Arguments
Is the CO2 effect saturated?
What the science says...
| Select a level... |
Basic
|
Intermediate
|
Advanced
| ||||
|
The notion that the CO2 effect is 'saturated' is based on a misunderstanding of how the greenhouse effect works. |
|||||||
Climate Myth...
CO2 effect is saturated
"Each unit of CO2 you put into the atmosphere has less and less of a warming impact. Once the atmosphere reaches a saturation point, additional input of CO2 will not really have any major impact. It's like putting insulation in your attic. They give a recommended amount and after that you can stack the insulation up to the roof and it's going to have no impact." (Marc Morano, as quoted by Steve Eliot)
At-a-Glance
This myth relies on the use (or in fact misuse) of a particular word – 'saturated'. When someone comes in from a prolonged downpour, they may well exclaim that they are saturated. They cannot imagine being any wetter. That's casual usage, though.
In science, 'saturated' is a strictly-defined term. For example, in a saturated salt solution, no more salt will dissolve, period. But what's that got to do with heat transfer in Earth's atmosphere? Let's take a look.
Heat-trapping by CO2 in the atmosphere happens because it has the ability to absorb and pass on infra-red radiation – it is a 'greenhouse gas'. Infra-red is just one part of the electromagnetic spectrum, divided by physicists into a series of bands. From the low-frequency end of the spectrum upwards, the bands are as follows: radio waves, microwaves, infrared, visible light, ultraviolet, X-rays, and gamma rays. Gamma rays thus have a very high-frequency. They are the highest-energy form of radiation.
As our understanding of the electromagnetic spectrum developed, it was realised that the radiation consists of particles called 'photons', travelling in waves. The term was coined in 1926 by the celebrated physicist Gilbert Lewis (1875-1946). A photon's energy is related to its wavelength. The shorter the wavelength, the higher the energy, so that the very high-energy gamma-rays have the shortest wavelength of the lot.
Sunshine consists mostly of ultraviolet, visible light and infra-red photons. Objects warmed by the sun then re-emit energy photons at infra-red wavelengths. Like other greenhouse gases, CO2 has the ability to absorb infra-red photons. But CO2 is unlike a mop, which has to be wrung out regularly in order for it to continue working. CO2 molecules do not get filled up with infra-red photons. Not only do they emit their own infra-red photons, but also they are constantly colliding with neighbouring molecules in the air. The constant collisions are important. Every time they happen, energy is shared out between the colliding molecules.
Through those emissions and collisions, CO2 molecules constantly warm their surroundings. This goes on all the time and at all levels in the atmosphere. You cannot say, “CO2 is saturated because the surface-emitted IR is rapidly absorbed”, because you need to take into account the whole atmosphere and its constant, ongoing energy-exchange processes. That means taking into account all absorption, all re-emission, all collisions, all heating and cooling and all eventual loss to space, at all levels.
If the amount of radiation lost to space is equal to the amount coming in from the Sun, Earth is said to be in energy balance. But if the strength of the greenhouse effect is increased, the amount of energy escaping falls behind the amount that is incoming. Earth is then said to be in an energy imbalance and the climate heats up. Double the CO2 concentration and you get a few degrees of warming: double it again and you get a few more and on and on it goes. There is no room for complacency here. By the time just one doubling has occurred, the planet would already be unrecognisable. The insulation analogy in the myth is misleading because it over-simplifies what happens in the atmosphere.
Please use this form to provide feedback about this new "At a glance" section. Read a more technical version below or dig deeper via the tabs above!
Further details
This myth relies on the use of a word – saturated. When we think of saturated in everyday use, the term 'soggy' comes to mind. This is a good example of a word that has one meaning in common parlance but another very specific one when thinking about atmospheric physics. Other such words come to mind too. Absorb and emit are two good examples relevant to this topic and we’ll discuss how they relate to atmospheric processes below.
First things first. The effect of CO2 in the atmosphere is due to its influence on the transport of 'electromagnetic radiation' (EMR). EMR is energy that is moving as x-rays, ultraviolet (UV) light, visible light, infrared (IR) radiation and so on (fig. 1). Radiation is unusual in the sense that it contains energy but it is also always moving, at the speed of light, so it is also a form of transport. Radiation is also unusual in that it has properties of particles but also travels with the properties of waves, so we talk about its wavelength.
The particles making up radiation are known as photons. Each photon contains a specific amount of energy, and that is related to its wavelength. High energy photons have short wavelengths, and low energy photons have longer wavelengths. In climate, we are interested in two main radiation categories - firstly the visible light plus UV and minor IR that together make up sunshine, and secondly the IR from the earth-atmosphere system.
Fig. 1: diagram showing the full electromagnetic spectrum and its properties of the different bands. Image: CC BY-SA 3.0 from Wikimedia.
CO2 has the ability to absorb IR photons – it is a 'greenhouse gas'.So what does “absorb” mean, when talking about radiation? We are all familiar with using a sponge to mop up a water spill. The sponge will only absorb so much and will not absorb any more unless it's wrung out. In everyday language it may be described, without measurements, as 'saturated'. In this household example, 'absorb' basically means 'soak up' and 'saturated' simply means 'full to capacity'. Scientific terms are, in contrast, strictly defined.
Now let's look at the atmosphere. The greenhouse effect works like this: energy arrives from the sun in the form of visible light and ultraviolet radiation. A proportion reaches and warms Earth's surface. Earth then emits the energy in the form of photons of IR radiation.
