Sunday
Dec022012
by 
Bishop Hill
Bishop Hill Quantifying Uncertainties in Climate Science
Dec 2, 2012 Climate: Statistics Another date for your diaries - the Royal Met Soc's meeting on uncertainty in climate science.
Climate models produce different projections of future climate change under identical pathways of future greenhouse gases. This meeting will highlight recent studies that have attempted to quantify those uncertainties using different approaches.
Programme:
Time No. Presenting author Title 16:40 Prof Reto Knutti, (ETH Zürich) Projection uncertainties: The multi model perspective. 17:10 Dr Paul Williams, University of Reading. Climate models: The importance of being stochastic. 14:10 Dr Jonty Rougier, University of Bristol Background and philosophy 14:40 Dr David Sexton, UK Met Office UK climate projections. 15:10 Dr Tamsin Edwards, University of Bristol Palaeo-constraints on climate sensitivity. 16:10 Dr Lindsay Lee, University of Leeds Constraining aerosol models.
Details here.
Reader Comments (224)
Tamsin's posts are in jargon whereas the questions directed at her are in non-jargon.
I've just started reading this thread. I'd like to congratulate Tasmin Edwards for her excellent comments on 3 December, and thank her for standing up for integrity.
Structural and other uncertainties are all very well – interesting and challenging as they may be to define and estimate – but the elephant in the room for me might be regarded as a structural certainty.
I refer to the device whereby the patently internal and patently continuous (albeit irregular in space and time) growth of CO2 in the troposphere from the bottom up is modelled as instantaneous jumps in ‘external’ forcing at the top of the model atmosphere. The more CO2, the more 'forcing'.
This peculiar device would not have been entertained (as a possible thought experiment) back say in 1940. In 1940, in a reputable textbook, I note the author’s quietly confident words as follows:
'Hence, for this reason, as well as for the one given above, either doubling or halving the present amount of carbon dioxide could alter but little the total amount of radiation actually absorbed by the atmosphere, and therefore, seemingly could not appreciably change the average temperature of the earth, or be at all effective in the production of marked climatic changes.’
‘this reason’ refers to water vapour having ‘high coefficients of absorption in substantially the same regions where carbon dioxide is effective, leaves but little radiation for the latter to take up’
‘the one given above’ refers to experimental evidence that, for example, a 50cm column of CO2 ‘absorbs quite as completely as does a column 200cm long a the same density’, and noting that the ‘the amount of carbon dioxide in the atmosphere is equivalent to a column of the pure gas, at ordinary room temperature and atmospheric pressure, of, roughly, 250cm in length.’, the author concludes ‘doubling or halving the amount of carbon dioxide in the atmosphere …would not appreciably affect the total amount of radiation actually absorbed by it, whether of terrestrial or of solar origin…’
Source:‘Physics of the Air’, by W.J.Humphreys (1940, McGraw-Hill) – esp. pages 584-586.
As I understand it, the ‘external forcing’ methods just keep adding a step change in the energy budget at the toa for every projected change in CO2 levels (e.g. in monthly or annual increments), thereby sort of begging a question the naïve observer might hope that the models might help answer.
I understand also that the early results of using this forcing wheeze were so implausible that a fudge factor labelled 'flux adjustment' was added for ad hoc changes to energy transfers at the bottom of the atmosphere until 'proper' behaviour was obtained. I believe these are no longer required in current models, but whether that is due to their having been effectively incorporated in automatic numerical analysis algorithms or to improved model physics, I don't know. Some more background can be found here: http://theresilientearth.com/?q=content/secrets-climate-modeling-mystics.
In the 1940 book, experiment and direct physics calculations are the order of the day. In our time, the outputs of models are.
shub @ Dec 3, 2012 at 8:46 PM
The Australian government, and the now politicised government departments, listen to whoever tell them what they want to hear. We should be able to rely on the Australian Treasury Department to be the last line of defence. Treasury should be honest, impartial, objective, but that is not the case any more. Following are a coupe of examples of their advocacy for an Australia carbon tax and ETS:
Regarding Treasury’s assumptions used in their modelling of the carbon tax and ETS see Box 3.1: http://archive.treasury.gov.au/carbonpricemodelling/content/report/07chapter3.asp :
That’s just part of what treasury has done to justify Australia’s carbon tax.
Here is another extract from Treasury’s propaganda:
http://archive.treasury.gov.au/carbonpricemodelling/content/overview/page16.asp
That is just some of what Treasury is saying to justify the carbon tax and ETS.
Dec 5, 2012 at 10:10 AM | John Shade
"I understand also that the early results of using this forcing wheeze were so implausible that a fudge factor labelled 'flux adjustment' was added for ad hoc changes to energy transfers at the bottom of the atmosphere"
When I was modelling nearly 20 years ago one of the main issues were the tropics got too hot and the poles too cold so a quite large "Flux Correction" was hard coded in to move heat polewards. It is no longer used as I think the Ocean model used and parameterising to fit moves the acceptable amount of heat now.
