A recent BBC radio documentary on the possible over-selling of climate change, focussed on the link between high profile papers appearing in Nature or Science, the press releases and the subsequent press coverage. One of the examples chosen was the Stainforth et al climateprediction.net paper that reported the ranges of climate sensitivity within their super-ensemble of perturbed physics runs. While there was a lot of interesting science in this paper (the new methodology, the range of results etc.) which fully justified its appearance in Nature, we were quite critical of their basic conclusion – that climate sensitivities significantly higher than the standard range (1.5 – 4.5ºC) were plausible – because there is significant other data, predominantly from paleo-climate, that pretty much rule those high numbers out (as we discussed again recently). The press coverage of the paper mostly picked up on the very high end sensitivities (up to 11ºC) and often confused the notion of an equilibirum sensitivity with an actual prediction for 2100 and this lead to some pretty way-out headlines. I think all involved would agree that this was not a big step forward in the public understanding of science.
Why did this happen? Is it because the scientists were being ‘alarmist’, or was it more related to a certain naivety in how public relations and the media work? And more importantly, what can scientists do to help ensure that media coverage is a fair reflection of their work?
A point that shouldn’t need repeating is that the media like a dramatic statement, and stories that say something is going to be worse than previously thought get more coverage than those which say it’s not going to be as bad. It’s not quite a fair comparison, but witness the difference in coverage for the recent Hegerl et al paper, which presented evidence that really high sensitivies are unlikely (a half dozen stories), and the Stainforth et al paper (hundreds of stories). (As an aside, a comment in the documentary that the recent Annan and Hargreaves paper was deliberately ignored by the media is without foundation – GRL is not Nature, and no press release was issued (a press release was issued – apologies). Expecting mainstream press coverage in such circumstances would be extremely optimistic).
Secondly, the scientists also need to appreciate that most journalists will only read the press release, and possibly only the first couple of paragraphs of the press release. Very, very few will read the whole paper. This implies that the press release itself is the biggest determinant of quality of the press coverage, and of course, the press release is generally not written directly by the scientists.
Thirdly, though we are trying to do something about it here, most journalists are not experienced enough in scientific topics to be able to place new results in context without outside help. Often they have a small number of preconceived frames into which they will place the story – common ones involve forecasts of possible disasters, conflict within the community (the more personal the better), plucky Galileos fighting the establishment, and of course anything that interacts directly with politics, or political interference with science. This can be helpful if the scientific story fits neatly into one the boxes, but can cause big problems if the story is either more complex or orthogonal to the obvious frames. Scientists are aware of this, but often are not pro-active enough in preventing obvious mis-framing. This implies that even if a press release is 100% scientifically accurate refection of the original paper, the press coverage can still be terrible.
So what went wrong with Stainforth et al paper? The press release is available here. The only science result in the press release refered to the 11ºC outlier but the release itself is not incorrect. However, both the title ‘Bleak first results…’ and the first paragraphs do not provide any context that would correctly lead a (relatively ignorant) journalist to appreciate that there was even a distribution function of climate sensitivities. I’m pretty sure that the point that was trying to be made was that relatively small tweaks to climate models can change the sensitivity a lot, and that you can’t rule out high sensitivities based on model results alone, but that was not clear for people who didn’t already know the context.
Myles Allen, for whom I have the utmost respect, I think made a rather poor argument in the BBC program. He stated that “if journalists embroider the press release without reference to the original paper, [the scientists] are not responsible for that”. I disagree. Looking at the press release, one could have predicted with high confidence that much of the coverage would focus solely on the 11ºC number and that they would assume that this was a new prediction. As scientists, I would argue that we have to take responsibility for how our work is portrayed – and if that means we need to provide better context, then we need to insist that that is included in the release. Myles is on much stronger ground when he argued that the mean model response (~3ºC sensitivity) wasn’t terribly interesting because it is just a reflection of the basic model they started with before any perturbations, which is true. However, without some statement about the relative likelihood of any of the high-end numbers, I find it hard to see how the journalists could have got the message right. Having said that, implications aired in the program that the scientists deliberately misled the journalists or said things that knew would be mis-understood are completely without foundation. (Update: Please see the response of the journalists listed in this comment below to really underline that).
What can we learn from this? The first and most fundamental lesson is that scientists should not relinquish control of the press releases. Public relations professionals are talented and useful when it comes to writing releases for media consumption, but the scientists have to be fully involved in the process. If there are obvious frames that the scientists want to avoid, they need to be specific within the press release what their results do not imply as well as what they might. A clear statement in the Stainforth et al release that placed the 11 C result in context of how unlikely it was and specifically stated that it wasn’t a prediction would have gone a long way to allay some of the worst coverage.
For an example of how this can work, the Solanki et al paper on solar sunspot reconstructions had a specific statement that their results did not contradict ideas of strong greenhouse warming in recent decades, neatly heading off simplistic (and erroneous) interpretations of their paper. On the other hand, much of the poor reporting related to the ‘methane from plants’ story could have been avoided if the authors had been more upfront in their release that their work was not related to greenhouse gas changes and had no significant implications for reforestation credits under Kyoto!
In summary, I would emphasise that the scientists and the actual papers discussed here and in the BBC documentary were not ‘alarmist’, however there is a clear danger that when these results get translated into media reports (and headlines) that scientifically unsupportable claims can be made. Scientists and the press professionals they work with, need to be very clear that, for the field as a whole, the widest possible coverage for any one paper should not be the only aim of a press release.
All publicity is not good publicity.
Thanks Stewart Argo!!
Re. 99:
I see Mark beat me to it. :)
Still a good read, though.
Re: Raypierre’s response to comment #32—
As nice as your statement sounds (about any relatively educated layperson being able to understand and critically evaluate the basic arguments regarding AGW), I do have my doubts about this. I think that some people have taken the notion of “democracy” a little too far. Sure, everybody has a right to their own opinion, but I think that a democracy does not mean that you should not at least put more respect in the opinions of experts…After all, everyone can’t be an expert on everything. And, quite frankly, especially before RealClimate came along, I think the denialists were generally “winning” on the web as they had arguments that often were more convincing to the average person.
For example, I think the “paper” accompanying the Oregon Petition is quite convincing if you don’t have the sufficient knowledge and background to see through it. And, the biggest problem I saw was that many climate scientists seemed to believe that things like the IPCC report and the NAS report could somehow counter the messages being put out by things like the Oregon petition. Unfortunately, I don’t think it works to just tell people the correct science…You also have to explicitly explain to them why the incorrect arguments propagated by the denialists are incorrect. Again, I applaud RC for finally filling this void. [Another way to look at this is to note that a dry report, like what the IPCC and NAS put out, that critically evaluates the science will lose almost everytime in a “debate” against something put out by the denialists with the expressed intent of convincing people of a certain point of view by cherry-picking the evidence and so forth.]
