Guest Post by Willis Eschenbach
Yeah, I know Nature doesn’t have human emotions, give me a break. I’m aware it is unscientific and dare I call it atavistic and perhaps even socially unseemly to say Nature “hates” straight lines, but hey, it’s a headline, cut me some poetic slack.
My point is, everyone is aware that nature doesn’t deal in straight lines. Natural things move in fits and starts along complex paths, not straight from point to point. Phenomena have thresholds and edges, not slow linear changes at the perimeter. Tree branches and coastlines are jagged and bent. Things move in arcs and circles, relationships are complex and cyclical. Very little in nature is linear, particularly in complex systems.
Forcing is generally taken to mean downward radiation measured at the TOA (top of atmosphere). The IPCC says that when TOA forcing changes, the surface temperature changes linearly with that TOA forcing change. If there is twice the forcing change (twice the change in solar radiation, for example), the IPCC says we’ll see twice the temperature change. The proportionality constant (not a variable but a constant) that the IPCC says linearly relates temperature and TOA forcing is called the “climate sensitivity”.
Figure 1. Photo of impending change in climate sensitivity.
Today I stumbled across the IPCC justification of this linearity assumption. This is the basis of their claim of the existence of a constant called “climate sensitivity”. I quote it below.
I’ve removed the references and broken it into paragraphs it for easy reading. The references are in the original cited above. I reproduce all of the text on the web page. This is their entire justification for the linearity assumption. Having solved linearity in a few sentences, they then proceed to other matters. Here is their entire scientific justification for the assumption of linearity between forcing and temperature change (emphasis mine):
Linearity of the Forcing-Response Relationship
Reporting findings from several studies, the TAR [IPCC Third Assessment Report] concluded that responses to individual RFs [Radiative Forcings] could be linearly added to gauge the global mean response, but not necessarily the regional response.
Since then, studies with several equilibrium and/or transient integrations of several different GCMs [Global Climate Models] have found no evidence of any nonlinearity for changes in greenhouse gases and sulphate aerosol. Two of these studies also examined realistic changes in many other forcing agents without finding evidence of a nonlinear response.
In all four studies, even the regional changes typically added linearly. However, Meehl et al observed that neither precipitation changes nor all regional temperature changes were linearly additive. This linear relationship also breaks down for global mean temperatures when aerosol-cloud interactions beyond the cloud albedo RF are included in GCMs. Studies that include these effects modify clouds in their models, producing an additional radiative imbalance.
Rotstayn and Penner (2001) found that if these aerosol-cloud effects are accounted for as additional forcing terms, the inference of linearity can be restored. Studies also find nonlinearities for large negative RFs, where static stability changes in the upper troposphere affect the climate feedback (e.g., Hansen et al., 2005).
For the magnitude and range of realistic RFs discussed in this chapter, and excluding cloud-aerosol interaction effects, there is high confidence in a linear relationship between global mean RF [radiative forcing] and global mean surface temperature response.
Now, what strikes you as odd about that explanation of the scientific basis for their claim of linearity?
Before I discuss the oddity of that IPCC explanation, a short recap regarding climate sensitivity. I have held elsewhere that climate sensitivity changes with temperature. I will repeat the example I used to show how climate sensitivity goes down as temperature rises. This can be seen clearly in the tropics.
In the morning the tropical ocean and land is cool, and the skies are clear. As a result, the surface warms rapidly with increasing solar radiation. Climate sensitivity (which is the amount of temperature change for a given change in forcing) is high. High sensitivity, in other words, means that small changes in solar forcing make large changes in surface temperature.
By late morning, the surface has warmed significantly. As a result of the rising temperature, cumulus clouds start to form. They block some of the sun. After that, despite increasing solar forcing, the surface does not warm as fast as before. In other words, climate sensitivity is lower.
In the afternoon, with continued surface warming, thunderstorms start to form. These bring cool air and cool rain from aloft, and move warm air from the surface aloft. They cool the surface in those and a number of other ways. Since thunderstorms are generated in response to rising temperatures, further temperature increases are quickly countered by increasing numbers of thunderstorms. This brings climate sensitivity near to zero.