Greenhouse gases in the atmosphere, such as CO2 molecules, absorb some of this IR radiation, then re-emit it in all directions - including back to Earth's surface. The CO2 molecule does not fill up with IR photons, running out of space for any more. Instead, the CO2 molecule absorbs the energy from the IR photon and the photon ceases to be. The CO2 molecule now contains more energy, but that is transient since the molecule emits its own IR photons. Not only that: it's constantly colliding with other molecules such as N2 and O2 in the surrounding air. In those collisions, that excess energy is shared with them. This energy-sharing causes the nearby air to heat up (fig. 2).
Fig. 2: The greenhouse effect in action, showing the interactions between molecules. The interactions happen at all levels of the atmosphere and are constantly ongoing. Graphic: jg.
The capacity for CO2 to absorb photons is almost limitless. The CO2 molecule can also receive energy from collisions with other molecules, and it can lose energy by emitting IR radiation. When a photon is emitted, we’re not bringing a photon out of storage - we are bringing energy out of storage and turning it into a photon, travelling away at the speed of light. So CO2 is constantly absorbing IR radiation, constantly emitting IR radiation and constantly sharing energy with the surrounding air molecules. To understand the role of CO2, we need to consider all these forms of energy storage and transport.
So, where does 'saturation' get used in climate change contrarianism? The most common way they try to frame things is to claim that IR emitted from the surface, in the wavelengths where CO2 absorbs, is all absorbed fairly close to the surface. Therefore, the story continues, adding more CO2 can’t make any more difference. This is inaccurate through omission, because either innocently or deliberately, it ignores the rest of the picture, where energy is constantly being exchanged with other molecules by collisions and CO2 is constantly emitting IR radiation. This means that there is always IR radiation being emitted upwards by CO2 at all levels in the atmosphere. It might not have originated from the surface, but IR radiation is still present in the wavelengths that CO2 absorbs and emits. When emitted in the upper atmosphere, it can and will be lost to space.
When you include all the energy transfers related to the CO2 absorption of IR radiation – the transfer to other molecules, the emission, and both the upward and downward energy fluxes at all altitudes - then we find that adding CO2 to our current atmosphere acts to inhibit the transfer of radiative energy throughout that atmosphere and, ultimately, into space. This will lead to additional warming until the amount of energy being lost to space matches what is being received. This is precisely what is happening.
The myth reproduced at the top – incorrectly stating an analogy with roof insulation in that each unit has less of an effect - is misleading. Doubling CO2 from 280 ppm to 560 ppm will cause a few degrees of warming. Doubling again (560 to 1130 ppm) will cause a similar amount of additional warming, and so on. Many doublings later there may be a point where adding more CO2 has little effect, but recent work has cast serious doubt on that (He et al. 2023). But we are a long, long way from reaching that point and in any case we do not want to go anywhere near it! One doubling will be serious enough.
Finally, directly observing the specific, global radiative forcing caused by well-mixed greenhouse gases has - to date - proven elusive. This is because of irregular, uncalibrated or limited areal measurements. But very recently, results have been published regarding the deep reinterrogation of years of data (2003-2021) from the Atmospheric Infrared Sounder (AIRS) instrument on NASA's Aqua Satellite (Raghuraman et al. 2023). The work may well have finally cracked the long-standing issue of how to make finely detailed, consistent wavelength-specific measurements of outgoing long-wave radiation from Earth into space. As such, it has opened the way to direct monitoring of the radiative impact (i.e. forcing + feedback) of greenhouse gas concentration changes, thereby complimenting the Keeling Curve - the longstanding dataset of measured CO2 concentrations, down at the planet's surface.
Note: Several people in addition to John Mason were involved with updating this basic level rebuttal, namely Bob Loblaw, Ken Rice and John Garrett (jg).
Last updated on 31 December 2023 by John Mason. View Archives
Stealth, I have replied to your comments about the quality of models on a more appropriate thread: the counterargument to "Models are Unreliable."
Scaddenp @242: I don’t think I’m confusing models. I think understand the differences between them and their various purposes, at least in a high-level way.
As for the 5.35 value, I read a paper (which I cannot find now) that was trying to find the CO2 finger print in satellite measurements, and they only looked at clear skies over the Pacific Ocean. I thought it was Harries et al 2001, but it must have been something related. Nevertheless, the forcing value for CO2 has been stated as:
ΔF = 5.35*ln(C/C0) W/m2
Is the 5.35 value true for clear skies only, or entire average of the whole atmosphere? Others in this thread have implied that this is the forcing for CO2, which implies all conditions. But the empirical measurements seemed to be done only for clear skies. If it has been measured in clear skied only, then what does the forcing equation look under cloud conditions? If clouds completely absorb all IR in the bands that CO2 can affect, then this would imply that CO2 cannot absorb any more and the forcing function would be zero when there are clouds. If 65% of the earth is covered with clouds then wouldn’t the 5.35 value be multiplied by .35 since only 35% of the sky is clear? This would imply that the value is 1.87. Or, is the value for clear skies really 15.29 and then when it is averaged over the globe it becomes 5.35? I am not making any claims about the value, just asking if this forcing equation for CO2 is valid through clouds.
Stealth - "Is the 5.35 value true for clear skies only, or entire average of the whole atmosphere?"
Myhre 1998 refers to and builds upon Myhre 1997, which states that "Spatial and temporal variation in the radiative forcing due to variations in temperature, humidity, and cloudiness has been taken into account on the basis of observed data." This supported the rational for (and accuracy of) using multiple atmospheric columns in computing the Myhre 1998 results. So, yes, the CO2 forcing function is for the atmosphere as a whole, not just clear skies.