When I talk to people about climate change (and the one time that I gave a talk on climate change at a physics colloquium), I always like to emphasize the fact that I am a PhD physicist who has spent considerable time reading up on the issue, including many of the actual papers in the peer-reviewed journals, but even with that background I still am not arrogant enough to believe that this qualifies me to have a truly independent opinion on the subject. (And, thus, I explain, what I will present to them is basically what the consensus opinion is in the field as I, as an outsider, understand it.)
So, to sum up, I think democracy works best when those involved do have enough humility to put at least some trust in experts. Of course, as much as possible, I think it is good for these people to educate themselves on the subject…but they should do so with great humility…i.e., with the idea that they are still going to be at a considerable disadvantage in understanding to those who are trained and working in the field. Or, as I like to tell people, even if I read up quite a bit on flying airplanes, I would still not be silly enough to believe when I step on a 747 that I have just as good an opinion about how to fly the plane as the pilot does.
Wow, I applaud your post Joel Shore. I have posted several messages earlier about how the general public is under-educated in science [and often this is not their fault]. If people were offered a better education in science [any science] in high school, I think they would be able to respect expert opinion more. This is like anyone watching a ballet performance thinking they are qualified to judge the skill of the dancers. Each person may form an opinion of how enjoyable the performance was, but only a professional ballet dancer is qualified to judge the merits of the performance. To think otherwise is simply ignorant arrogance.
RE #100[Response by raypierre]:
As the OP originally stated:
However, this points not so much to “methodological” problems with the paper (a poor choice of words on my part), but to an outstanding issue with GCM models in general. I think what climateprediction.net are doing is in fact exactly what needs to be done to understand the extent to which GCM models are too heavily parameterized to generate reliable predictions.
The correct interpretation of their results is not that climate sensistivities are significantly higher than “the standard range”, but that under plausible parameter variations, the models exhibit implausible sensistivities. That points to fundamental problems with the model, not necessarily fundamental problems with the climate.
I am a skeptic precisely because a lot of what passes for climate science is in fact “climate model science”, and the models are nowhere near well-enough understood yet to equate the two.
Alexi,
On the subject of disclosing and defending assumptions, you should make it more explicit that underlying all of your arguments about climate behaviour and prediction is the assumption that climate is chaotic, I have even seen you defend the statement that climate is chaotic on all timescales.
I may have asked you this before, but woud like to know on what evidence do you base this assumption? While the climate may be chaotic on some timescales (ie millions of years at one end and very short timescales where it interfaces with weather) it seems to me that all evidence thus far indicates that climate is broadly deterministic in its response to forcings, at least on any timescale that concerns policy decisions.
Why are you so convinced it is chaotic and fundamentally unpredictable?
I remember that before the ozone depletion science was proved, that the science (models and understanding) underestimated the CFC and HFC damage potential to the ozone layer. Science can indeed lean to the conservative side.
I hope the GW science is not also being too conservative. There’e lots we still don’t understand about GW positive feedbacks…such as when glaciers suddenly started collapsing when a critical part of it disappeared, or of all the mechanisms for polar ice melt which accelerated polar ice melt beyond many predictions.
RE #112: I don’t understand. What does fuzzy logic have to do with it?
If a model is over-parameterized with respect to the available data, you can’t be confident of its out-of-data predictions. That’s basic statistics – it’s called overfitting.
One way to (empirically) determine whether a model is over-parameterized is to vary its parameters within reasonable physical boundaries, and look at whether the predictions continue to be reasonable. That’s pretty much what the cp.net folks did, only instead of interpreting their results as casting doubt on the models’ predictive capabilites, they interpreted their results as cause for greater alarm about global warming.
Re #108: Coby, “on what evidence do you base this assumption [climate being chaotic]”? and “it seems to me that all evidence thus far indicates that climate is broadly deterministic in its response to forcings”.
I am not sure which evidence do you have in mind. For example, this very relevant framework (and the subject of this thread BTW) obviously deviates from your line of evidence:
http://www.climateprediction.net/science/strategy_adv.php
It is useful to try to approach the problem from the opposite end. Every motion that involves more than three molecules is chaotic, and only a few extremely purified problems can be approximately treated as non-chaotic for certain narrow practical uses. Given obviously irregular character of all historical data regarding climate, it is a tough sell to start with an assumption that everything is in a global equilibrium, and only spontaneous external events like eruptions or large meteors are the cause of that variability.
People here are trying to artificially separate short-term “internal variability” (weather and maybe some seasonal trends) from effects of “external forcing”. The problem here is that the “forcings” are in fact some other inherent variables of the very same climate-bearing system, but there is no clear separation of time scales that allow for any sort of coherent theories like “averaging of fast motion” that results in Landau-Ginzburg-type equations for “slow envelopes”. This is the same ugly face of the general problem of turbulence – continuous spectrum of scales, so any artificial limit on the length of averages will always miss something, and the simplfied equations cannot form a closed solvable set (someone here was confused about importance and relevance of this problem).
However, no matter how paradoxical it may sound, there is no contradiction between globally-chaotic behavior of climate variables, and “broadly deterministic response” to “forcing”. While the underlying climate variables are obviously fluctuating chaotically within their own ranges and time scales, it is true that at almost any given time the system would react almost deterministically to perturbations, it is a well known fact that geodesic flow on any dynamical system is simple when current state is far from a degenerate point, one can define and calculate Lyapunov exponents along the flow etc. The problem is that the values of Lyapunov exponents usually vary along the attractor (so the direction of “response” may change), but you don’t know where you actually are on the global cycle of nature. Without knowing the bigger picture (i.e having right global model of glaciations-deglaciations), the “local’ responses may be misleading, especially for practical purposes.
For example, if you examine the ice core data, say from Vostok station
http://www1.ncdc.noaa.gov/pub/data/paleo/icecore/antarctica/vostok/
you may find many episodes with “bizzare” behavior. For example, for a whole period of 3000 years, between 234ky BC and 231ky BC, CO2 was generally rising while temperatures were dropping steadily. For another thousand years from 224ky BC to 223ky, the trend was opposite – T was on the rize, while CO2 was going down. From 215ky to 212, for another 3000 years, again the CO2 was up while T was going down. I am sure one can find many more episodes of this sort. The irony is that that were outputs from the same system, that’s why I question completeness of most current models.