Finally, thunderstorms have a unique ability. They can drive the surface temperature underneath them below the temperature at which the thunderstorm formed. In this case, we have local areas of negative climate sensitivity – the solar forcing can be increasing while the surface is cooling.
As you can see, in the real world the temperature cannot be calculated as some mythical constant “climate sensitivity” times the forcing change. Sensitivity goes down as temperature goes up in the tropics, the area where the majority of solar energy enters our climate system.
So the IPCC claim of linearity, of the imagined slavish response of surface temperature to a given change in TOA forcing, goes against our daily experience.
Let me now return to the question I posed earlier. I asked above what struck you as odd about the IPCC explanation of their claim of linearity regarding forcing and temperature. It’s not the fact that they think it is linear and I disagree. That is not noteworthy.
Here’s what made me stand back and genuflect in awe of their claims. Perhaps I missed it, but I didn’t see a single word about real world observations in that entire (and most important) justification for one of their core positions.
I didn’t see anyone referenced who said something like ‘We measured solar radiation and downwelling longwave radiation and temperature at this location, and guess what? Temperatures changed linearly with the changes in radiation.’ I didn’t see anything at all like that, you know, actual scientific observations that support linearity.
Instead, their claim seems to rest on the studies showing that scientists looked at four different climate models, and in each and every one of the models the temperature change was linearly related to forcing changes. And in addition, another model found the same thing, so the issue is settled to a “high confidence” …
I gotta confess, that wasn’t the first time I’ve walked away from the IPCC Report shaking my head, but that one deserves some kind of prize or award for sheer audacity of their logic. Not a prize for the fact that they think the relationship is linear when Nature nature hates straight lines, that’s understandable, it’s the IPCC after all.
It is the logic of their argument that left me stammering.
Of course the model results are linear. The models are linear. They don’t contain non-linear mechanisms. And of course, if you look at the results of linear models, you will conclude with “high confidence” that there is a linear relationship between forcing and temperature. They looked into five of them, and case closed.
I mean, you really gotta admire these guys. They are so far into their models that they actually are using the linearity of the model results to justify the assumption of linearity embodied in those same models … breathtaking.
I mean, I approve of people pulling themselves up by their own bootstraps, but that was too twisted for me. The circularity of their logic made my neck ache. I kept looking over my shoulder to see if the other end of their syllogism was circling behind to strike me again. That’s why I genuflected in awe. I was overcome by the sheer beauty of using a circular argument to claim that Nature moves in straight lines … those guys are artists.
Meanwhile, back in the real world, almost no such linear relationships exist. Nature constantly runs at the edge of turbulence, with no linearity in sight. As my example above shows, the climate sensitivity changes with the temperature.
And even that change in tropical climate sensitivity with temperature is not linear. It has two distinct thresholds. One is at the temperature where the cumulus start to form. The other is at the slightly higher temperature where the thunderstorms start to form. At each of these thresholds there is an abrupt change in the climate sensitivity. It is nowhere near linear.
Like other natural flow systems, the climate is constantly restructuring to run “as fast as it can.” In other words, it runs at the edge of turbulence, “up against the stops” for any given combination of conditions. In the case of the tropics, the “stops” that prevents overheating is the rapid proliferation of thunderstorms. These form rapidly in response to only a slight temperature rise above the temperature threshold where the first thunderstorm forms. Above that threshold, most of any increase in the incoming energy is being evaporated and used to pump massive amounts of warm air through protected tubes to the upper troposphere, cooling the surface. Above the thunderstorm threshold temperature, little additional radiation energy goes into warming the surface. It goes into evaporation and vertical movement. This means that the climate sensitivity is near zero.
Now it is tempting to argue that the IPCC linearity claim is true because it only applies to a global average temperature. The IPCC only formally say that there is “a linear relationship between global mean RF [radiative forcing] and global mean surface temperature response.” So it might be argued that the relationship is linear for the global average situation.