As for getting source, I’ve chased down through many of the links provided and yet have been able to get to any source code. I recall doing this a while ago, and getting code is not as easy as you all imply. The NASA GISS ModelE is restricted to those with a nasa.gov IP address. When I try to download one of the ModelE snapshots, I just get an invalid ZIP archive. CESM from NCAR/UCAR only talks about the various versions, but I don’t see any accessible source code. EdGCM 4D appears to be built on top of ModelE and doesn’t provide source code. It looks like a pre built package that one can install and play with, but no source code. The subversion source code repository for CCSM4.0 requires a username and password.
Well, this is enough wasted time looking for model source code with no luck. I’ll look more at a later time. Have any of you that provided links (most of whom implied that I'm an idiot that can't do a 2 second Google search) actually download any source code for any model?
But I was stuck by this statement from the Azimuth project (http://www.azimuthproject.org/azimuth/show/Climate+model):
“General circulation models used to simulate weather and climate do not operate at fine enough grid resolutions to resolve many observed regional weather and climate features.”
This comment from the Met Office on the same page is also interesting:
“Clouds could create positive or negative climate feedbacks and are an ongoing area of research. One example is low-level clouds, especially stratocumulus, which help reflect sunlight and keep the Earth cool. The more stratocumulus we get over the planet, the more cooling effect. If our warming climate creates more low cloud, this would be a negative feedback — helping to offset the heating by reflecting more sunlight away from Earth. If our current climate change means there will be less low cloud overall, then this would be a positive feedback — contributing further to the warming by allowing more sunlight in.”
The climate models seem to be coarse in resolution, and there seems to be uncertainty as to whether or not clouds are a positive or negative feedback. This seems to be an admission that the models a) do not model clouds well, and b) that they could either warm or cool the planet, and c) induce a lot of doubt about the accuracy of the models. How do you develop so much confidence in the accuracy of the climate models when the model developers make statements like this? They seem to be agreeing with me and saying there is sizable uncertainty.
[TD] Cloud feedback claimed to be negative is covered here. But the bottom line of course is how well the models have actually performed, both in hindcasting and in forecasting; that topic is covered here. If the models perform well enough to tell us what action we should take, then it really doesn't matter whether anyone personally believes that they should perform poorly, nor does it matter how much they might be improved.
Stealth, how about Clear Climate Code?
Stealth @252, from Myhre et al 1998, the mean global clear sky forcing of CO2 in 1994 was 1.78 W/m^2 (model range 1.759-1.8 W/m^2). The mean global cloudy sky forcing of CO2 was approximately 1.32 W/m^2 (model range 1.313 to 1.37 W/m^2). The global value for 1994 CO2 concentrations (358 ppmv) using their simple formula is 1.315 W/m^2. From that I deduce that by "cloudy sky" they meant what would now be called "all sky" conditions.
Although clouds will reduce the CO2 forcing, only thick clouds with very high tops will eliminate it entirely. Other than cumulo-nimbus clouds, most low clouds will have very little effect on the CO2 forcing because their tops are well below the altitude at which CO2 radiates to space. Those clouds still have a greenhouse effect because their absorb and radiate across the entire IR spectrum, including in regions in which well mixed gases do not absorb and radiate.
In the more recent Lacis et al (2010) (Preprint version) shows the clear sky CO2 greenhouse effect is 26% stronger than the all sky effect. The single factor addition effect (ie, no overlaps with water vapour) is 29% stronger than the all sky effect. That compares with about 33% stronger for Myhre et al.
Stealth @254, of course the models have large margins of error. Have you not seen the error bars on the temperature predictions:
That does mean we might luck out and bring in a temperature increase of 2 C (the upper limit of what is considered safe) by 2100 even doing nothing; but we equally might get unlucky and see a temperature increase over 5 C. Further, the two scenarios are not equal, with the increase in damage for each extra C of increase in temperature being highly non-linear. Therefore rational policy making should weight the unfortunate outcomes more.
One thing is very clear, however, even given the wide margins of error. Not curtailing emissions will see temperature increases to values that have never been seen by Homo sapiens, and which are likely to radically reshape our world.
Stealth,
As for getting source, I’ve chased down through many of the links provided and yet have been able to get to any source code. I recall doing this a while ago, and getting code is not as easy as you all imply.
Step 1: Click on the link I provided to RealClimate.
Step 2: Click on the very first link in the list, "GISS ModelE".
Step 3: Scroll down to the section "CMIP3 Model Configuration and Description". Where it says "The frozen version of the ModelE code used for CMIP3 simulations and the controls for model description papers is denoted as ModelE1 (internal version number 3.0, dated Feb. 1, 2004). This code can be freely downloaded (as a 1.2 MB gzip-ed tar file) from modelE1.tar.gz.", click on "modelE1.tar.gz". This is the source code for the version used in AR4.
Step 4: Un-zip and un-tar the source code from modelE1.tar.gz. I use 7-zip under windows.
Step 5: Open the "model" folder. Inside you'll find "all the model source code used in all possible configurations".
There are also instructions for downloading the historical forcings to be used as input. doc/modelE.html gives a brief overview. The NASA GISS page links to papers, etc.
I hope that wasn't too difficult.
The climate models seem to be coarse in resolution, and there seems to be uncertainty as to whether or not clouds are a positive or negative feedback.
Of course they're coarse. They're not trying to forecast weather, they're trying to forecast climate, and they're trying to do so in a reasonable time frame. You can predict that summer will be hotter than winter on average, and Miami will be hotter than Boston on average, without knowing what the exact weather will be on a micro-scale.