It is interesting that Gavin has stated on one of your blog thread,
http://illconsidered.blogspot.com/2006/03/models-are-unproven.html
“Let me state my essential point more clearly – in NO model of the atmosphere that passes basic screening for a reasonable climate and is in equilibirum with the atmospheric composition and sea surface temperature will the sensitivity to 2xCO2 be negative.”
So, it looks like the observaions I mentioned above prove that something is wrong in the process of model construction, some basic assumption. I wonder, which one?
(I’m afraid the above might look as another “multisyllabic gobbledegook” for some, so I will stop right here;-)
[Response: Almost all of these issues have already been thrashed out in the article Chaos and Climate In short, there is no real evidence that climate is chaotic in the usual sense. One cannot infer chaos by just an irregular or “bizarre” appearance of a time series. It is not, in principle, impossible for coupled ocean-atmosphere climate to be chaotic, but all evidence so far points to the likelihood that the strength of the response to GHG radiative forcing changes overwhelms the effect of any chaos there may be in the system. The observations you mention do not by any means prove something is wrong in the process of model construction. You have taken Gavin’s statement out of context, and have completely failed to appreciate that other climate factors need to be taken into account in interpreting the Pleistocene record. Orbital parameters are changing, as is atmospheric dust, and glaciers have a dynamics of their own which can lead to short-term decorrelations of temperature and CO2 over periods of a few thousand years. The CO2 radiative forcing changes over the time periods you cite are far smaller than what would be caused by a doubling of CO2. –raypierre]
Getting back to the issue of “How not to write a press release” readers may well be interested in spiked-online.com’s take on that particular press release http://www.spiked-online.com/Articles/0000000CB027.htm. In difference to the “alarmist” 11 degrees C interpretation they are suggesting that the model is flawed. Its worth getting past their tabloid like headline “Climate change: a model cock-up” to where they do try to address the issues regarding climate modelling.
Re response to #115 by raypierre:
I see some continuing misunderstanding of the subject, or what I tried to say. But first, nothing so far has been “thrashed” or “debunked” in that article, starting form completely false statement (and non-sequitur argument) like:
“Chaos is defined with respect to infinitesimal perturbations and infinite integration times, but our uncertainties in the current atmospheric state are far too large to be treated as infinitesimal”
Or this false construction:
“Although ultimately chaos will kill a weather forecast, this does not necessarily prevent long-term prediction of the climate. By climate, we mean the statistics of weather, averaged over suitable time and perhaps space scales”
As I said, since the original attractor is chaotic, there is no “suitable” time nor space scales, just remember the problem of turbulence.
Or this misappropriate statement:
“We cannot hope to accurately predict the temperature in Swindon at 9am on the 23rd July 2050, but we can be highly confident that the average temperature in the UK in that year will be substantially higher in July than in January.”
Dare to predict whether the 2050 summer season will be dry or rainy in that area, to plan accordingly for agricultural activities?
Or this example of illogic:
“Imagine a pot of boiling water. A weather forecast is like the attempt to predict where the next bubble is going to rise (physically this is an initial value problem). A climate statement would be that the average temperature of the boiling water is 100ºC at normal pressure”
What if your pot has a lid that has certain weight and might melt down over the time? How accurate now is your statement?
About examples of Lorenz attractor. Why do you assume such a limited case of one? Why don’t you consider that you might have a dozen of Lorenz-type attractors, all operating on a certain spectrum of time and space scales, and intertwined to different degrees via their coefficients and “forcing” terms? Say, some internal variable (like ice mass) would modulate the “r” parameter of your main Lorenz attractor on a time scale of 100 years? Dare to predict your 30-years average now?
And what the heck is this, “the directly policy-relevant portion is on the multi-decadal and centennial time scale”? Are you doing a science here, or trying to fit it to the purpose of politics? Why don’t you simply say, hell with all this multisyllabic gobbledegook about chaotic-non-chaotic, we need to act now no matter what the freaking science might be questioning?
In short, the article
https://www.realclimate.org/index.php/archives/2005/11/chaos-and-climate/
completely failed its purpose.
About “forcings”. As I said, all your forcings (and glaciers and dusts) are in fact another variables of the same coupled climate system, and their scales are not separable so you can’t consider them as independent variables. Yet you say that I failed to appreciate other climate factors. How do you “appreciate” these factors yourself when building your current models? How do you know that we are not near that “short term de-correlation” period of some-1000 years long? And which one of the two “de-correlations”? How do you know that the amplitudes of “CO2 radiative forcing changes” were “far smaller” 200,000 years ago while all data are effectively smoothened over 3000-5000 years by specifics of gas occlusion process?
[Response: You’ll find that as a technique of getting answers (assuming that’s the goal), keeping to one topic at a time is more useful. In this post you confuse a number of issues to nobody’s benefit. To take just one – the relevance of the boiling water example. The question you ask is what is the dependence of the boiling temperature as a function of pressure – this is a climate question that can be addressed very simply without considering the details of the turbulent flow. What we are talking about in climate is are small changes in the attractor as a function of different forcings, and we have gone over this before. Simply switching venues and starting over is no way to have a conversation. – gavin]
i find your article and the responses you received very interesting. the real problem here (well the one that you stated) is media coming in the way of science and forming a statement that the masses rely on as the “truth”. every one who has seen scientific briefs know that there is some area within the study that is left unexplored (if everything were explored nothing would be a mystery). just look at that micheal crichton book, i heard that NBC called him up to have him explain his “theory”. what theory? the guy is a fictional writer, for crying out loud…
Apologies all round for some of the confusion in this thread with regard to who is posting. I would like to make this crystal clear. Any comment that is posted where someone pretends to be one of the other posters, will be deleted immediately and the poster permanently blocked.
The busy lay reader generally just wants to know where current scientific thinking is regarding the question: “how hot might the earth get and what is the chance of this occurring?”
If I have it right, the discussion in this thread is about climate sensitivity to CO2 levels and the predicted temperature increases mostly being disputed here do not include any consideration of the possibility of increasing temperature causing increased release of CO2 or methane from natural carbon sinks.
If this is the case, all this discussion about climate sensitivity, while centrally important from the point of view of getting the science right for modelling, is not the immediate key real world question the lay reader wants answered.