@ Rob Burton
So its not physics, its a heuristic?
Dec 5, 2012 at 9:04 AM | Dolphinhead
"one of the problems with blogland is that people can and do just walk away"
But hasn't there been a big effort by the science community in 'communicating the science' this year. Hopefully unanswered questions could at least be replaced with a 'I don't know' to support the effort...
Dec 5, 2012 at 10:55 AM | ThinkingScientist
Doesn't tweaking parameters in general count as heuristic. Actually it sounds like my general approach to problem solving, though you want to back up your intuition with evidence. (From Wikipedia Examples of this method include using a rule of thumb, an educated guess, an intuitive judgment, or common sense.)
In reply to your list of questions to Richard have you looked at the www.climateprediction.net 'experiment'. I doesn't really answer your questions but as a pretty good and succinct explanation of what they are doing. That project in particular to me seems to be curve fitting taken to an extreme.
Just quickly checking in to say I am reading your comments and I do intend to reply more if I possibly can. I was up until 3am working on something that had to be finished and today I have to do something for Doug (McNeall) that's already a week late. It's the fun of the pre-Christmas rush...
ThinkingScientist and Alex - it's starting to sound like an All Models Are Wrong comment thread :) (sorry I haven't replied to you there Alex). There's a whole load of interesting stuff here to discuss - I need to read up a bit to compare the details of history matching with similar (differently named) things in our field - think perhaps I should consider a proper blog post and discussion rather than have it here though.
OK, got to dash, late for a meeting, again...!
Rob Burton
'But hasn't there been a big effort by the science community in 'communicating the science' this year. Hopefully unanswered questions could at least be replaced with a 'I don't know' to support the effort..'
Regrettably, the science community's idea of communicating the science is to limit themselves to the arenas where they control the dialogue. Do they not see that their unwillingness to enter open debate is compelling evidence that their arguments cannot stand up to proper scrutiny. I suspect that when put to the test there would be an awful lot of 'I don't know's' from the science community. The irony is that they would regard that as a sign of weakness whereas I would regard it as a sign of strength. Good science has the strength to admit its limits.
Tamsin:
And if you can't, we still appreciate what you've already given on this thread (hugs included, though I may be biased there). As I've just written elsewhere, nirvana isn't a viable option, whether in the real world of energy policy, with powerful vested interests already in place, or in blog-based discussions of the scientific underpinnings (or otherwise) of any level of carbon concern.
Everyone has a different view of what is most important but I continue to think Nic Lewis's post of 10:00 PM last night is one of the most important I've seen on Bishop Hill. On that two further questions in different directions:
1. Of course the code and data necessary to replicate published results should be made available the moment results are, for any kind of model, however simple or complex. But especially the complex ones, because the likelihood of this kind of error greatly increases in that case, as Nic says. So what are the practical implications of that for the big beasts of the IPCC AGW forest - the GCMs - and the availability of the supercomputers they take weeks or months to run on?
2. If a software error had given rise to an estimate of sensitivity of .7ºC rather than than the correct value (for the observed data) of 1.6ºC would we ever have heard about the .7ºC figure? Somehow I doubt it. What does this tell us about the field? (And of course I applaud the programmers in this case for admitting and correcting the error that made the answer 2.5ºC. As the apostle James says, we err in many ways in what we code. Or something like that.)
Er ... what exactly has been done on this thread?
Shub I think it is called foreplay
Dec 5, 2012 at 3:28 AM | Hilary Ostrov
Hi Hilary,
Sorry, in my haste I genuinely thought I was addressing the most important point in advising Nic on how to get his voice heard (following Don's comments about "silencing" earlier). And as you probably also saw on Twitter last night, I also got distracted by supporting Barry Woods in telling somebody that use of "Denier" is not acceptable!
Anyway...
Nic, I agree with your calculations, if the input numbers are correct - but we need to be careful there. Your AF "best estimate" number does reflect more recent literature on aerosol forcing, of course there are still fairly large uncertainties, but I think it's fair to say that the evidence is pointing strongly towards a net anthropogenic RF & AF that is larger that we assessed in AR4. The big question mark is over ocean heat uptake, which cannot really be measured directly, it has to be inferred from changes in ocean heat content, which again has large uncertainties. Previously people have used 0.6 W/m^2 for ocean heat uptake, but as you say it probably isn't as large as that. I'm not convinced we can estimate it to 2 significant figures like you did though. However I couldn't claim to be up to speed with the literature on that anyway - I just know it's a challenging area. But, yes, the upward revision of net anthropogenic RF implied by recent aerosol measurements does have very interesting implications for climate sensitivity. It would be good to be able to use observations to constrain it better.