And as Tom points out, those uncertainties are already baked in to the forecasts. But note that the uncertainty about what we're going to do about AGW is larger than the models' uncertainties about what the consequences of each course of action will be.
Regarding clouds, the reason why we can't be certain whether they will be a positive or negative feedback is precisely because so far it seems to be a wash; the latest results indicate they may be slightly positive. If they were a strong feedback either way then it would be easy to identify. To argue that the models significantly over-state future warming on that basis is to argue that clouds will become signficantly negative in the future despite no evidence to support that — in other words, wishful thinking.
KR @ 248: I get the impression that you do not understand physics; fudge factors are not an accusation of fraud in any way. The comment of “fudge factors” is from Dr. Freeman Dyson, a world renowned physicist -- I am certain he is qualified to speak to both physics and fudge factors. Fudge factors are values that are derived from observation or educated guesses. Often they are fit to curves in order to simplify an equation. As an example, in the CO2 forcing equation, ΔF = 5.35*ln(C/C0) W/m2, the 5.35 value is a “fudge factor”, and so is the natural logarithm function. Over a much larger sample of data, for example, a logarithm with a base of pi may fit better than one with a base of e.
The “laws of physics” are not something given to us from on high. They are simply a mathematical representation of what we think is happening, or a way to describe how something behaves. The laws of physics (i.e. equations with coefficients) are used to make predictions, not some ordained fact. Your statement of “They are full of physics” as a way to assert truth or correctness of the models, is both meaningless and ignorant.
Tom Dayton @255: I downloaded CCC, but it appears to be a model to reconstruct the global temperature anomaly from land and sea records in GISTEMP. While I did ask for “any model”, I was really looking for something like a GCM, and (drum roll) I have finally downloaded CMIP3 source code.
Tom Curtis @ 256: I’m glad you chimed in – I like your responses the best. You provide nice data, graphs, and explanations. Some of the other folks here are a little combative or condescending. That can get me whipped up and want to lash back, and I really try to avoid that because it is not productive.
Based on your explanation, it seems that the forcing function for CO2 is more complicated than stated, and I fully expected that to be true. The coefficient of 5.35 isn’t static, but a function of the amount of clouds (and probably other things). It may be 5.35 for conditions seen today, but if the amount of clouds changes over time (which they will), then that equation begins to break down. This is what I mean when I say “all models are wrong.” This doesn’t mean they are not useful, but it means they are limited in their effectiveness because of the underlying assumptions. How GCMs handle this effect is critical across every equation they use. My primary concern is that almost every equation in GCMs have various limitations or assumptions because they are calibrated to measurements made recently. How those equations, coefficients, and assumptions hold up over time is critical for accuracy.
Tom Curtis @257: What model, or models, produced the chart in your example? I see that it is an IPCC chart, so it might be an aggregation of many models -- if so -- do you know which ones?
If I am reading the chart correctly, even if CO2 emissions were stopped in 2000, then the models predict that temperature will continue to increase at a low rate for far into the future. If this is correct, then I think we’re about to have one heck of a big test relative to solar activity. As everyone knows, the sun is entering an exceptional quite spell, and something that may approach the Maunder Minimum and the Little Ice Age. Many people, and myself included, believe that the sun plays a fairly large role in the climate, whereas many people here and in the peer reviewed literature believe it doesn’t. If global temperature falls in response to the quite sun, and CO2 continues to rise, I think this will be proof that the models and peer reviewed literature are not accurate enough in order to take action about curbing CO2 emissions.
JasonB @258: I followed your instructions and still failed. Trying to extract the ModelE1 tarball using Windows and WinZIP fails. However, I did go to my Linux machine and tried there using the ‘tar’ command, and succeeded. So now I have actual Fortran code for a 10 year old model. That’s a start, I guess. I fully expected this to be hard, if only because model developers and their organizations tend to be protective of their code due to competitiveness.
[TD] Facts trump belief, so for the strength of the Sun's influence versus greenhouse gases see the counterargument to "It's the Sun," and note that it has Basic, Intermediate, and Advanced tabbed panes. For how much effect a Maunder minimum would have, see the Intermediate tabbed pane in the counter to "We're Heading Into an Ice Age."
Please help keep Skeptical Science comprehensible and therefore useful by putting your comments on the appropriate threads. Your comments are valuable to other people only if those people who are interested in your topics can find your comments. Someone interested in the accuracy of GCMs is not going to find your comments on that topic in this "CO2 is Saturated" thread, because they will be looking in the "Models are Unreliable" thread. Your off-topic comments in this CO2 is Saturated thread make it more difficult for people interested in the saturation topic to find comments that are related to that topic. Everybody else, please set a good example by replying in the appropriate threads to off-topic comments here. Stealth, you can see all recent comments regardless of the thread they are in by clicking the "Comments" link in the blue horizontal bar that is at the top of every page.
Stealth @259, thank you for the complement, though I am not sure it is deserved. On occasion I have been quite acerbic myself.
Regarding the models, they are just the GCM used for the IPCC AR4, which are listed on table 8.1 of the WG1 report.
Regarding Clear Climate Code, I believe they are working cooperatively with NASA GISS to produce a python version (from memory) of the GISS Model E. I am unsure how far that has advanced, and if completed I am unsure who is hosting it.