The way I read the discussion here (sorry I have not read the original papers being discussed) is that the new work which showed a lower range of temperatures (Hegerl study noted in #2) shows a 5 percent chance of the temperature going over 11 degrees F. That probability needs to be increased by whatever probability estimates climate science can provide of temperatures up to 11 degrees initiating reversals of natural carbon sinks and racking up CO2 equivalent levels further. That then gives a better estimate of the actual real world estimated probability of temperatures going over 11 degrees F. This figure is of more relevance to risk management than raw climate sensitivity estimates (essential though they are for doing the calculations).
Are there estimated probabilities for the initiation of carbon sink reversals for temperatures up to 11 degrees F and how would you best factor them into the probability estimate for going over 11 degrees F.
Re: #111
I agree with one of the ideas that you’ve expressed in previous posts, namely, that we can’t be certain that model parameters or model behavior correctly reflect the actual dynamics, so we have to take model simulation results with a grain of salt. But, your recent posts suggest that you don’t just want to put a grain of salt on it, you want to bury it in a mountain of NaCl.
You remind me of Churchill’s definition of a fanatic: someone who can’t change his mind, and won’t change the subject.
Some of your objections are utter nonsense. Like this one:
Nobody pretended to be able to predict whether summer 2050 will be dry or rainy. The claim was high confidence that summer in the UK will be hotter than winter. You said nothing about that. And the example makes an excellent point, that even chaotic systems can show predictability, especially when they’re subject to forcing or constrained by conservation laws. You said nothing about that, either.
This is ridiculous (by which I mean, worthy of ridicule). I might as well say, “What if your pot is made of green cheese? How accurate now is your statement?”
What’s next? Will you invoke the Heisenberg uncertainty priciple to claim that no computer simulation will ever truly reflect actual climate dynamics?
Re #114 – what you speak of is known as positive feedback and yes I would imagine that most climate models are models of CO2 sensitivity due to CO2 levels rising due to human output and known natural means only and not by additional events such as the permaforst melting, or rain forests drying out of which both would potentially add large amounts of additional CO2 into the atmosphere.
Maybe real climate can clear this one up for us all. Do climate models predict anything other than human CO2 release ?
[Response: Some climate models include full carbon cycle models – these predict CO2 in the atmosphere as a function of human emissions, and feedbacks in the carbon system (uptake by plants, ocean, releases from soils etc.). These generally show a postive feedback on CO2 levels in a warming climate, though the magnitude of the effect is rather uncertain. However, when we talk about climate sensitivity or about the models in the IPCC AR4 comparison, these have CO2 levels and their future trajectory prescribed according to different sceanrios. – gavin]
Here’s a good press release (from the food/safety area)
http://www.nutraingredients.com/news/ng.asp?n=58372-sudan-top-scientists/
#re 116
And those alternatives from the IPCC show us that the worse case scanario is around 11 F of warming?
I have a question that may be considered irrelevant, and I apologize in advance if it is. Just ignore it please in that case. A couple of my students made a “greenhouse” in the form of a glass box. We saw a significant temp increase when we left it under the sun for a few hours. Then overnight, we put some dry ice in the box and sealed it with masking tape on all edges. Left it over night so the dry ice would sublimate and the gases in the box would reach room temp. The next day, we put it under the sun for the same amount of time as the day before. But we didn’t get any more temp increase in the box than the day before. So what did I do wrong with the experiment? I mean shouldn’t the CO2 in the box have “trapped” more heat? I thought this would be a relatively cheap and easy way to demonstrate the greenhouse effect to a bunch of freshmen, but clearly I am wrong. Please help with any suggestions.. Thanks in advance.
[Response: It’s not exactly on the topic of this thread, but there ought to be some place where people can ask such questions. It’s perfectly reasonable. The answer is that the greenhouse effect needs more than just absorption by CO2. It also relies on the radiating temperature being different from the surface temperature, which generally requires a temperature decrease with height. You get that in the atmosphere from compressibility effects, but you don’t get that in your box. For some introductory material on how the greenhouse effect really works, take a look at my post A busy week for water vapor, or at Chapter 3 of my ever-evolving Climate Book. Dave Archer’s global warming book, linked through his web site at U. of Chicago, has a more elementary explanation of the same things. Actually, as a model of the greenhouse effect, the box experiment has a lot wrong with it. It’s actually dominated by the effect of the lid on turbulent heat transfer out of the box, not on the infrared blocking properties of the glass. Many people blame Fourier for the faults of this experiment, but it isn’t generally recognized that the experiment was actually designed by Horace de Saussure, who used it as a means of measuring the intensity of solar radiation on mountains. Fourier used it only as a loose analogy for the principle of energy balance, and stated clearly in his treatise that he knew the shortcomings of the experiment as a model of the greenhouse effect. Some of the historical perspective concerning Fourier’s work can be found in my Fourier essay on my web site (geosci.uchicago.edu/~rtp1 , then follow links to publications), my annotated translation of Fourier’s treatise, and in William Connolloys excellent online translation cited in the latter. Probably at some point I’ll do a post on Fourier and the box experiment, but right now more pressing duties call –raypierre]
What grade level and school system are you teaching in?
I have friends and inlaws teaching first and fifth (US) elementary grades who have teaching material available.
You’ll need to understand some basics, where do you get help as a teacher? Do you mind my asking your credential?
Ok I think I can answer this. Even according to the conventional (wrong) story about how greenhouses work, this would not have been successful. According to that story visible light passes the glass some gets converted to longer wave length light and thus when it trys to exit the greenhouse through the class bounces off. More CO2 and less oxygen or nitrogen would have no easily affect in this scenario. (You need a lot of co2, the amount in an atmosphere to have a similar affect.)
However, when it comes to greenhouses, this explaination of the greenhouse affect is wrong anyway. Many greenhouses are made of plastic which let the long wave lengths in the light spectrum through in any case. What happens in a greenhouse is unlike the larger atmosphere you have traped the air. Hot air rises and is replaced by colder air outside of a greenhouse. Inside a greenhouse, hot air can rise only so far; so the cold layer of air on the bottom can get heated too. Also I suspect that there is less air movement (convection) within a greenhouse than outside, so air in a greenhouse is an insulatior (albeit a poor one). That is what produces the greenhouse affect in greenhouses. The greenhouse affect does work in the atmosphere beacause thousand of meters worth of greenhouse gases do reflect the really long waves back.
OK that is amateur’s attempt to explain. The real scientists can correct anything I have wrong.
Regarding land resposes to co2 emissions This study on maximum limits of land storage for atmosperic co2 claims that at best the land biosphere will remove about 30% the co2 emissions over 50 years.