No doubt you gave the D&A chapter authors plenty to think about in your review coments..... :-)
This is making me wish all the more that I could attend the RMS meeting and hear Tamsin's talk on palaeo-constraints on climate sensitivity! (Tamsin, please can I see your slides afterwards? BTW good to see you are imminently going to deliver to Doug.... :-) )
Dec 5, 2012 at 8:53 AM | ThinkingScientist
Hi
Great questions, but unfortunately asking 7 questions at once makes it harder for me to find time to answer. We're back in the situation where lots of people fire questions or comments, sometimes with several questions from one person at the same time, which in themselves are good questions, but it is difficult for one person to keep up - and all this does is let people say "you didn't answer my question" later. This leads to the risk of thinking that it is far better to not get involved in the conversation in the first place. I will try to answer in due course, but it will take a while I'm afraid - sorry!
Hi Richard,
Thanks for your reply. One answer at a time is fine...
@Tamsin
Like others I appreciate any of the people who come here who are actively working on these problems and bother to engage with this dialogue.
RE: "ThinkingScientist and Alex - it's starting to sound like an All Models Are Wrong comment thread "
I would say gain, I prefer the full quote: "All models are wrong, but some are useful".
Regarding Rob Burton's comment about heuristics, modfiying a physical parameter is ok for example if your model is a solver and includes a description of the data misfit in its objective function. It becomes a heuristic when it no longer is a physically reasonable property, or when it is not a proper property at all. In a reservoir model, for example, the ultimate "fudge factor" is the permeability multiplier. IF you then multiply permeabilities on an aread by area basis you are fitting the data, but your physical model must be broken.
ThinkingScientist: That's a really helpful example to this outsider to reservoir modelling. I'm familiar with the idea of overfitting and use of something like the Akaike Information Criterion in trying to build robust ARMA or ARIMA models that will do predict usefully in finance. But of course every real world problem has a different shape, dependent on the physics.
Sorry by the way not to respond to your long and (I agree with Paul Matthews, not for the first time) instructive post addressed to me early yesterday morning. I was indeed up early - because I had a lot else to do, away from Bishop Hill. It's a minor example of something we're all used to on the blog - but sometimes we interpret negatively. I appreciate the way you, Tamsin and Richard are sticking at it. And (as I've made clear) the conversation about aerosols, clouds and reflectivity that's sprung up really grabs me. But I appreciated your response yesterday morning. And Josh's, come to that :)
When someone doesn't reply its easy to interpret motive into what may a simple or innocent oversight. Like you have a life or job to get on with.
For example, I really wanted to get responses from Richard Betts about my questions from March, but if he hasn't seen them he can't reply to them. Now I know he has seen them, so I will wait until he has time to reply (or, if like me, maybe he just won't find time).
Interestingly those questions came up on a very similar thread theme and I believe Tamsin was on that thread too, so its a generally interesting topic.
I also find the posts by Nic Lewis particularly interesting and useful.
Further down I'll give my thoughts on Thinkingscientist's questions. But first a short comment on (coding-)errors in GCMs (as asked for by Nic Lewis and Richard Drake). It's not really a satisfactory answer, but a good starting point (from my point of view). This paper deals with so called "defects" in climate models and compares it to similar large projects.
Are there bugs in the GCMs? A former colleague basically collected them. Were they big? Yes, but not in the sense that correcting them changed the results significantly (compared to e.g. the tuning). Are codes available? I think, Judith Curry would say no. I say, as far as I know, you cannot always download them directly, but you should be able to obtain the code. (and I know what kind of reaction that provokes). Anyway, the bigger problem is certainly that only few people have the access to computers big enough to efficiently run the code. That is especially a problem if the code or input data is optimised for specific machines. How to solve that? No idea.
On Nic Lewis quote from the Ring et al paper. Do I think the change in CS insignificant? No. But I am more astonished that also the made choices (on input-data) produce relatively prominent changes.
Some thoughts from someone who is not really a modeler. But first links to Judith Curry, Lucia and to the paper they both refer to (it’s open-access). However, some of the references the paper gives may be of even larger interest.
Re:1. General circulation models for climate studies are not generally initialized. The tuning is performed based on agreement with certain defined metrics – e.g. matching a certain global mean temperature or global radiation metrics (see the linked paper above). For simpler models or rather in the original development stage such a "metric" may be that patterns and energetics of atmospheric planetary waves are matched in principle.
Re:2. In case of initialized runs – e.g. for the decadal hindcasts that are going to be used for the next IPCC assessment report – and in my understanding (and I’m way out of my depth) there is of course a dependence on the initial conditions. However, these initial conditions do not represent a current climate state but a shorter term average (maybe day to season).
Re:3. Not really on the question. Mentioned initialized hindcast experiments suggest that it makes a huge difference whether we initialize only the atmosphere, the atmosphere and the sea ice, or even add the ocean. However, ocean and sea ice initializations are problematic. For the ocean it is my understanding that the available datasets notably differ (e.g., dx.doi.org/10.1007/s00382-012-1310-7 ($), again out of my depth and only what I learned in talking to former colleagues at the Max Planck Institute in Hamburg).
