the winter, and also receive heat from gradual cooling of the ocean in winter.
Every graduate student should be required to present their research, stage by stage, on the internet. The feedback and learning value would be tremendous.
Willis: If I correctly understood your argument, you have calculated “climate sensitivity” by divided the mean summer/winter temperature differences (dTsw) by the mean difference in daily summer and winter insolation (dWsw) in watts/m2 and then multiplied by (3.6 W/m2)/2XC02. This derivation assumes that seasonal temperatures respond quickly to changes in solar insolation. This assumption is not true. The coldest winter temperatures (in the US) come at the end of January, not on the shortest day of the year on December 21 and the warmest temperatures often come in late July, not on the longest day of the year. Where I grew up in coastal California, the average temperature on June 21 is 1 degC colder than on September 23 because the nearby ocean takes so long to warm up.
If temperature were determined solely by solar insolation, the summer/winter temperature difference would be the same in the center and at the coasts of continents at a given latitude. If temperature were solely determined by insolation, the temperature of the spring and fall equinoxes would be identical. There is a 6 degC difference where I live, about half of the spread between the daily low (with no insolation) and daily low (with 2X average daily insolation.)
So one can easily see that it takes at least one month for surface air temperature dominated by land to respond to changes in insolation and longer for surface air temperatures dominated by water to respond. However, persistent circulation of air (both vertically and horizontally) takes place on time scales faster than monthly, so the same air is not actually responding to the changes in insolation that reach the earth.
You can find information about how the location of the thermocline varies with season at: http://www.villasmunta.it/oceanografia/the_three.htm See Figure 1-2-3. Seasonal temperature changes in the thermocline reach 100 meters.
Finally, we know that warm air holds more water-vapor, so the greenhouse effect is stronger in the winter. If one takes the IPCC’s estimate for water-vapor feedback of about 2 W/m2/degC, you’ve got a substantial summer vs. winter forcing. Then there is cloud feedback to worry about. If winter skies are dominated by low clouds and summer skies by higher clouds, radiative cooling from cloud tops will be much less efficient in the summer. Is albedo the same in summer and winter?
Willis Eschenbach (13:15:19) :
tallbloke (02:14:54)
Hi Willis,
I think this is a promising approach, but needs some further consideration on the absorption of energy by the ocean.
Always good to hear from you, tallbloke.
Willis, many thanks for your considered and detailed reply to my post. I hope you saw the answer to some of the issues you raised in the short addendum I added a couple of comments later.
I can see you have your hands full here with the many excellent and more immediately relevant replies you have recieved on this thread, so I will take my time over considering your full response. Perhaps one thing we might agree on is that your error bars might be tightened down by a decadal length study which removes some of the signal introduced by the changing solar-oceanic energy balance over the solar cycle length.
Thanks again.
0.05C/3.7W is 370 W for the 5 C increase since last glacial. Wonder where those 370 watts came from.
lgl (00:17:00) :
You say:
“0.05C/3.7W is 370 W for the 5 C increase since last glacial. Wonder where those 370 watts came from.”
The paleoclimate record strongly suggests that the climate system is bistable (i.e. stable in each of two states; viz. glacial and interglacial). If this apparent bistability is real then the energy accumulation you assert is an indication of a change from one state to the other.
The fine analysis conducted by Willis Eschenbach only concerns the climate sensitivity of the present interglacial state.
Therefore, your comment is not relevant to the present discussion unless you can provide an explanation for the apparent bistability of the climate.
Richard
Willis,
The Albedo data comes from “Thermal Environments” by James F. Clawson of the Jet Propulsion Laboratory. It is used to control spacecraft and satellites and I believe it uses ERBE data. I haven’t found an on-line copy of it but the data is provided as an extract in a few places. Here’s one.
http://www.tak2000.com/data/planets/earth.htm
When I run all the numbers using these figures, I get a global Albedo value of 0.2983 which is exactly the amount in Trenberth’s latest Earth Radiation Budget paper.
You can also download the climatologies from ERBE here (as usual, they do not make it easy). Use Total Albedo which incorporates the effects of clouds versus Clear-Sky or Surface Albedo which doesn’t.
http://iridl.ldeo.columbia.edu/SOURCES/.NASA/.ERBE/.Climatology/.total/albedo/?help+datatables
Steve Goddard,
Increased greenhouse gases make it more difficult for heat to escape, so temperature rises to keep equilibrium. Heat is not “trapped.” It finds it’s way out.
Exactly! That is global warming caused by greenhouse gases. The temperature rises in order restore the radiative balance.
My “saucepan” analogy was not meant to show that heat is permanently trapped, but that the temperature of the contents in the saucepan rise higher if there is a lid on it than if there is not (I am assuming the contents are not boiling of course!)
Willis,
If you have something scientific to add, please bring it on.
Okey dokey, how about Karl et al 2006?
Old stuff I know but as you will be aware the Greenhouse effect predicts that as more heat is trapped in the troposphere less reaches the stratosphere. The effect will be to cause a rise in temperature in the troposphere and fall in the stratosphere. This is exactly what the satellites and radiosondes are reporting. Look at the graphs on page 8 of the executive summary here:
http://www.climatescience.gov/Library/sap/sap1-1/finalreport/sap1-1-final-execsum.pdf
Also go to
http://discover.itsc.uah.edu/amsutemps/
and look at how the troposphere (ch5) temps are rising year on year while those in the stratosphere (ch12) at 31km up are cooling.
The observations from the satellites are showing exactly what the theory predicts. How that increase in troposheric temperatures will affect the weather and clouds is uncertain. We may see some negative feedbacks but apparently no major ones yet. Overall the satellite data is showing a slightly smaller trend rise in temperatures than the surface stations – but there is barely 20 years of data to work with. It takes a long time for a 0.2c – 0.3c a decade increase in temperatures to become apparent when the weather “noise” is tens of degrees and the year-to-year “noise” in average temperatures is 0.5c or so. Add in 30 and 60 year variations in NAO / PDO to add longer term “noise” and it becomes difficult to tease out a mere 3c in 100 years.
My argument with your piece was primarily the last two paragraphs – which make the common falacial comparison between short term variations in insolation and temperature between summer and winter (that I would call “weather”) and longer term trends in average temperatures (that I would call “climate”).