Regarding "fudge factors", there is in fact a quite reasonable use of the term in physics as you point out. If this was just physics, however, nobody would be disputing Michael Mann's Hockey Stick, and nobody would be disputing the existence of global warming. There are, unfortunately, people whose approach to this debate is entirely governed by politics. Those people would, and have in the past, take words like "fudge factor" and interpret them out of context to use as just one more weapon in their propaganda war. Therefore we have to recognize words that could be so used, and be very carefull to use exact words that are not prone to misinterpretation.
Regarding the Sun, currently many solar physicists are predicting an extended and significant minimum. Those predictions, from what little I know, have been aound for ten years at most, and are not based on models of solar physics, but on short term correlations. They may be correct, or we may simply be getting a repeat of the very low solar activity around 1910 (which we have not yet dropped below). However, it has been shown that a genuine repeat of the Maunder Minimum would only very slightly delay global warming. That is, with such a minimum and ongoing business as usual, the temperatures we would otherwise have reached in the early 2090s will be reached in 2100.
Finally, forcing is always defined by reference to a particular year. Therefore, any change in cloud cover with respect to that year would constitute a feedback rather than a reduction or enhancement of the CO2 forcing. As it happens, the reference year used by convention is 1750, so any difference in the extent or type of cloud cover between 1994 and 1750 would result in an error in calculating the greenhouse forcing. Of course, whatever the size of that error, it will be matched by an equal and opposite error in calculating greenhouse component of the cloud feedback. So, while your point is technically correct if properly stated, it is largely an academic issue about the use of terms. It is also an unresolvable issue in that we cannot hope to have sufficiently accurate historical records or proxies of cloud cover in 1750 to resolve the point.
[TD] Will everyone please help get the conversations onto the right threads, since Stealth is not cooperating but his/her comments and everyone's responses are too useful to delete?
Moderator [TD] @260: My apologies; I will try harder to stay on topic.
Back to the CO2 forcing equation: ΔF = 5.35*ln(C/C0) W/m2
I have read the Myhre 1998 paper, and it appears that the 5.35 coefficient was derived from a combination of using LBL, BBM, NBM, and HITRAN models.
I have read the Schmidt 2010 paper (http://meteo.lcd.lu/globalwarming/Schmidt/attribution_present_GH_effect_2010.pdf) and it uses the NASA GISS ModelE as an attempt to arrive at CO2 affect at absorbing IR radiation. The paper also references peer reviewed sources for CO2 effects that are all over the map: Houghton 1990 states water vapor is 60% to 70% of the greenhouse effect and CO2 is 25%. Lindzen 1991 states water vapor is 98% and CO is 2%. The effects of clear sky, clouds, water vapor distribution vary enormously. Schmidt also states that many “cloud treatments tended to be quite simple.”
It appears that coefficients for the CO2 forcing function equation have been derived from models -- is this true, or have I misread these papers? Are there any empirical measurements made with any land or space based instruments, and if so, how did they separate out CO2 absorption from IR absorption of other gases?
SASM, is this what you're looking for: http://www.cccma.ec.gc.ca/papers/jli/pdf/puckrin2004.pdf?
Stealth @261:
1) You appear to not understand what models are. They are simply the worked out predictions of scientific hypotheses. In the case of LBL models, they are the worked out predictions of hypotheses that have been repeatedly tested to very greate detail in laboratories along with more or less accurate estimates of atmospheric temperature, humidity and gas concentration profiles. The working out is not mathematically complex, although it is tedious. Nor is it chaotic, so errors in initial conditions will be proportional to errors in the final results. In that context, saying that an LBL model predicts something is no different from saying that the only well confirmed radiation theory predicts that thing. It would be like disputing the planned trajectories of space craft as entirely hypothetical because they were worked out on a computer using only Newton's laws of graviation and the known masses and trajectories. Indeed, more absurd in that Newton's laws of gravitation and motion for more than two bodies and a sufficient time are chaotic.
2) Despite that, satellite observations have been used to directly confirm the accuracy of LBL models. One of the earliest attempts to do so was published in 1970:
In this case, the data about temperature and humidity profiles was gathered by a radiosonde near simultaneiously with the satellite observation. The accuracy, even back in 1970, was extraordinary.
More recently (2008), a comparison between two LBL models and satellite observations was made. The following is a scatterplot of measured OLR for 134,862 observations between the satellite and the more accurate of the two LBL models:
And here is a fuller comparison of both models with the satelite:
Note that the least accurate of the two models is never less accurate than 1.33% error, while the more accurate is never less accurate than 0.33% error.
Similar comparisons can be found for even broad band models such as Modtran:
Broadband models use the same well understood physics as LBL models, but use simplifying approximations to reduce the number of calculations needed.
It should be very clear from this that uncertainty from radiative models is almost entirely from uncertainty about the specific composition of the atmosphere; and that Myhre et al's model based estimate of 3.7 W/m^2 +/-10% is sufficiently justified by the data.
3) Your suggestion that estimates of the CO2 contribution appears to be based solely on Lindzen's 1991 throw away comment. Certainly the 1990 IPCC (Houghton) estimate of 60-70% water vapour and cloud contribution is reasonably close to the 1997 Khiel and Trenberth estimate of 72-79%, which is yet closer to the Schmidt et al 2010 estimate 75%. That seems like a fairly smooth refinement of a value as methods (and computers) improve, with the initial estimate being by no means unreasonable.
In contrast Lindzen's "estimate" has no basis. He does not even provide sufficient information as to how the estimate was done to know whether he was estimating the current increase in radiative forcing of CO2 as a proportion of the total greenhouse effect (as assumed by Schmidt et al), or whether he estimated the relative contributions to back radiation (as I have previously assumed). In the first case he is, completely without warrant, assuming that H2O and clouds accounted for 100% of the natural greenhouse effect. In the latter case he is showing a fundamental misunderstanding about the nature of the greenhouse effect (which is about the TOA energy balance). In neither case is his estimate a scientific estimate because he has not provided the necessary detail for reproducibility.