Re:theother3. James Annan notes at climate-lab-book :
I would have only noted the adjoint or tangent-lineaer approaches which indeed are pursued but not always successfully.
Re:4. If I understand it correctly, the QUMP project where Tamsin worked modified 31 parameters. And Yokohata et al. (2012) also report on an ensemble with 31 perturbed parameters. The paper above mentions 8 commonly altered cloud-parameters and shows changes for 5. However, these have nothing (well not directly) to do with the referred aerosols. With respect to them the “increases” base on observations or – rather – estimates which in turn rely on some observations. Parameterizations become relevant in how they and their temporal changes are handled / interpreted / acted on by the coded model-physics (certainly especially the radiation-code). But this also bases to certain extents on observations, laboratory experiments and theoretical consideration which are not necessarily climate specific. The number of such, potentially time-varying, boundary conditions depends on the complexity of the GCM. In simple cases we may only prescribe constants for e.g. the solar constants, carbon dioxide concentration and ozone concentration.
On aerosols, the AEROCOM project may be of interest.
Re:5. I have to admit, I’m not sure what is meant. So I have to ask: do you mean a GCM in which CO2 doesn’t interact with thermal radiation, or a GCM where such interaction has no effect on water vapor, or without CO2, or …? Maybe the QUMP-project had some parameter-settings approaching such a state?
Re:6. First, water vapour feedback is “not a built-in assumption” as Gavin Schmidt put it some years ago @realclimate. Can we tune a model fitting the observations well but with weak or negative water vapour feedback? I don’t know. Maybe some of the perturbed physics ensembles or the emulator-projects looked for such a model-world, but I can’t find it right now.
Only slightly on the question: Ingram proposes in various papers to take a different route to the “water vapour“ feedbacks. While I can’t find author-versions, these are his slides summarizing it.
Re:7. As shown by the linked paper the tuned absolute mean temperature is … arbitrary. However, from my point of view, the simulated evolution is reasonable and in principle reliable. Thus, no update is necessary in this respect.
For decadal – or seasonal – hindcast experiments, observational estimates are assimilated for initialization in the beginning, but afterwards the model runs free. The GCM is not tuned to match the initial state – which would be too costly – or to match the trend exactly but rather to match the metrics (see Re:1.). A note: Hindcast does not necessarily refer to the same thing. So, if you used it in the sense of Lucia Liljegren, I likely used it differently. That is, I don’t see the historical CMIP-simulations for the 20th century as hindcasts. In turn, the simulations into the future are not meant to be predictions or forecasts. They try to project the climate and its dynamics under the applied scenarios. These scenarios mean to represent "best estimates" on the future evolution of those boundary or forcing conditions assumed to be relevant.
That further implies – in my understanding – if we note that our observations contained an error (and this applies to all observations that may influence the development of the model or the simulations), we are going to have to reassess the quality of the model under the prior tuning (see for example the linked paper or the last section quoted by Lucia) against this changed observations.
Later tuning would follow the same procedure as the first tuning of the model (as for example outlined in the linked paper).
I think J. Rougier prefers the term simulators for our models and I do agree.
Sorry for too much jargon. And I hope none of the scientists visiting here are going to stone me for massive mistakes.
A.Bothe:
Only if you're lucky enough not to be stoned first for the jargon. It's equal opportunities stoning around here - we have a policy of not discriminating against the totally ignorant. :)
But I speak in jest, I'm almost certain. I'm very grateful for the response.
Me neither. Except to make use of Moore's Law (while it still lasts) to get to the point where every published result of a GCM, not least in IPCC assessment reports, can be replicated on an affordable machine by any intelligent third party. That means the complexity of models will have to be capped. If it's good enough for carbon ...
Dec 5, 2012 at 1:15 PM | Richard Betts
"Nic, I agree with your calculations, if the input numbers are correct - but we need to be careful there."
Richard,
Thanks for responding on this. I agree, but I did stick pretty closely to mainstream estimates, so the main question seems to me how much uncertainty there is.
I think the estimated standard error (SE) in my total adjusted aerosol figure is around 0.15 W/m^2. Allowing for recent mainstream uncertainty in other anthropogenic forcings, that brings the SE for total anthropogenic forcings to 0.25 W/m^2. I can't cite the source for these figures - you can no doubt guess why. Likewise for the HadCRUT4 trend error.
I was only estimating the change in ocean heat uptake to one and a half significant figures - 0.45 looked closer to me than 0.4 or 0.5, but it makes little difference. Let's guess the SE for this figure as 0.2 W/m^2. Levitus would no doubt claim it is smaller.
