Monday, March 15, 2010

Has the IPCC been too cautious?

Sir John Houghton writes in the Times that the IPCC has been 'too cautious, not alarmist'. He dedicates one paragraph to the question:

In truth, it’s far easier to find what now looks like excessive caution in IPCC reports. For instance, the 1990 report stated that increases in greenhouse gases were causing global warming but added that, because of natural climate variability, this warming could not be clearly detected in the observed record. As warming has continued at about the rate projected by the reports, each subsequent report has in general shown increasing confidence in its conclusions. Let me give you another example: the 2007 report declined to estimate the possible effect of accelerated melting of ice caps, as it considered no reliable estimates were available at that time.
There are two separate issues here. One is the trend over time which could easily be shown to contain more 'alarming' statements. Early reports were about detection and attribution, centred around climate sensitivity. Later reports included extreme events much more prominently. So Sir Houghton, unwittingly perhaps, confimrs that the trend over time has become more alarming indeed.

The more difficult question is how to measure alarmism (or complacency) in one time period. What is the yardstick? Is it at all possible to measure? If the debate is characterised by uncertainty, we only ever know with hindsight that something was 'too alarmist' (or the opposite) -- after we have collected reliable observations.

The recent examples of alarmism in the IPCC AR4 (glaciers, draughts in Africa, disaster losses, etc) were either the result of faulty sources, of faults in the peer review process, or of faults in the editing of the SPM documents (this is my reading of the situation). In all cases the consensus among practitioners seems to be the yardstick for assessing excessive alarmism. At this point in time the practitioners do not agree among themselves how to assess the IPCC record with regard to erroneous statements (see other post about IPCC Open letter and comments). A nice example of social construction of knowledge.

You will not be surprised to see this statement coming from a sociologist. Guess how Sir John ends his article: "We scientists have facts on our sides — we must not be afraid to deploy them."


PaulM said...

Is he joking?
It is interesting read the comments under the Times article. There are now 73 comments, of which 3 support Houghton.

Among the more ridiculous of Houghton's comments are:

1. The idea that scepticism is dangerous.

2. The claim about Himalayan glaciers was not 'marginal' and 'poorly sourced' - it was complete nonsense.

3. 'The IPCC is not a self-selected group of scientists with a political agenda' Yes it is, sceintists who do not sign up to the agenda are ostracized (example Roger Pielke excluded from the next AR5 despite expertise in the field) and/or forced to resign (example Christopher Landsea)

4. 'The IPCC is too big an organisation to be captured by an ideological cabal or fall foul of group-think' False, Houghton and colleagues show all the classic delusions of groupthink.

5. 'But a report from Greenpeace or any other campaigning body would not be included' False, 8 Greenpeace reports were included!

6. 'Each chapter of an IPCC report goes through three reviews', True but many valid criticisms are ignored and many of the worst exaggerations are slipped in at the and after the review process.

7. 'A further myth is that the IPCC is alarmist. In truth, it’s far easier to find what now looks like excessive caution in IPCC reports. ' Complete nonsense, there are dozens of examples of alarmist exaggeration.

8 'A third myth is that the IPCC has refused to recognise that there has been no significant increase in global average temperature in the past decade or so.' Completely false. AR4 claims (falsely) that there was warming between the TAR and AR4, and falsely claims that warming is accelerating when in fact it has been slowing.

In summary, virtually every statement Houghton makes is false.

Aslak Grinsted said...

The IPCC sea level projections is a clear case where the IPCC has been too cautious. Everybody knew that the projections were too low, as important processes simply was not modelled. This information was also included in the AR4 as "scaled-up ice sheet discharge", and "larger values cannot be exluded". However, several of the IPCC authors of the relevant chapters were very unhappy with the final presentation (I think they wrote a comment in Nature at the time, but I cannot find it).

This is the relevant quote from the SPM:
"Models used to date do not include uncertainties in climate-carbon cycle feedback nor do they include the full effects of changes in ice sheet flow, because a basis in published literature is lacking. The projections include a contribution due to increased ice flow from Greenland and Antarctica at the rates observed for 1993 to 2003, but these flow rates could increase or decrease in the future. For example, if this contribution were to grow linearly with global average temperature change, the upper ranges of sea level rise for SRES scenarios shown in Table SPM.3 would increase by 0.1 to 0.2 m. Larger values cannot be excluded, but understanding of these effects is too limited to assess their likelihood or provide a best estimate or an upper bound for sea level rise."
I personally think this SPM bullet-point was not enough. Tables and figures should also have communicated the same info.