To illustrate the uncertainty, the same authors claim, “..40% of total anthropogenic emissions have remained in the atmosphere. The remaining 60% was absorbed by the oceans and the land biosphere”
Computing the land bioshpere equilibrium point is a hard problem. The greatest uncertainty is in how we react to the opening up of greater northern hemispherical land mass to longer seasons, the rate at which we abandon older cropland for more intensive agriculture, and the manner in which we introduce plant species, especially in forestry.
Re #115:
Yes, you have mis-interpreted my prior statements about model parameters as being only inaccurate. Now you get it correct: I claim that the linear response models (“sensitivity studies”) are fundamentally insufficient for the purpose of prediction and “policy making”. As an example, when a direct long-term simulation of weather (sort of) is conducted (as in ClimatePerdiction.Net framework, see #67 above), the model shows all sort of behaviors, including measurable number of scenaria when adding CO2 results in _cooling_, which contradicts “more CO2 == always more T” for time horizons longer than 5-10 years. Another historical records of 1000-years-long “de-correlations” are given in #109. Modelling the sensitivity to a parameter is grossly insuffucient when you failed to consider the self-contained dynamics of the parameter itself. You want to measure sensitivity? Fine, but don’t claim a doom please.
About the 2050 summer in Swindon. The original claim was absurd, there is no new information in the statement that summer will be hotter than winter. This is a trivial statement that holds no predictive power. The whole topic is and always was about climate prediction, and specifically for practical purposes and “policy making”. I just tried to demostrate that the climate (even as it is defined) is no more predictable than weather when it comes down to practical purposes. However, I have to make my statement more adequate, it should say: “Dare to predict whether the 2040-2060 summer seasons in the area will be more dry or more rainy than today?”
About the pot with boiling water. You miss the whole point. It does not matter what the pot is made off, green cheese or chicken poop, but it does fundamentally matter if the lid is included in the model or not. Since our real system does have a sort of lid (which builds or releases excess of “pressure”, flipping quasi-randomly), the statement that “average boiling temperature is 100C, and we perfectly know how it depends on pressure” does not have any predictive power since you deliberately excluded ways/models to predict what the actual pressure might be. Therefore the original example about how easy the climate can be predicted is plain incorrect and grossly misleading for “policy makers”, which is not a laughing matter at all.
What is next? I will try to exclude “multisyllabic gobbledegooks” as “uncertainty priciple”, “differentiable manifolds” and “Anosov diffeomorphisms”, and use simple understandable terms like “green cheese”, “pot lid”, and “crackpot”, if you prefer :-)
[Response:What is next is that I’m not going to bother responding to any more of your posts, because you have shown yourself incapable of or unwilling to learn from what you have read. I’m not impressed by your name-dropping either. Plenty of us here know all about differentiable manifolds and diffeomorphisms, and are not particularly impressed that you know how to spell these words. –raypierre]
Re #119: Alexi, I previously posted, on a previous thread I think, regarding my view that we are all ‘gentlehommes’ until proven otherwise. Your first three paragraphs were mostly ok, making good points. Your last paragraph strikes me as an attempt to be insulting, and beyond my understanding of ‘gentlehommes’. Moderate it or I’ll call for a vote.
In the meantime, if you can do a bit better job to sticking to one point at a time, avoiding anything that can possibly be considered insulting, I want to read what you have to say. Sometimes it is actually valuable. Thank you.
Yes, I would also appreciate dropping the hostile back and forth among posters and, instead discussing and criticizing ideas. Eleanor Roosevelt said: simple minds talk about people, average minds talk about events, great minds talk about ideas. I am trying to learn about global warming science and value everyone’s opinion, but the hostile static has got to go.
I should also clarify the specifics of our experiment [Thanks Gar Lipow for your response and maybe specifics would be more helpful]. Our glass box is 2 feet by one foot by one foot. We used about 1.5 pounds of dry ice. And Gar, you are right we should have used plastic and maybe that would be a good change to make for next time. But we weren’t trying to make a greenhouse. We were trying to see the potency of CO2 as a greenhouse gas in trapping heat. Just some details, and I really appreciate any input I can get. Thanks again.
“but the hostile static has got to go.” that’s great. But what about the relentless censorship. Could you put your attitude toward ending that also?
[Response:At the risk of again derailing the topic, you are wrong about this. If people make serious points in constructive ways, it gets through and is responded to. Meaningless attempts at point scoring, repetition of already discussed items, personalisation of scientific topics, trolling for effects etc. do not. It really is quite simple. – gavin]
Ocean, do you have access to public school teaching materials? I know first and fifth grade teachers who have materials for this sort of study.
Seems to me you need some very basic physics first. Heat transmission is not the same as heat storage. When you fill an aquarium with CO2 you make it transmit less of a certain range of infrared wavelengths. If you can get an infrared camera, or infrared film and filter, suited to taking a picture at those wavelengths, you can demonstrate how an aquarium full of ordinary air transmits infrared light vs. the same aquarium full of carbon dioxide.
But that’s going to require understanding the wave theory of light, atomic/molecular absorbtion and so on.
You seem to be thinking that CO2 somehow “traps” or “holds onto” heat as though a tank full could be used to store it, and that an aquarium full of CO2 sitting in the sunlight is going to store up more heat than an aquarium full of air? That’s “specific heat” — explained here:
http://hyperphysics.phy-astr.gsu.edu/HBASE/kinetic/shegas.html#c1
I can’t comment on whether the specific heat has anything to do with the ability to block infrared wavelengths transmitted in the atmosphere. Anyone?
One aside that might be helpful — the amount of CO2 in the atmosphere is in parts per million. Seems like not much. I found this analogy somewhere: ordinary window glass is something around 91 percent transparent, passes that much visible light. A thin sheet looked through seems clear. The same sheet looked through along a long dimension, ‘edge on’, will look very greenish.
Re #120:
David, I am curious where have you been when I was accused of dazzling the thread with “multisyllabic gobbledegook” (#68), and the accusation was even supported by the host, raypierre(#69)? I was depicted as “less than honest about”, “simply making unfounded assertions”, “had any credibility in criticizing climate science”. I am just trying to maintain the same friendly level of conversation. Please be consistent.