Then add a further, say, 0.15 W/m^2 for uncertainty in small differences in the internal state of the climate system that are not reflected in either surface temperature or ocean heat uptake, and in solar and volcanic forcing (low in both decades). That takes the total SE for {change in forcing - change in ocean heat uptake} to 0.35W/m^2 (adding the squares of the SEs since the errors are independent).
So my +/- 1SE 'likely' range for {change in forcing - change in ocean heat uptake} is 1.3 to 2.0 W/m^2 - or +/1 21% of the central value.
The corresponding 1 SE range for the HadCRUT4 1880-2010 temperature trend, as a fraction of its value, is under half as large - about 10% of its central value.
Since the temperature error is independent of that in the change in forcing - ocean heat uptake, the probability distribution for climate sensitivity (the ratio of the change in forcing - ocean heat uptake to the change in global temperature) can be obtained by randomly simulating a large number of values from the distributions (assumed normal) of the numerator and denominator, as in Gregory et al. 2002. Taking 100,000 of each and computing the cumulative probability distribution for climate sensitivity (done in an instant with R) gives a 5 to 95% confidence range for climate sensitivity of 1.2 to 2.6 C. Quite well constrained, with the top of the 95% bound below the IPCC's central estimate of 3 C.
Anything that you or your colleagues disagree with here?
" No doubt you gave the D&A chapter authors plenty to think about in your review comments..... :-)"
Yes; I expect that they wish I would go away, particularly as I am also taking them to task about what Ch. 10 says (and doesn't say) about Bayesian inference. :-)
BTW, thanks for your response about ensuring my comments on Ch. 7 and 8 were noted. I did give them full references, so that should be OK.
O. Bothe wrote:
That's a miracle, or the mother of all incorrect statements.
O.Bothe:
"Were [bugs in GCMs] big? Yes, but not in the sense that correcting them changed the results significantly (compared to e.g. the tuning)."
I was also under the impression that bugs were generally not very significant, but see Nic Lewis' observation. Perhaps this is the exception which tests the rule.
[I add a personal observation that this sort of bug appears to be a simple cut-and-paste error such as I have often encountered in my professional career. They are often insignificant in effect, perhaps only noticeable under unusual circumstances. But every once in a while it's a doozy...and seeing that fixing this bug changed the climate sensitivity from 2.5 K/doubling to 1.6, this is one of those occasions.]
O. Bothe,
Thank you very much for taking the time to give the responses to my questions. I will need to digest some of these answers, but they give me food for thought.
Richard, if you are reading, this doesn't let you off the hook!
Tamsim,
I downloaded and scan/read the Bayesian emulation of a reservoir simulator paper you linked to. Interesting. However, I did pick up on a point that links quite nicely to the point I was trying to make above, which Richard Drake picked up. It is the issue of non-physical properties.
The paper you linked to gives quite a nice example of this: firstly, the emulator is based on multipliers of the model properties in the cells. This is not unusual in reservoir engineering as I understand it. More unusually in this example the permeability is not an important factor, but the porosity (and also the critical water saturation) are found to be the critical parameters. The coarse model results led to the following statement:
"Closer investigation revealed that this set can be well-approximated by the restriction that φ should be constrained to the sub-interval [0.5, 0.79]."
Remember these are parameter multipliers. This says that that porosity in the model needs to be "corrected" by a factor in the range 0.5 - 0.79. That is a huge multiplier for a property that is probably quite well calibrated by observations (well logs). Now, in this example we can state that this is therefore potentially inconsistent with observation and this might lead us to deduce that there is another inherent problem with the model which we failed to take into account, but in a climate model, where the factor in question may be unknown, we can be creating an artificial change that appears to give a good match to the overall prediction, but we are unaware that the parameter is non-physical and may actually indicate some major physical shortcoming of the model.
A further point is the assumption of multi-scale, ie that a coarse or simplified model run is linearly related in some way to a finescale run. This may be true in some models, but it may also be false. For example, a thin boundary impermeable layer in a reservoir model can completely change the flow perfomance of the reservoir.
Today on Twitter I pointed out to Ed Hawkins how climate science is the one science that doesn't want to male use how all the time possibly given for free by hundreds of volunteers. So climate scientists refuse to share data (and code) long after publication has been achieved.
And now we know from Richard Betts that the BH comment area might get filled with too many questions.
I feel like there's a lot of unappreciated love around here... :)
Maurizio, to be fair, the two problems reflect two different kinds of love or lack thereof. Refusing to "share data (and code) long after publication has been achieved" is inexcusable. This betrays a lack of love for science and for the world, which has become so dependent on the impartial advice of climate science (as it hopes) for vastly costly policy decisions. If the world doesn't now love climate scientists as much as it might, those responsible for withholding code and data have a great deal to answer for. This is independent of the quality of any one individual or group that wants to have a go with the code and data or indeed their 'love' or otherwise. Perhaps some wanting the means to check and replicate published results really are evil fossil fuel interests. It makes not a scrap of a difference. The data and code should be open.