[See also sealevelgate at realclimate.]

_Flin_ said...


1. While scepticism in itself isn't dangerous, to deny a very probable scenario is. If you are going 50 mph an you see a wall 100m in front of you, and the people in your car say: "It would be wise to break now", it doesn't help to say "There is no wall", "I've seen walls before, and never hit one", "The impact into a wall doesn't hurt" or "It's better to repair the car than to use the brakes". The problem is that the noticing of the wall was about 60m ago. We can still break comfortably, but we should start to do so right now.

2. The Himalayan claim was nonsense. In Working Group 1 there was no such claim. In the policy summary in Working Group 2 was no such claim. So the claim was there, it was nonsense, but it didn't have a prominent role at all. Marginal: "of minor importance". One might interpret this error as marginal.

3. Roger Pielke jr. is excluded from AR5? How comes? That is news to me. As far as I know nominations just ended 3 days ago.

4. The statement you cite is as wrong as your conclusion. Size of a group has nothing to do with possibility to err.

5. Among the contributors, futhermore, were people from Exxon, the Federation of Finnish Insurance Companies, Insurance Australia Group, Delft Hydraulics, the World Bank or Fisheries and Oceans, Canada. Which tells us... right, absolutely nothing.

6. The worst exaggerations are slipped in? Sounds like a conspiracy theory statement to me, since it is generalizing and not focusing on precise debatable points.

7. There are not dozens of examples, there are a few. The Himalayan example. And criticism about the correctness of the economic costs of mitigation. On the other hand is sea level and Antarctic land ice. Which kind of balances it out in the eyes of some people.

8. Please point out the exact phrase that there was accelerated warming between 2001 and 2007 in AR4.

And comments really say nothing. It depends where you look for comments. If you look in "The Sun" about comments on the quality of german football, comments might point in the opposite direction than in "BILD".

_Flin_ said...

@Aslak Grinsted: Maybe you might want to take a look at the sea level discussion here on Klimazwiebel: To scare or not to scare.

A very interesting discussion. I will keep an eye on that topic and am very curious whether Rahmstorf/Vermeer's projections will be close to observed sea level rise.

eduardo said...

The most pessimistic emission scenario, SRES A2, is based on a world population of 15 billion by 2100. Some results for this scenario were shown in another thread.
This scenario is now outside the 95% range of the revised UN projections, which were available for the last IPCC Report but were not taken into account.
The median UN projection is now a population of 9 billions in 2050, declining thereafter.

eduardo said...

@ 3
Dear Aslak,

thank you for commenting here.
You write that everybody knew that the projections were too low because some of the processes were not modeled. why were higher projections not included 'if everybody knew?'. could everybody give a number for these higher projections ?
I attended a talk by Peter Lemke, coordinating lead author of chapter 4, as recently as October 2009, and he stated that 'I dont buy sea-level rise projections of 1.5 meters for 2100', explaining that some processes will not contribute to future sea-level rise. Among them are the disappearance of many non-polar glaciers, the diminishing pressure for Greenland glaciers that would contribute to slowdown flow, and Greenland topography that is not very conductive for unhindered glacier flow. So it seems to me that there are competing opinions in a subject that it is very difficult to simulate properly. Of course, in the presence of uncertainty, if one wishes all projections are an underestimation. There will always be additional potential contributions that might worsen still the worst scenario: if a meteorite of the size of the moon falls on Greenland, sea-level rise will certainly be much higher. The question is how probable this event is. The sentence ' this or that cannot be excluded' is devoid of information. Nothing can be excluded.

I think the IPCC conclusions are quite reasonable here, displaying what can reasonably be simulated and pointing to the need of further work. We are discussion about sea-level for 2100. The present trend is about 3 mm/year. Cant we wait 10 years to pin down better estimations of sea-level rise?

Hans von Storch said...


you claim Everybody knew that the projections were too low, as important processes simply was not modelled. - This is obviously false, as at least some persons, namely Peter Lemke and the majority in his group of authors, did not have such knowledge. Such categorical statements are certainly not helpful, and have lead to the problems we are now in. What is true: "Some people claim that the projections are too low. And these people may be right."