But the average of non-linear data is almost always non-linear. Since daily forcing and temperature vary non-linearly, there is no reason to think that real-world global averages vary linearly. The real-world global average is an average of days during which climate sensitivity varies with temperature. And the average of such temperature-sensitive records is perforce temperature sensitive itself. No way around it.
The IPCC argument, that temperature is linearly related to forcing, is at the heart of their claims and their models. I have shown elsewhere that in other complex systems, such an assumed linearity of forcing and response does not exist.
Given the centrality of the claim to their results and to the very models themselves, I think that something more than ‘we found linearity in every model we examined” is necessary to substantiate this most important claim of linearity. And given the general lack of linearity in complex natural systems, I would say that their claim of linearity is an extraordinary claim that requires extraordinary evidence.
At a minimum, I think we can say with “high confidence” that it is a claim that requires something more weighty than ‘the models told me so’ ...
In your tropical example isn’t it the radiative forcing thats changing not the climate sensitivity. Clouds and the like are blocking solar radiation to the surface?
Brilliant yet again Willis.
This phrase struck me from their justification: ‘…………. have found no evidence of any nonlinearity …………’ Now that double negative is spin in action; it is not the same as saying: ‘………… have found evidence of linearity ……….’. Just thought I’d mention it.
Here’s the doom-monger argument in two sentences. It’s obvious that the sun warms the earth and there’s a nonlinear relationship of sensitivity to solar input over the course of day (same over seasons in temperate areas). But the linear relationship is from “forcing” which we define as the delta in radiative power (solar, CO2 or other) from some baseline to the new regime with greater “forcing”.
They justify this crap (mostly the ridiculous definition of “forcing”) using models which contain constant parameterized responses (e.g. parameterized convection) which produce linear responses. It also requires the mind boggling assumption that the world and its weather is somehow affected by this hypothetical world-wide average delta. Ask yourself if the tropical daytime nonlinear response described by Willis is affected by it being 0.5 degrees warmer in that location (unlikely since it is dwarfed by the daily forcing change) or 0.5 degrees warmer elsewhere (impossible).
Someone comes up with an idea for a computer program, what it should do and what the input and output should look like.
The computer programmers create a program that creates the required output.
The computer operator runs the program and generates the required output.
The climate scientist writes a paper about the output from the computer program.
The scientist’s employer writes a press release about the scientist’s paper.
The media publishes the press release as a news article.
The people read the article and think it is based on actual data.
It is not based on actual observations in nature; it is the output of a computer program.
The climate scientist is just a ghost writer for the computer programmer.
Perhaps the media should just talk directly to the computer programmers, since they are the ones that actually know why and how the program generated the press release.
No one even needs to go outside or actually measure anything.
HR: clouds are sensitivity.
Are you comparing apples to oranges?
Is it possible the models are working on much longer time scales than say hourly at a location?
Perhaps you are both right (Wills and ‘the models’) Willis is correct on very short time scales? The models are correct on very long time scales?
I don’t like the assumption of linearity. But I don’t find the reasoning in this post very persuasive. I’m not arguing for climate linearity here; I’m arguing that sceptics need to present a well-reasoned position and that this post isn’t it. In two parts:
There are a lot of scientific laws that claim linearity, where if you look closer things don’t behave anything like what the law describes. Ohm’s law says V=IR; if you look closely, you find that electrons are jumping all over the place. Some will go backwards, some will go sideways, the majority will go in the right direction but with a wide spread of velocities. But when you zoom out a bit, the average law works quite well. Same with the kinetic theory of gasses; PV=NkT at a wide scale, but at a smaller scale you can see that actually atoms are crashing around in entirely random ways and locally concepts like “pressure,” “volume” and “temperature” don’t make a lot of sense. So you can’t point to local weather effects and claim that they disprove the aggregate average theory of linear climate sensitivity; the randomness of small-scale weather processes doesn’t disprove the overall theory of linearity any more than observing individual electron tunnelling in silicon disproves Ohm’s law.