But if the questions on one blog are too many for one particular climate scientist, in one exciting and busy week, that indeed shows interest from outside the field but the lack of immediate answers doesn't imply moral peril for the scientist. He may judge some questioners as more worthy of attention than others. This is just normal life, especially with the internet and pseudonymity allowing so many to ask so much so quickly, some of them people one cannot possibly know.
Richard Betts seems to have made Nic Lewis priority so far. I personally think that's a good call. Unlike Hilary Ostrov I didn't interpret one of Richard's terse answers as dismissive of Nic; quite the reverse, I thought it showed a busy man was really trying to help. Nic seemed to feel the same.
The importance of Nic's detailed questioning of the IPCC's range of likely sensitivity to a doubling of CO2 cannot for me be understated. Anyone could ask those questions (given enough skill in maths and time to read the papers) but it happens to be Nic. In the most important specific area raised Dr Betts has clearly been paying attention. I think that is love, from both men, though it's true that one is doing all the work on his own account and that is an inspiring thing indeed.
[O/T]
[O/T]
[O/T]
Dec 5, 2012 at 8:41 PM | Richard Drake
Richard, with all due respect, I would appreciate it if you would consider refraining from "reframing" the observations of others in your interpretations.
The point of my observation was definitely not that RB's "answer was terse" - in fact, I found his response far from "terse" (please see below). But rather that RB had failed to even acknowledge Nic's lengthy response to RB's questions.
YMMV, but to my mind 201 words on one bureaucratic procedure does not fall within my definition of "terse".
Which reminds me that one of the observations I had actually deleted from my observations was to the effect that the extensiveness of this particular paragraph contained the implicit suggestion that the intended recipient might lack the intelligence or competence to have read the instructions prior to submitting any review comments. Either that or the actual instructions provided by the TSU lacked clarity. You may take your pick :-)
In the event, as Nic (who I am inclined to think is far more forgiving and diplomatic than I!) subsequently indicated [Dec 5, 2012 at 3:57 PM]:
Hilary: sorry if 'terse' was wrong. I thought Richard's key move was saying
and I generally thought that he and Nic should be left to get on with it, without much commentary. With that in mind, .
Richard Drake - the love am talking about is by outsiders for climate science not by climate scientists.
Can you imagine, any claim they make would be scrutinized hard and by many people if only they'd let the data out. This would result in climate science rapidly becoming extremely strong because it would be composed of statements capable of surviving the hardest of scrutiny. We could easily know more about climate than anything else.
Instead no, those in charge don't see any of that, grey employees the lot, sinking the Met Office and the Royal Society under the weight of advice and relevance, that is money. Fools with PhDs.
100% agree Maurizio - and privileged to do so.
Dec 5, 2012 at 9:56 PM | Hilary Ostrov
Hi Hilary
You have a remarkable ability to read more into what I write in a quick blog post that I could ever possibly imagine! I suppose I should be flattered that my hastily-written words are analysed so deeply, and that my tweets are stored away by someone so that they can be quoted back at me as examples of not answering questions (or whatever it was). In case you are in any doubt, here is how I contribute to a busy but interesting thread on Bishop Hill when I have little time:
1. I skim through the latest set of posts
2. Those that catch my eye, for whatever reason (eg: it's by someone I know, or seems to be asking an interesting question) I quickly make a response, writing it simply as the words come to me without agonising over them
3. When I have used up whatever time I had available, which may be just a few minutes, I go off and do whatever else it is I need to do.
My words here really do not deserve in-depth analysis on the subtleties of my motivation or nuances of how I view the person I am talking to. They are blog posts, not scientific papers or even newspaper articles. I find interesting and hopefully others do too, but please do not over-think it:-)
If I have "failed to acknowledge" somebody, the chances are that I either didn't quite have time, or forgot. Nothing sinister. Certainly no deliberate attempt to avoid answering questions.
Dec 5, 2012 at 3:57 PM | Nic Lewis
Hi Nic, thanks for your response. I think I'd better read your paper in J. Clim, which I have to admit I have not managed to read yet. NB Hilary, please note that this is not dismissing Nic - it is late at night here, I've just got in from the pub and I have quickly checked in here before going to bed. I am lecturing and writing all day tomorrow so am unlikely to have much time to blog, but I acknowledge Nic's response and will read his arguments as published in the scientific literature.
Goodnight all! :-)
Dec 5, 2012 at 1:15 PM | Richard Betts
Thanks for your reply, Richard.
I fully appreciate that you thought you were addressing "the most important point ..." The sheer number of your words on that matter (which, as Nic has subsequently confirmed, were really not necessary!) made that perfectly clear ... to me at least. To Richard Drake, it would seem, perhaps not so much ;-)
Perhaps you might consider taking a leaf out of Tamsin's posting book. One of the things I greatly appreciate about the pattern of Tamsin's participation here is that she rarely (if ever) fails to acknowledge questions and/or observations of others that she might not have time to answer at the time, and clearly indicates her intention to do so (or not, as the rare case might be!)