What is more, you are showing a throrough inconsistency. You are calling into doubt detailed and accurate estimates because they are "model based", but are prepared to give Lindzen's "estimate" credence when it appears to be based on no more than a back of the envelope calculation. Frankly, all Lindzen's estimate shows is that he stopped doing science and became a simple denier far earlier in his career than I had previously estimated.
Stealth, in addition to Tom Curtis's excellent reply to you, see Chris Colose's post "Adding Up the Greenhouse Effect: Attributing the Contributions," being sure to click on his links to coverage of that topic by RealClimate among others.
As Tom Curtis correctly points out, the LBL codes are quite accurate and are direct computations from the spetroscopy, not complex models, confirmed by satellite meaasures. And the primary uncertainty in direct effects is from varying composition of the atmosphere, not the LBL models.
However - the uncertainties (+/- 5-10%?) in direct forcing are once again smaller than, and subsumed by, uncertainties in climate sensitivity, which is estimated at 2°C to 4.5°C per CO2 doubling. A high estimate in direct CO2 forcing will lower the CS estimate, and vice versa, as they are related by the total effects on the climate.
So: ΔF = 5.35*ln(C/C0) W/m2 is a demonstrably solid estimate for CO2 direct forcing, quite small uncertainties, and incidentally not affected by the computation methods of GCMs. To bring this discussion back to the opening post, CO2 direct forcing is by no means saturated; it has a logarithmic relationship to CO2 concentration. And the total effects on the climate of GHG increases are bounded by climate sensitivity, not the far far smaller uncertainties with line-by-line radiative code.
Can we leave this particular dead horse in peace now?
Tom Curtis @263:
It has been a little while for me. Sorry for responding slowly and appearing to drag out a topic, but I read what you guys send, I do some research, and I think about things a bit. I’m trying not to jump to conclusions and I am really trying to dig into things to get a better understanding.
1) I think I have a pretty good idea of what models are, why they are built, how they are used, and what many of their weakness are: I have a physics degree and computer science degree and have been software building models for 30 years for aircraft, threats, radars, missiles, various aircraft sensor and weapons, and earth components such as terrain and weather (not climate, but weather as it impacts these systems). I test my models for aircraft with other models, measurement data in labs and anechoic chambers (http://www.tdk.co.jp/tfl_e/chamber/chamber01/5.html), tests from poll models (http://www.thehowlandcompany.com/radar_stealth/Bluefire_Helendale.htm) and actual flight tests. I know from first hand experience that testing is critical and that models and lab tests do not always match very well to what happens in the real world.
My question @261 was more related to verification and validation of the various LBL models as they relate to the atmosphere. My concern was for the accuracy of the 5.35 coefficient in the CO2 forcing equation. If this value was derived from models without any real world measurements to back it up, then I would be highly suspicious of it.
I would not dispute Newtonian mechanics for orbital prediction since that is easy physics. The climate, however, is very complex and making accurate predictions is hard. Think more along the lines of trying to predict where a gold ball will land after being hit with a golf club. Let’s say we need 1% or 2% accuracy too. That is wicked hard even if you know the exact initial conditions of the ball when it leaves the club face. You will have to run a complex and iterative numerical-methods simulation that considers many things: winds, ball drag (which is a function of many things), spin on the ball, lift induced by rotating body, and so on -- hard stuff to be accurate. Think about why we have GPS and laser guided bombs (which I know a lot about) – we do not use dumb unguided bombs because they are very inaccurate -- we can model the physics with extreme precision but still cannot hit anything with a dumb bomb. There simply is too much uncertainty in the environment, even on a very small scale. Therefore, we install guidance and targeting systems to compensate for errors and drift as it develops during the bomb delivery. (I also think the iterative growth in errors within GCM is a huge problem, but I’ll get into that over on the models topic).
2) I dug into Dessler et al 2008 in JOGR, and that was an impressive study and test. That helps me gain confidence that the models are at least in the ball park in terms of accuracy with respect to the real world. Excellent charts, data, and a great study on Dessler’s part.
3) I was wondering if the Lindzen reference was the same guy at MIT. I wasn’t sure and I wasn’t claiming or supporting it at all. It was just a data point in a list of points about the effects of CO2 and how they varied a lot. Granted, Lindzen was an extreme outlier and I withdraw his data point from my previous point @261.
I spent a bit of time playing with MODTRAN running numbers on the reduced IR flux from CO2 and cloud effects. Since the chart you reference above shows that MODTRAN matches fairly close to IRIS Satellite data, I think my analysis below should be accurate.
I found that for no clouds and CO2 at 294.3 ppm the IR flux is 288.97 W/m2. At 800 ppm CO the IR flux is 284.39 W/m2. The ratio of 800 to 294.3 is approximately e, so the natural log of this is very close to1. Using this I compute that coefficient for the CO2 forcing function is 4.58. This is 15% less than Myhre 1998, and his value of 5.35 is 15% less than the previous IPCC value of 6.3. I am sure that Myhre is a good scientist, but it seems that MODTRAN has been shown to be accurate, yet its output disagrees with Myhre. I find the wide range of values from MODTRAN, Myhre, and IPCC over the last decade not very reassuring in that scientists have a solid grip on these values. I know and fully accept that it is a very noisy world, so I fully expect that this value is hard to determine.