Secondly, whether proving linearity from a model is justified depends on the details of the models and whether they do actually assume linear sensitivity. You would need to follow all those references you deleted to find out. Perhaps the model is based on some low-level empirically-derived theory that allows you to infer whether the aggregate behaviour is linear or not; we can’t tell from the text you cite. Perhaps they actually model all the thunderstorm creation processes in the tropics, and snowfall at the poles, based on observed data, and find that when you add it all up it comes out linear. The kinetic theory of gasses is again a great example, though in reverse; the aggregate gas law was developed empirically and the model of an ideal gas was developed by reasoning about what underlying processes might produce that aggregate behaviour. There was no direct evidence that gasses were made up of tiny particles when the model was developed; that didn’t make it wrong. The model agreed with the empirically-derived law, and was successful in explaining it. It turned out to in fact be pretty much right. If you followed your reasoning in the mid-19th century, you would decry the KTOG as fairy-tales that no sane man would believe and you would later look very foolish.
“I mean, you really gotta admire these guys. They are so far into their models that they actually are using the linearity of the model results to justify the assumption of linearity embodied in those same models … breathtaking.”
Straightline thinkers like Gavin are never going to see the circularity of their own arguments.
Here’s a classic control system with feedback.
1. Increased solar radiation heats area of planet, causing increased evaporation of ocean, lake, river, soil, whatever is down there.
2. Increased evaporation results in higher percentage of H2O in higher pressure lower layer of atmosphere. H2O is LIGHTER than the mixture of N2/O2, and tends to “want” to rise, but is prevented from doing so by boundary layer.
3. Eventually, enough H2O pools that it starts pushing through boundary layer and rising. As it cools by entering lower pressure areas, some of it condenses into droplets or crystals which cause visible clouds. These clouds block incoming sunlight, reducing incoming energy.
4. At this point, the system can achieve a level of equilibrium, where clouds control incoming radiation and manage to hold the temperature at a level appropriate for an area’s current pressure.
5. In some cases the amount of H2O is so great in a concentrated area that it bursts through the boundary layer and large amounts from the area go with it, creating massive upwellings that can exceed 100mph and carry millions of tons of moisture to areas of low enough pressure that they are in the -40C range. Welcome to a thunderstorm, or hailstorm. These types of storms haul HUGE amounts of heat high enough in the atmosphere that excess energy can radiate directly into space, eventually dragging supercooled H2O down to ground level in the form of rain or hail.
6. No matter how much cooling or heating happens during the day, then there is night. At night, all energy radiation is exactly one direction: outward. At any given moment, slightly more than 1/2 of the planet is experiencing night. At any given moment as much energy is radiated outward from the back of the planet as is being absorbed in the front.
7. Thus, it is physically impossible to achieve dramatic temperature changes from minor trace gas changes, since equilibrium is based on total mass of atmosphere and its major gas balance. Now, if the percentage of O2 and N2 were to significantly change, so would the equilibrium temperature of the planet.
It’s those dam models again!
So linear computer models find no evidence of non-linearity? Makes you want to weep!
The “Sunday Times” (normally a respected newspaper in the UK) yesterday had an article “2010 the warmest year on record”, but now buried on page 15 rather that the normal spot on the front page. Perhaps the MSM are now taking the hint, along with the Royal Society and all the other clowns who were taken in by this rubbish. But I notice that last week’s draconian (and necessary) spending cuts by the UK government still left billions for wind farms and carbon capture demonstration plants. Why are governments the last ones to “catch on”?
“Rotstayn and Penner (2001) found that if these aerosol-cloud effects are accounted for as additional forcing terms, the inference of linearity can be restored.”
Sounds like a fudge factor to me. Take dust, an aerosol. If the concentration dust is greater from ground to, lets say, 100 meter then dust would have a warming effect on surface temperature. But if the concentration of dust was greater form 100 meter to 200 meters than 0 to 100 meters, then that would have a cooling effect on ground temperature would it not? After all, dust is a solid and is absorbing short wave radiation.