Re: 122. CO2 in a box that size will not give you a great difference from air. A 2X1X1 box filled 100% with CO2 will not give you significantly more heat than a box that size filled with air. Remember global warming is caused by layers of CO2 25 km’s high – though only a fraction of a percent CO2. To equal that thickness of CO2 you would need pure CO2 33 feet thick. In another words you would need a box 33 feet high. And even that would not do it because much of the greenhouse effect comes from feedbacks. As the atmosphere warms it can hold more water; that additional water vapor provides more of the warming than is directl caused by CO2. In short to get the affect equivalent to the current affect of CO2 you have to fill a 33 foot high box with CO2 and have the floor cover mostly pans of water . Even so this omits gases other than CO2 which are responsible for about half of human cause global warming. Still if current human contributions have created almost a degree of warming to date, that box might create about half a degree of warming compared to a second box filled with air, floor covered with the same pans of water. Of course there are probably other factors that might make the difference less. Given that small difference you would have to have two boxes at the same time, not the same box at diffent times, cause normal variations in temperature from day to day would swamp your results. And you would have to have awfully sensitive thermometers to capture difference of half a degree or less. Also you would have to be careful about placement. I could think of a few other things that could lose these small differences in noise.
I’m not sure about the experiment except to say that in the greenhouses I built in Maine, I installed heat-activated vents on each gable end otherwise the plants would fry depending on sun intensity. Maine’s very cloudy and humid anyway, cold too, but as an aside they are tapping maple trees in February these days.
Hi all, and thanks for all the responses I just received all of the sudden :) As far as credentials go, I have a PhD in geology. I am a professor at a university. But I just wanted some simple “heat trapping” experiment to give my students first hand experience. But as I said earlier, I am not a climate scientist and have much to learn [as you can see]. But maybe infrared camera is a way to go? I hadn’t thought about that. Hank, my e-mail is [please don’t laugh] ocean_chem_angel@yahoo.com If you have any teaching material they could e-mail me, I would be much appreciative.
Thanks Pete and Gavin for your response in #116 that the estimates for future temperature change being discussed in the climate sensitivity studies (discussed in this thread) do not generally take into account the effect of increased temperature on initiating further natural carbon release. I think that the vast majority of lay readers who read the headlines and the text of stories on climate sensitivity do not know this and they simply presume that the scientists concerned are talking about their absolute best estimates of the possible temperature increases which may be faced. Do you think that in the same way that the Solanki et al paper on solar sunspot reconstructions had a specific statement that their results did not contradict ideas of strong greenhouse warming in recent decades, this (the fact that climate sensitivity projections are not best estimates of possible future actual temperature increases) should be clearly noted in media releases put out by scientists when reporting climate sensitivity studies?
[Response: Yes. That would definitely be useful. – gavin]
Secondly it seems really important to me for the wider discussion (beyond just the topic of this thread) that some estimate of the “uncertain” potentiating effect of predicted temperature increases be provided and integrated with the climate sensitivity predictions so that we can have a statement that actually attempts to predict possible real world temperature increases. Presumably such temperature predictions are the ones that should form the basis for policy decisions.
[Response: Indeed. And in fact they do. Climate sensitivity is something that we (as scientists) get excited about because it is a relatively well-posed question (none of that messy economic analysis or human behaviour include). But the projections that end up having ‘policy relevance’ are only loosely dependent on climate sensitivity. They are more directly related to what is called the transient climate response (which takes into account the ocean lags) and the greenhouse gas and aerosol scenarios. These scenarios are the element that needs to include more of the geo-bio-chemical feedbacks that affect CO2, CH4 and aerosols. That is not yet common practice, but it is starting to be done. -gavin]
I’m slightly surprised no-one has mentioned Piani et al. Constraints on climate change from a multi-thousand member ensemble of simulations also from the ClimatePrediction.net group. It is in my opinion a more significant paper as it manages to get a PDF out of the results (the Stainforth et al. paper does not attach a probability to any of the outcomes, 11 degrees could not be ruled out, but the [very low] probability attached to such high temperatures could not be estimated).
The Piani et al. paper gives “the 5th and 95th percentiles are 2.2 K and 6.8 K respectively.” while the “best estimate of climate sensitivity is 3.3 K”. If it is correct it shows three things:
1. There is almost no chance of a very low climate sensitivity which would enable us to emit CO2 to the limits of available fossil fuels without harmful effects being predominent.
2. 2 K warming is almost inevitable. Even drastic cuts in CO2 emissions are unlikely to be enough, and there is no sign of even modest cuts happening for the next few years.
3. There is a significant probability of extremely harmful temperature rises, even the most drastic cuts do not reduce this probability to miniscule levels.
As interesting as how the media latched on to the 11 K climate sensitivity in Stainforth et al. is their total lack of interest in a much more robust (but equally frightening) result.
I’ll defer to experts at this point.
Gar Lipow and others,
I have a question based on your response to my experiment. Please bear with me, my question may sound stupid, but it is a real question to see if I understand what you are saying correctly :) So the percentage of CO2 in the box is not as important as having layers of CO2? Like one wool jacket won’t keep you as warm in winter as layers of flannel or cotton. Is that the right analogy? Because the atmospheric CO2 concentration is like 0.03% and let’s say we have 50% CO2 in my box. But gas pressure would be higher if not the same as atmopsheric pressure in my box. So more CO2 and more pressure, shouldn’t that enhance warming? Like both Mars and Venus have over 96% CO2 in their atmospheres but Mars barely has any atmopsheric pressure, and Venus’ atmophere has 90 times Earth’s atm pressure. So Venus has the runaway greenhouse effect, right? [Disregarding all the other positive and negative feedback mechanisms involved.]
“aerosol scenarios”…
Let’s see…many models show that aerosols could have been artificially keeping the world’s average surface temperature cooler by about 3-5 degrees C from 1900-2000 – (sulfate aerosols certainly have some certifiable cooling effects cancelling out the warming effects of CO2).
So now get rid of most of the world’s sulfate aerosols in the next 50 years because it’s currently killing people, plants and destroying ecosystems (acid rain, pollution).
…and all by itself…woops…a possible isolated, independent temperature rise of 3-5 degrees C average world surface temperatures by 2100, not even including any other positive forcings, because the forcing is already there waiting for the cancelling aerosol cooling effect to be removed…
That is a possible scenario that would certainly push the actual realized temperature increase to the higher end of the “most likely” projected temperature range in a hurry.
136. No. Basically what we are saying is that normal greenhouse is based on having less thermal mass to heat than the atmosphere as whole due to seperation from that atmosphere, and to a lesser extent on insulation due to the air not moving as much. For a greenhouse a foot or two high, filling it 100% with carbon dioxide won’t make much of difference. Between a two foot high greenhouse filled with atmosphere and a two foot high greenhouse filled with carbon dioxide there would be less than a thirtyieth of a degree difference – provided you had water at the bottom of the greenhouse to provide vapor feedback.