And sorry, I suppose I don't pay as much attention to twitter as I could, so I definitely missed the evidence of your "distraction"! I find twitter a useful vehicle for announcing my own new posts - and for becoming aware of and sharing notice of new posts from others that I find interesting.
As you may (or may not!) have noticed, it is far from my preferred mode of "communication" and engagement. To my mind, as a conversation tool, twitter is a dressed up (but very much truncated) version of IRC channels, which preceded the advent of the web by several years.
Conversations on IRC channels, with so many voices clamouring and chattering simultaneously, were quite limiting - even with a significantly higher (although I can't remember the exact number!) permissible character count per "send". If one wanted to have a serious conversation or discussion, one did so via E-mail - or by "invitation" to a behind-closed-screens "private" channel!
The other disadvantage that I see in twitter as tool for discussion is that - with a mere 140 character count per "send" at one's disposal - it almost invites what I would consider to be tantamout to "handwaving" responses such as: "Plenty of other evidence that AGW poses risks. A decade on, MBH98/99 no longer key to argument. Superseded by more recent work."
When I spotted that particular tweet, btw, I was sorely tempted to reply, "More recent work? You mean like that of Gergis et al? & surely the notion of "risk" is far from catastrophe or dangers we’ve been told to expect" But, alas, too many characters. So I didn't :-)
Oh, well, FWIW, that's the view from here!
Tamsin -
Comparing your categories of uncertainty to a perhaps more well-known description, would it be accurate to equate "parametric uncertainty" to "known unknowns" and "structural uncertainty" to "unknown unknowns"?
Dec 5, 2012 at 10:03 PM | Richard Drake
Richard, I'm not sure who died and made you the great determinator of what "commentary" should be permitted by whom and when amongst this congregation!
But next time I have some comments regarding my observations of posting patterns, perhaps I should consult you first to make sure it's OK for me to post.
And ...
Dec 6, 2012 at 12:45 AM | Richard Betts
Just saw this post of yours, which has a faintly familiar ring to it, btw! Be that as it may ...
You seem to have mistaken Richard Drake's reframing and "interpretation" for that which I had actually written [Dec 5, 2012 at 3:28 AM]
Because in my books, at least, an observation that any poster's failure to acknowledge strikes me as being "discourteous" is a far cry from accusing anyone of being "dismissive".
And if you consider my observations of the pattern of your response choices here to constitute "analysis" of any kind ... well, all I can say is that it strikes me as being your problem, not mine ... unless you've redefined "analysis" ;-)
But then again, I come from the school of "say what you mean, and mean what you say", so what do I know, eh?!
Hi Bish - I'm presuming you are doing your own moderating so I'm directing this to you.
Please can you clarify why prodding Richard Drake over his real name nonsense is considered o/t? In case it wasn't obvious I'm fed up with his continual harping on this subject and I find his rhetoric particularly misplaced on a solid thread which has largely been based on the interest and quality of a pseudonymous poster? Or is ThinkingScientist a real name?
NBY
I am also tired of discussion of real names versus pseudonyms. I cut Richard's comment on the subject and your comment too (which IIRC made no sense without Richard's).
Bish: The snips are very good ones.
Hilary Ostrov to Richard Betts:
Aha, my monstrously evil influence now extends to invading the mind of a Met Office scientist to read other posters the way I decide, not as those posters intended. Or there may have been a nugget of truth in the way I read things, perish the thought. Either way, let's get over it. Nic Lewis and ThinkingScientist seem happy, if waiting for answers, and that's surely what matters.
@NBY " is ThinkingScientist a real name?"
Of course it is :-)
Thanks Bish - I'm happy with that.
@TS - Ok - I'll keep an eye out for you! ;-)
Dec 6, 2012 at 4:03 PM | Richard Drake
You should be so lucky! Unfortunately, for you, there's no evidence of the latter in this instance. Unless, of course, "the way [you] read things" overrides the poster's very own words [Richard Betts Dec 5, 2012 at 1:15 PM]:
Clearly, it's not the first time this individual's posting choices have been influenced by what he's read from someone other than the person to whom he was responding. So, since "dismissive" was your very much mistaken "reading" of my words, while there is no evidence of any "nugget of truth", there certainly is both precedent and contemporaneous evidence of your "influence".
Although I would not go so far as to call it "monstrously evil". Not even "evil", just wrong.
Obviously you don't seem to think it matters; but I'm still not so sure that anyone has vested in you the authority to determine "what matters" - nor, for that matter, who should be commenting on what or when!
So anytime you'd like to apologize for misleading Richard B. and quite possibly others - and/or attempting to boost your own credibility by diminishing mine - will be just fine with me.