As an aside, Connolly may be gone from Wiki, but his spirit lives on!
http://motls.blogspot.com/2010/10/wikipedia-william-connolley-was-just.html
An edit by Lubos of Briffa’s page and it’s quickly erased and threats of sanctions are made!
It’s quite likely that the IPCC is wrong and the forcing is not linear. But that’s not the most important point.
What really matters is the IPCC’s almost obsessive dependence on computer programs (calling them GCM’s sounds much grander, but basically they’re just computer programs written by people who are desperate to prove AGW).
There was a recent example from NASA, in which they use a computer program to ‘prove’ that CO2 is the driver of climate.
In reality, they create computer programs which have a series of assumptions that fit their own beliefs. They then examine the outputs of the computer programs to see if theirt assumptions are correct. And – who would have guessed it! – the computer programs do indeed confirm that those assumptions were correct.
To any outside observer, this process is so circular as to almost defy belief. It’s not science. It’s propaganda.
So why do they do it? I think the answer is pretty obvious. Their glorified Playstation programs are brilliant at forecasting climate that has already happened. But any fool can forecast what has already happened. But, because they have little connection with the real world, they are useless at predicting future climate. Therefore, studies based on real observations of the real world will tend to contradict the Playstation models.
It’s much safer to check their assumptions against the Playstation programs. By doing this they can be sure they will get the right answer every time.
I’d better stop now as I’m beginning to feel a little sick.
Chris
An interesting post Willis. I guess when you live in the virtual world of the IPCC conflicting observational evidence is of little importance.
Does this mean its worse than we thought?
Jerry says:
October 25, 2010 at 2:45 am
Nature doesn’t work like that. You can look at the fluid motion of milk in a cup of tea and the motion of a massive gas cloud nebula in space and describe both using the same basic rules, yet the cup of tea is an event measured in seconds and the motion of that nebula is measured in millions of years. Possibly an extreme example and I’m certain someone will find a “yes, but” that they think counters the argument but bioth are governed by the same rules of fluid motion and will both behave in the same chaotic and apparently random way as a result, just over different time scales.
We live in a fractal universe. The same rules apply at the top and the bottom, the only difference appears to be the amount of time needed to observe them. Things don’t suddenly become linear after some ill-defined threshold of convenience.
Playing devil’s advocate here, I think you’re trying to refute a climate-theoretic concept, which plays out over decades, with weather-theoretic concepts spanning a few hours. Climate sensitivity (‘lambda’) is also one of those ‘first-order’ approximations, which no one expects to hold exactly at all geospatial scales.
delta-Temp = lambda x delta-Forcing
You’re correct, Willis, the relationship between temperature and forcing is not exactly linear. But all _observed_ processes tend to be non-linear because of process and measurement ‘noise’.
But having said that, most non-linear processes can be approximated by integrating enough tiny linear steps.
Also, the issue here is CO2 forcing, not solar forcing, no one will dispute the general idea that the Sun warms the Earth. But if you believe that there is a positive contribution from CO2 to the net imbalance between incoming and outgoing irradiation (‘radiative forcing’), then lambda merely expresses a first order approximation, over climatic epochs, of the temperature change that would be ‘observed’, at the climate level, due to that forcing.
I don’t see where proving this relationship doesn’t exactly hold on a hourly basis necessarily proves or disproves any climate theory. And, yes, I’m a CAGW skeptic.
While this issue is indeed a major issue , one has to be careful with the arguments .
I have to mention that any parameter Y related to a variable X will answer in a linear way (e.g Y=a.X) provided that the variation of X is small .
This is a trivial consequence of the Taylor expansion .
Sure it is necessary that the relationship between Y and X be sufficiently smooth but this is just a technical detail .
So if one considers a very small variation of a radiative flux (a couple of W/m²) compared to the overal flux (hundreds of W/m²) , then any variable , not just temperature will answer in a linear way .
Especially if you choose a variable like global temperature that has been so averaged that it almost stays constant whatever happens .