Multiple layers would make a difference but for still a third reason, having nothing to do with either the greenhouse effect or how human made greenhouse work. Air has insulating value, but because it moves a lot this is overcome by convention. Trap the air in smaller values; it moves less, and some of the insulation value overcomes convection. That is why multi-paned windows have more insulating value than single paned one. (Though there are other things you can do to give windows insulation that give better results for the money.) For that matter, this is how some foams work – trapping air in tiny cells so there is no convection to overcome. (In practice most foams use gases other than the atmosphere; but there a few ‘green’ foams that do in fact use normal air, and they work pretty well.)
Bottom line, I can’t imagine offhand how you can demonstrate the atmospheric greenhouse affect in a greenhouse two feet high. I’m not saying it can’t be done; I just can’t think of it. Greenhouses work fine but they work on different principles than the atmosphere does.
Since obviously I was not clear, try this explaination from the Alaska Science Forum.
http://www.gi.alaska.edu/ScienceForum/ASF8/817.html
The effect from greenhouse gases is a different matter.
Re #119 and posts following (the greenhouse box experiment): I think there was a post here at RealClimate that explained that the forcing effect due to CO2 (or other greenhouse gases) is pretty subtle…even if you ignore feedbacks and all that. As I recall, it turns out that the reason why CO2 produces warming is that it effectively causes the energy emitted back out into space from the earth to be emitted from a higher average height in the atmosphere where the temperature T is colder (since the temperature generally drops with altitude through the troposphere). Since the energy emitted goes like T^4 power, the earth thus emits less energy back into space, which is why it has to warm (until it reaches a temperature when the earth is again emitting as much energy back out into space as it receives from the sun and so is back in equilibrium).
Clearly, a little box is not going to have a measurable “lapse rate” of temperature from top to bottom like the atmosphere has and thus this effect will not be observed.
Hopefully, the RC folks will correct me if this is garbled or wrong…but I think this might be the most fundamental problem with your experiment.
[Response: You are exactly right. The greenhouse effect needs TWO ingredients: a lapse rate which makes temperature aloft colder than temperature at the ground, and infrared opacity of the greenhouse gas, which allows the infrared radiation to escape only from the higher altitude (colder) parts of the atmosphere. –raypierre]
Comment 139 was an answer to Ocean.
Thanks Gar Lipow :)
Thanks Joel Shore and to all. This has been quite educational for me. And I appreciate it being called “the greenhouse box experiment.” It sounds so sophisticated. I just wish there was a simple demo I could do with my students to illustrate the “greenhouse effect” other than using glass or plastic as analogies…
[Response: I mentioned some suggestions for additional reading as a comment to your original query up above. There really isn’t any feasible laboratory analogue making use of CO2, but it would be possible to do a reasonable “glass box” analogue which used the infrared absorbing properties of glass — the trick is, you’d have to do the experiment entirely in a vacuum, with the glass box evacuated, but also under a bell jar so that the glass surface couldn’t lose heat from its top by turbulent convection. Interestingly, Fourier says that the experiment ‘would work’ (he uses the subjunctive in the original) in a vacuum, but it sounds like it hadn’t actually been done at the time, and I’m not aware that it’s been done since either. Now, the usual glass box experiment done in air is not entirely pointless either. The essence of the greenhouse effect is that the atmosphere inhibits energy loss (to space) so that for a given rate of solar energy input, the temperature of the surface has to be greater in order to allow the necessary amount of heat to be lost per unit time. Putting a glass layer over a box has a similar effect, except that it inhibits turbulent heat loss rather than infrared heat loss. This is the sense in which Fourier used the analogue in his original paper. One can also illustrate the principal of energy balance with a bucket having a hole in the bottom, put under a running faucet. The input water flow == solar input. The output flow through the hole == infrared loss. The level of water in the bucket == temperature. It’s a very close analogy, also used by Fourier. Apropos of that, Spencer Weart’s book is good, but the only thing I hold against him is that he didn’t seem to understand what Fourier actually did; I don’t think he even read Fourier’s paper. –raypierre]
Re #67. Alexi Tekhasski said about the Stainforth article
“Finally, runs that show a drift in Tg greater than 0.02 Kyr21 in the last eight years of the control are judged to be unstable and are also removed from this analysis.”
Also, from the same source: http://climateprediction.net/science/secondresults.php
“Most models still maintain a temperature of between 13 and 14 °C, however some get colder – these are not stable and the heat flux calculated in phase 1 was not correct to keep the model in balance.”
These statements pretty much nullify results of the whole article IMO – it looks like the data were filtered in favor of desireable result – GW.’
———————————————————————
I should point out that in the ClimatePrediction models used, the first two of their three phases were hindcast, control phases using pre-industrial CO2 levels. These phases were included in order to test the stability of each model’s particular set of parameters before increasing CO2 in phase 3. A small number of models showed a slight drop in temperatures, particularly in phase 2. Of course these models had to be excluded from the final study – they did not represent pre-industrial climate accurately! I do not think many (perhaps even any) sets of parameters produced a temperature rise during these control phases. If any had done so, they would have had to be eliminated from the study for the same reason.
The ensemble of models included in the study therefore consisted of those shown to reliably represent pre-industrial climate. Had this not been the case, criticism regarding the reliability of the ensemble used to show what may happen in the future would indeed have been justified.
Thanks very much Gavin for response on #133. Regarding #138 do any of the climate scientists (apologies Richard if you also are a climate scientist) want to comment on Richard’s suggested figure of an independent 3-5 degree C rise by 2100 just from a reduction in aerosols.
usually a publication date is announced in a press release, usually the press also receives an embargoed copy upon request so the newspapers can write about the paper that has appeared in the journal
usually
Re: 138
The human generated aerosols that have been selectively eliminated — CFCs — don’t have climate sensitivities anywhere near the order of 3-5C.
Re. 138
“The human generated aerosols that have been selectively eliminated — CFCs”
Could you please be a little more specific? Do you mean simulations isolating sulfate and nitrate aerosols or do you mean simulations isolating CFCs (Chlorofluorocarbons), a relatively unimportant greenhouse gas-due to its relatively extremely low concentrations in the atmosphere?).