Furthermore, when I run similar numbers using clouds in MODTRAN, I get an IR flux of 261.78 W/m2 for 294.3 ppm CO2, and 258.11 W/m2 at 800 ppm CO2. This yields a CO2 forcing function coefficient of 3.61, which is 21% less than MODTRAN with no clouds, and a full 33% less than Myhre. So while CO2 is not technically fully saturated, its effect is small relative to clouds. The reduced IR flux just from clouds alone is 27 W/m2 when compared to clear sly. Since 60% to 70% of the earth is covered with clouds, it seems logically correct to me that clouds are a major player in IR flux. If doubling CO2 reduces IR flux by 3.7 W/m2, but clouds reduce it by 27 W/m2, then clouds are over 7 times stronger than CO2. A small change in cloud coverage could easily overwhelm the effect of CO2, making the issue of whether or not CO2 is fully saturated really a moot point. Is this not correct?
Given the magnitude of the effect of clouds, how has climate science determined that most of the global warming since 1950 is most likely due to CO2? That seems like an impossible thing to determine. In 1950 CO2 was about 310 ppm and today it is 400 ppm. Even using Myhre’s CO2 forcing coefficient of 5.35, CO2 has only blocked 1.36 W/m2 in the clear sky portion of the atmosphere. Where there are clouds the effect is even less. Isolating CO2 when so many other moving parts of the earth’s energy balance are also changing seems impossible, especially given we do not have very good data on clouds since before satellites (about 1970). How can the IPCC claim that it is very confident that CO2 has caused most of the recent warming?
KR @265. The devil is always in the details. The horse is never dead. There is nothing that is *true*. Science can only falsify things and cannot prove something to be true. Some scientific theories are accepted as true only because they have withstood the test of time and have not been shown to be false. I seriously doubt that climate science is any where near achieving this status.
[DB] You are all over the map here, with much of your comment off-topic for this thread. Please direct comments towards models to one of the model thread. Your question about attributions is discussed on a number of threads, summarized here.
Assuming scientists do not know what they are doing because it disagrees with your preconceptions is simply arguing from your personal ignorance. Much of what you claim is unknown is actually fairly well-understood by science.
It would be helpful to adopt a more streamlined approach to both your comment construction, the threads on which you place them and operate with less of a presumption that the science must be wrong.
Please can we make sure that you are not confusing models in the sense of GCM (which would have a complexity considerably larger than say CFD models for airplane simulation) and LBL models which would be at least an order of magnitude simpler, mathematically and computationally. I cant see why you think the accuracy of 5.35 is so important - how much difference does it make whether it is 5.1 or 5.8? To my mind, this part of the equation pales to insignificance compared to uncertainities in feedback.
It also puzzling how you can read and Dessler (read the more recent papers) and yet make a statements about effect of cloud uncertainties. The best we have would put the net effect of clouds about approximately zero. If you have a substantial criticism then perhaps follow up here or here. Note that clouds are both positive and negative.
It would also be interesting to know what your alternative hypothesis is. If the obviously change in net forcings is not cause of observed climate, what do you propose instead and what is your case for this making physical sense.
Sorry looking further up, I see you have answered the question about the distinction in models.
Moderator [DB] post to my comment @266: I cover several things in my post, but they covered the three items in Tom Curtis’ post at @263. He was not called out for “being all over the map.”
He claims in 263:1 that I do not seem to understand what models are, and I show that I build software models for a living. I may not know a lot about LBL models, but I know a lot about software, physics based models, testing models, and so on.
Then Tom Curtis covers lots of excellent data in 263:2 and the Dessler 2008 paper. This shows that LBL models are accurate in computing reduced IR flux for CO2. This is directly related to whether not CO2 is fully saturated, is it not? I was questioning this because I was concerned about the accuracy of models because Myhre 1998 use models to arrive at the 5.35 coeffcient in the generally accepted CO2 forcing function (ΔF = 5.35*ln(C/C0) W/m2). TC showed support that models (including MODTRAN) are pretty accurate. I believe that now, for the most part.
In 263:3 he jumped on Lindzen with valid comments, so I with drew my usage of his comment about the effects of CO2.
Then my longest part is a lot of data about MODTRAN and the reduction of IR flux. MODTRAN has been shown to be accurate by Tom Curtis in 263:2 and Dessler 2008, so I am using its data. My ultimate point is that both Myhre and MODTRAN cannot be correct. The entire topic is about whether no CO2 is fully saturated. This is all about the CO2 forcing function (ΔF = 5.35*ln(C/C0) W/m2) from Myrhe 1998 and the coefficient of 5.35. I run MODTRAN and I get 4.58 in clear skies. In cloudy skies I get a value of 3.61. Most of the planet is cloudy, so the 3.61 value, if MODTRAN is correct, is probably more accurate. These values from MODTRAN disagree with Myrhe 1998. I am not claiming incorrectness on anyone’s part – only showing that they are inconsistent. You guys cannot claim that CO2 is not fully saturated and it has a large effect because of the 5.35 coefficient, then claim that MODTRAN is accurate.
Does this help clear up my point? And does it show that I am actually on topic?
I've returned to see what discussion followed the questions I asked a while back and have a few comments.
First, I would like to say that I have enjoyed reading the questions posed by Stealth. I am six years into a graduate program in physics and appreciate that even some of the most "basic" phenomena can be hard to model.
That said, I would like to point out that Stealth comes across to me as a true skeptical scientist. A quick Google of define:skeptical yields:
As I was reading the responses to his questions, I also felt they were needlessly "combative". Some of them, I think, would certainly qualify as "ad hominem"...