Let’s imagine that there exists an extremely highly non linear relationship F such as the Global average of temperature Tg :
Tg = F(Global radiative flux , albedo , sun activity , wind speed , polar ice surface , cloudiness , humidity , O3 concentration , microalgae mass and time) .
Now if you consider a small variation of the variable “Radiative flux” during a short time , all other variables being considered constant , the Tg response will be almost exactly linear .
One can even give easily the value of the proportionality constant (e.g climate “sensitivity”) – it is the partial derivative of F with regard to the Global radiative flux .
Of course there are at least 2 problems .
One is lethal – you don’t know F in the real world and it is not available to experimentation . So you have no clue and will never have about what the partial derivative of F are .
Second is almost lethal too . The linearity holds only for a short time as long as the system didn’t move far from the point where it was before .
As soon as any or all variables acting on F can no more be considered constant , the linearity no longer holds . In other words you can make predictions only for a very short time .
Of course in your example with thuderstorms there are no “small” variations , no “equilibriums” and no “everything else being constant” .
That’s why the Taylor development of your G (note that your G is NOT the same as the above F !) is useless and you are exposed to the full non linearity .
However this local result does not necessarily prove something about the averages .
In other words the uselesness of a Taylor development of G (e.g your thunderstorms) does not necessarily imply the uselesness of a Taylor development of F (e.g global averages) .
You will have noticed that I said “not necessarily” .
Actually the implication is often correct but the point is that it is not automatical , one has to go a bit more in depth to prove it .
Giving just the example of G is not enough .
Tom says:
October 25, 2010 at 2:49 am
——-
Wonderful, physics acknowledges linear relationships when they are found. You reference V=IR. How about VT=D? There are plenty more. Your mention of dancing electrons is ridiculous in this context. At the quantum level, nothing is certain.
The point you are missing is that in every case of a settled physical relationship there is direct measurement and observation that supports it. Here, in the case of the behavior of climate, there is no data to support these claims, nothing but models that perform as they were written to perform. How much money was wasted paying for people to write computer programs to demonstrate a preconceived result without supporting data??
The sad fact here is that there are individuals so invested in this shell game that they write supposing intellectual arguments to obfuscate the obvious and attempt to continue the charade. Wake up.
Willis — you made your case that nature is not linear. IMO IPCC claim for linearity is juvenile. But having “proved” that, what is missing from this post is addressing the obvious SO WHAT? Without that it just looks like a hit and run.
Straightline thinkers using circular logic. How much more convoluted can that be? (Maybe they’re completely dedicated disciples of http://www.art.com/gallery/id–a51898/m-escher-posters.htm?RFID=274465&CTID=999901394
M. C. Escher?)
Straightline thinkers using circular logic. How much more convoluted can that be? (Maybe they’re completely dedicated disciples of
M. C. Escher?)
When talking to a member of the Wet Office back Feb/March, when he drew his little famous Wet Office graph of the slow rise in temps, followed by a slight cooling, then a warming up to the end of the 20th Century, he explained all about how they programme these models ot obey the laws of physics, etc. I then challlenged him on the validity of the models & their output he responded by asking “how did we get that result to match surface temperatures?” He couldn’t see the logic of his argument that a graph had been produced by computer/hand in earlier attempts, but now models replicated it therefore the models must be right! I reminded him that somebody has to tell the computer to show warming for given inputs of CO2, it cannot do it all by itself, this seemed to leave him dumb-founded & he couldn’t counter my point! They are all so up their own models as Wllis says that they cannot contemplate that they could be wrong, it’s inconceivable to them it would appear. All I did extract from him the admission that there are huge uncertainties, & that these are played down, although denied the Wet Office did this but I suggested he was not looking at the articels in the msm etc, where no doubt exists, & Dr Vicky Pope et al never mentions the word “uncertainty”, well not in public anyway!
“Why are governments the last ones to “catch on”?”
An easy question to answer.
The world warms and they did nothing despite being warned: they take the blame.
The world warms and they wasted billions because they were warned: they blame the scientists.