Then perhaps we can all "get over it" ... at least until the next time;-)
If you spend a few minutes reading my paper and at least the abstract of the paper published by the American Institute of Physics (cited in reference (8) in my reference [13]) you might understand what happens in the atmospheric physics of both Earth and Venus.
I'm still waiting for a satisfactory alternative explanation from anyone in the world regarding the Venus surface temperature.
Pressure does not maintain high temperatures all by itself, anywhere, not even on Venus. So forget that "explanation."
My paper is up for PROM (Peer Review in Open Media) for a month, so feel free to publish a rebuttal or debate it with some of these members of PSI. Such a review system far outstrips the "peer-review" system used for typical pro-AGW publications.
Doug Cotton
I would like to add my support to the comments thanking the scientists for contributing here. I realise they are busy and taking time out to do this is unlikely to assist with getting their work done or in helping their careers. However, I would emphasise that I believe the sorts of informative discussions that have been posted on this thread are enormously valuable. They are a great help in improving the tarnished reputations of climate scientists and in recovering some confidence that we can find reasonable - i.e. economically rational solutions - for addressing the issues.
Climate science and all science has been seriously damaged, IMO, by the hockey stick affair, Climate Gate and the scientists 'surrounding the wagons'. AR4 reads like an alarmist document throughout, IMO. The first time I read AR4 SPM, Chapter 6 (my main area of interest) and WG2, I concluded it was written by alarmists involved in group think. I certainly hope AR5 will avoid this, but I am very doubtful this will be the case.
I look forward to the debate on climate sensitivity on 'Climate Dialogue' which I understand is soon to start:
http://www.climatedialogue.org/
I hope it can be presented in a way that gives no cause for concern about objectivity and partiality.
Hilary: your original comment wasn't to me, I mentioned you only in passing and you shouldn't have bothered with me further. 'Sorry if "terse" was wrong' was part ironic. Invading the mind of a Met Office scientist was an enjoyable conceit. Read Richard carefully and learn that less would be have been more for him too. I retain a high regard for your contributions in other areas.
Hi all,
A couple of quick responses to old and new comments.
@ThinkingScientist:
No, that's not what I meant: after all, the distance between model and reality, like financial investments, can go up as well as down :) As you can well imagine with such complex models, there have been cases in which adding or improving representation of processes or increasing model resolution have made a model worse, even after re-tuning.
And I would say that distance can never be zero, because the dimension of the output and parameter spaces are too high. You can't match every observable quantity exactly (e.g. tuning to temperature might make rainfall worse), and the best value for one parameter might contradict that of another.
It's more a statement (perhaps a manifesto!) that there is *always* a discrepancy between a model and reality, so it is essential we keep that in the forefront of our mind and try to estimate it rather than think only about the model envelope.
I absolutely agree that minimising the model discrepancy in the present day is no guarantee of a small discrepancy for future climate (or past climates, one of my own interests). In some ways we are stuck with "hoping that it is". In other words, we do (of course) know that getting the present day / recent climate right is only a necessary, and not sufficient, condition for getting other climates right.
Personally, I think the only way we can help our situation is by using as many different past climate states as possible. If a climate model can get the recent past *and* several other palaeoclimate states (ideally both warmer and colder than today) right, then we have more trust in it than if we only test against the recent past.
That's why Jonty and I (along with Mat Collins, organiser of the RMetS/RSS meeting) are working on a joint constraint of climate sensitivity using two different palaeoclimate states at once. That's the subject of my talk (and one of my past blog posts: A sensitivity subject)...
------
I also wanted to say that I'm signed up to do the climatedialogue.org climate sensitivity topic... yet another example of saying yes to too many interesting things... wish me luck!
My PLOS blog debut should be near the end of next week too, so look out for that.
@HaroldW
That's quite a good shorthand, though I'd argue structural uncertainty also contains some known unknowns. People have also called the unknown unknowns "deep uncertainty".
---
On the topic of what we choose to reply to. For me it is some combination of:
a) how "fired up" I feel about replying, which is a function of
(i) how wrong it is ;)
(ii) how well I feel able to comment, i.e. personal expertise
(iii) how quickly I think I can reply, which is related to expertise but also the length and depth of the comment.
b) how much time/internet/computer access I have within an appropriately short time of the posting.
Comments that I think would take more research, thinking, or reference-finding tend to slip past. I'd rather reply to three or four quick things from different people than one long and quite specific thing from one person. On the other hand, the latter often get added to my "list of possible blog topics".
Another consideration is how often something comes up. I note down about twenty times more things to read or think about than I get round to following up. But if there's an interesting question that keeps popping up, perhaps from different people (and not one I feel has been answered sufficiently elsewhere), it's more likely to stick in my mind for commenting or blogging.
As you know, there are many more of you commenters than us scientists. A many-to-one problem. So I'm afraid it's too much for us to answer every single thing. I don't answer every single one of my emails either (sorry to those awaiting a reply from me)...