Thank you so much Raypierre for your response to my box experiment[#119]. I would appreciate it if in the future you would have a thread about basic stuff like that. I am a scientist but not a climate scientist; and climate science has LOTS of independent factors and counter-intuitive principles in it. So I, for one, have a steep learning curve here. But maybe this would be helpful for other readers as well? Anyway, thanks again, that was very helpful! And I will happily check out the lit you suggest.
Teacher Ocean,
I’m a former science teacher and thin-films engineer. One problem you will have with your design is convection. Your greenhouse [assuming glass] is like a fat, dual-pained window. Gasses conduct heat between two plates of glass at a rate proportional to the speed of the gas, and less efficiently with greater distance between the plates [the effect also varies in different pressure regimes, it is very efficient at low vaccuum, because gas molecules can travel from plate to plate without being intercepted. At higher vaccuum regimes it becomes less efficient with fewer molecules impinging.]. As Temp is proportional to K.E. = 1/2mv^2, heavier gasses travel slower at the same temp as lighter gasses. Hydrogen cools the best because it’s the fastest [2 g/mol], Air slower [~29 g/mol], Argon slower [40 g/mol], CO2 slower [44 g/mol], and various noble gasses like Radon slower still. As a result, you may have noticed that dual-pained windows filled with Argon are typicaaly thinner than those filled with Air.
So, if you have two identical glass greenhouses with thermally isolated mercury thermometers at equilibrium in the sunlight [One with Air at Press=P, and the 2nd w/ CO2 at Press=P], and you close the blinds – you will see the thermometer in the CO2 greenhouse retain its temperature longer – not because of any ‘global warming’ type effect, but simply because Air conducts heat to the walls of the greenhouse better than Air does. By the way, 1.5 lbs of dry ice that you used should make about 15 cubic feet of CO2 gas at STP, a lot more than you need in your class.
I believe the ‘greenhouse effect’ normally associated with CO2 is more of a diffusion problem, rather than a heat sink problem. [i.e. – like a blanket, not a hot water bottle.] It’s not that CO2 ‘holds’ heat like a bank. The CO2 emits any heat absorbed virtually instantaneously – but some portion of the emitted thermal energy goes the ‘wrong’ direction. IR heat energy is absorbed and reemitted numerous times, Since the biggest greehouse gas, by far, is water, why not do your demonstration comparing the transmission of IR heat through liquid water and Air. Compare an empty and a water filled glass container. A thermometer on the other side of the water should take longer to warm up than one on the other side of an empty container, because the heat is having trouble finding its way through the water. [Of course, this experiment is faulty b/c the thermal mass of the water is acting like a hot water bottle…] Conceptually, however, you can show your students the diffusion effect associated with CO2 and H2O, that the heat will eventually work its way out of the water into its surroundings [like heat trapped in a hot rock], and that climate science is a complex endeavor because the CO2 signal is not the sole factor out there.
Teacher Ocean,
I’m a former science teacher and thin-films engineer. One problem you will have with your design is convection. Your greenhouse [assuming glass] is like a fat, dual-pained window. Gasses conduct heat between two plates of glass at a rate proportional to the speed of the gas, and less efficiently with greater distance between the plates [the effect also varies in different pressure regimes, it is very efficient at low vaccuum, because gas molecules can travel from plate to plate without being intercepted. At higher vaccuum regimes it becomes less efficient with fewer molecules impinging.]. As Temp is proportional to K.E. = 1/2mv^2, heavier gasses travel slower at the same temp as lighter gasses. Hydrogen cools the best because it’s the fastest [2 g/mol], Air slower [~29 g/mol], Argon slower [40 g/mol], CO2 slower [44 g/mol], and various noble gasses like Radon slower still. As a result, you may have noticed that dual-pained windows filled with Argon are typically thinner than those filled with Air.
So, if you have two identical glass greenhouses with thermally isolated mercury thermometers at equilibrium in the sunlight [One with Air at Press=P, and the 2nd w/ CO2 at Press=P], and you close the blinds – you will see the thermometer in the CO2 greenhouse retain its temperature longer – not because of any ‘global warming’ type effect, but simply because Air conducts heat to the walls of the greenhouse better than CO2 does. By the way, 1.5 lbs of dry ice that you used should make about 15 cubic feet of CO2 gas at STP, a lot more than you need in your class.
I believe the ‘greenhouse effect’ normally associated with CO2 is more of a diffusion problem, rather than a heat sink problem. [i.e. – like a blanket, not a hot water bottle.] It’s not that CO2 ‘holds’ heat like a bank. The CO2 emits any heat absorbed virtually instantaneously – but some portion of the emitted thermal energy goes the ‘wrong’ direction. Some small portion of the IR heat energy is absorbed and reemitted numerous times by GHG, and the heat ‘loses its way’ – effectively running around in circles in the atmosphere [‘trapping’ it].
Since the biggest greenhouse gas, by far, is water, why not do your demonstration comparing the transmission of IR heat through liquid water and Air. Compare an empty and a water filled glass container. A thermometer on the other side of the water should take longer to warm up than one on the other side of an empty container, because the heat is having trouble finding its way through the water. [Of course, this experiment is faulty b/c the thermal mass of the water is acting like a hot water bottle…] Conceptually, however, you can show your students the diffusion effect associated with CO2 and H2O, that the heat will eventually work its way out of the water into its surroundings [like heat trapped in a hot rock], and that climate science is a complex endeavor because the CO2 signal is not the sole factor out there [although one of the only man-made ones – others: water, aerosols, sun, et al]
[Response: As an addendum to that, though it isn’t possible to make an infrared greenhouse effect analogue easily in the laboratory, it is easy to illustrate the infrared opacity of various materials. For $50 or so, you can get a broadband infrared sensor which you can attach to a volt-ohmeter, whose output will be proporional to the infrared flux. Point it at a candle. It will register high. Put a pane of glass in the refrigerator for a while, then interpose that between the candle and the sensor. It will register low, since the glass is opaque to infrared, and by Kirchoff’s law, also emits like a black-body — at its cold temperature, rather than the hotter temperature of the candle flame. Go out on a hot day. Point the sensor at the asphalt parking lot. Registers high. Point the sensor at the sky, not toward the sun. You’ll see some sign of infrared, which is the downward infrared from the atmosphere into the ground. Now, the best thing would be to be able to take your class into space and point your $50 sensor at the Earth from the Space Station, so you could see that the radiation going out is like a blackbody at 255K instead of the actual surface temperature of the Earth. For that matter, in space, there would be plenty of vacuum to do the glass box experiment properly! –raypierre]