Whatever. Just sort of disappoints me when I see the response to a well-thought out question on a website that intends to teach is criticism of the asker. That said, the parts of the responses that were science I liked.
Back to the original topic (and my prior question), it is true that the measured outgoing spectra and the atmospheric temperature profile are coupled. It is also true that, in Harries et al's more recent publication, no significant difference in absorption for CO2 is observed between 1997 and 2003. The original post here uses the observed difference as definitive proof that CO2 absorption is not saturated. Should the conclusion be changed to something like "Assuming the validity of the temperature profiles resulting from simulations to match the outgoing spectra, CO2 absorption is not saturated."?
basnappl - "...in Harries et al's more recent publication, no significant difference in absorption for CO2 is observed between 1997 and 2003"; a reference please? I am not aware of such a paper.
Chen and Harries et al 2007, "Spectral signatures of climate change in the Earth’s infrared
spectrum between 1970 and 2006" clearly indicates the same top of atmosphere (TOA) spectral changes as predicted by LBL models and described in Harries 2001.
Speaking for myself, some of the strong responses were due to Stealth dismissing (repeatedly) the accuracy of well supported LBL spectral calculations by inappropriately considering them to have the uncertainties associated with far more complex global circulation models (GCMs). Those are not equivalent, and his comments thereof were unsupportable.
Hello KR,
In post 234, I quote Harries 2007 paper (Griggs, J. A., J. E. Harries, 2007: Comparison of Spectrally Resolved Outgoing Longwave Radiation over the Tropical Pacific between 1970 and 2003 Using IRIS, IMG, and AIRS. J. Climate, 20, 3982–4001.)
Regarding Stealth's "unsupportable" comments: It seems to me that at least part of his question is to what extent are lab models of radiative transfer representative of what is actually occurring in the atmosphere. I think this is a fair question. He did add a perhaps unneccessary bit of opinion but I took that as supplying context for why he might feel compelled to post here at all. Essentially, I think, he's saying that he has a lot of experience with applying non-climate physics based models to the real world and that this experience, along with Dyson etc's statements and what appears to be variability in a pretty foundational number (CO2 forcing), has led him to be skeptical about climate models.
It bothers me that his skeptisism wasn't greeted more warmly. I understand that this is an enormously complex system and it must be tedious for those that know a lot about it to try to bring others up to speed. I imagine it doesn't help when those asking questions start with the hypothesis that this is all a bunch of hooey (which I don't think Stealth was). But, that's the way it is with climate science. There's no denying it's been politicized and lots of people who have no business voicing an opinion do. That's not (as far as I can tell) what the case was with Stealth. Bear in mind, the entire purpose of this website is at least partially political. It wouldn't make much sense for me to make a skepticalscience page regarding high-temperature superconductors even though it's an open, important problem. This site is, in part, meant to be a political counterbalance to non-science. I think that's a good thing, but I also think that attacking people is an ineffective way to change their minds.
Reading that paper I see: "In Fig. 8, both the observations (Obs) and simulation
(NCEP) show a rather flat difference spectrum, close to zero except in the 2
band of H2 O. This would be expected, since in the short time interval between 1997
and 2003 little growth of greenhouse gases occurred" Note Obs and simulation showing little difference.
On the other hand, the paper is yet another (using new instruments) showing that the real atmosphere very much behaves in accordance with the models. Skepticism is good, but it seemed to me in the end, Stealth was determined to misunderstand, more like someone looking desparately for a reason to dismiss climate science rather than understand it.
The fact remains -the observational evidence from multiple sources supports the contention that LBL models are very accurate. The errors in estimating CO2 forcing from change in GHG are very small compared to errors in estimating feedbacks so I struggle to understand the hang-up.
basnappl - scaddenp has an excellent point; 7 years is a very short time in terms of CO2 increase, and given little GHG change we can but expect little spectral change over such a short period. But _all_ of the Harries papers provide solid support for the accuracy of LBL models.
As I noted above, I believe some of the response to Stealth involves the fact that he repeated unsupported assertions of uncertainty in LBL models after and through multiple pointers to the data. IMO - It's one thing to discuss a point in question, another entirely to ignore responses.
I’m new to this thread, and I confess that I’ve not yet read all the preceding comments. I’m still trying to construct a good mental image of the role of CO2 in global warming.
According to the article that starts this thread, “Consider the CO2 absorption band around 15 μm (about 650 cm-1), it is strong enough to not let any light go through after a few tens of meters at surface temperature and pressure.”
Now I believe this statement to be even more generally true because the earth’s blackbody radiation in the entire 750 to 600 cm-1 band is around 3x1022 photons per square meter per second. Furthermore, I estimate that near the earth’s surface there are about 1x1022 CO2 molecules per cubic meter. Thus a few tens of meters of the near-earth atmosphere should be plenty to absorb all photons in the band, not just the central emission at 650 cm-1. The high rate of collision between bending-mode CO2 (v2 CO2*) and the other atmospheric molecules will transfer that vibrational energy to translational energy quickly, thus converting the photon energy to thermal energy. The v2 CO2* population should stay very close to the distribution predicted by Boltzmann statistics for the observed temperature and pressure ranges. Any increase in CO2 concentration within these “few tens of meters” will not lead to additional warming.
Of course the observed increase in CO2 is not confined to very near the surface of the earth.
This is where I’m at so far in my mental picture. I want to make sure I’ve got this right before I continue. Any corrections or suggestions would be most appreciated.