However, the task of the IPCC is not to determine the "truth" but to asses the knowledge available at a certain time, in this case 2007 or a bit earlier. This knowledge, to believed correct at that specific time, may turn out to be false at a later time. However, writing down the knowledge, as accepted in 2007, is the "right" decision for the IPCC AR4, even if the knowledge later turns out to be "false". It would be "false" by the IPCC to make "true" statements (in the sense, that they turnout being "true" at some later time) contradicting the knowledge at the time of the assessment.

Therefore, it is also not adequate to claim, as some have done, that the alarmists-errors in WG2 like the Himalaya-case would be "balanced" by "downplaying"-errors in sea level projections. The Himalaya is inconsistent with the IPCC task, the sea level-error (if it turns out to be so) would be consistent.

One of the reasons, why people thought in the final phase of AR4 that the sea level projections should be considerably higher, was the science-paper by Rahmstorf (2007), which however was criticized in writing by at least three independent groups for various reasons, among them statistical and physical inconsistency. There is nothing uncommon and dishonorable with such a process, but it demonstrates that newest results are not necessarily superior, and that a certain time for the self-correcting processes is needed before the knowledge claims is accepted. It shows also that modesty is a worthy attitude. -- Hans

_Flin_ said...

@Hans von Storch: ad 1: Your little story is unrealistic, because there is a multitude of threats at the same time, some of which are delusions, some very real, they all need our attention. Investing your attention on one issue means to not paying attention to others, which may deserve more attention. This is a normative decision, a subjective decision.

I do not understand the multitude of threats. Does that refer to multiple threats connected to climate science (e.g. extreme weather/sea level/ocean acidification/security issues) or to multiple general threats (e.g. financial crisis, shortness of ressources, population growth, religious and political extremism leading to unstable governments) which all need monetary ressources (which are limited) allocated to them by policy makers to try to handle them?

If it was the former, doesn't mitigation take care of most of them? If it is the latter, isn't there the need to quantify the available ressources for the task at hand? Like saying (to come back to the little story) "ok, there is a wall, we got airbags, we can only concentrate on braking for a given amount of time, but if we can slow down to 25 mph, the impact and damages will be manageable. And maybe the wall isn't as hard as it looks from here after all.".

In both cases, the wall exists and doubting it's existence in the cloak of scepticism while trying to delay the start of the braking process leads only to a harder impact or more uncomfortable braking.

(I just assume that acting now is better than acting later.)

Aslak Grinsted said...

@Eduardo & Hans:

I totally agree that IPCC should not pull numbers out of a hat. If there were no proper projections available taking everything into account, then ofcourse they should not make them up.

My main gripe is that the true uncertainty was not clearly communicated. I would even say that the way the uncertainty ranges were displayed were misleading. Especially when you consider statements regarding the confidence intervals such as this: "The ranges are narrower than in the TAR mainly because of improved information about some uncertainties in the projected contributions."

I am not advocating that we should put special emphasis on the worst-case scenario. Ofcourse people have to wait as long as it takes for the research to make proper projections. However, I think that it is necessary to communicate what our best guess is already now: That is the semi-empirical projections. It is not about scaring people, but allowing society to make rational choices based on the best information available. It can be as mundane as making new regulations to ensure that new infrastructure will not be overly vulnerable to a 1m sea level rise.

Here's my take on what semi-empirical projections (Jevrejeva et al. 2010, Vermeer & Rahmstorf 2009, Grinsted et al. 2009, Rahmstorf 2007) agree on:
* 21st century SLR in AR4 ~3 times too low.
* Large system inertia. SLR will continue for centuries. ~0.5 degC cooling needed to stop sea level rise.
* Mitigation. Policy-choice does matter: ~0.5m between projections for A1FI and B1.
* Significant SLR even in most optimistic scenario. Adaptation necessary.
Note these papers use different models, different input data, and different statistical methods.

I also think it is worth considering the long-term equilibrium sea level response, even if it looks scary.

Aslak Grinsted said...


OK, I take back the "everybody knew" statement. However, it does accurately reflect my own initial reaction to the AR4 projections (knowing nothing Rahmstorfs 2007 work at the time). My close colleagues had the same reaction. - And as I said, some of the authors of the relevant IPCC chapters were unhappy with the SLR presentation [unfortunately I still cannot find the reference].

You have to remember that there was evidence showing a 'rapid dynamical change' was already occuring (i.e. Jakobshavn). We knew that the ice sheet models of the time simply would not be able to model things like the Jakobshavn decay. By design they can only respond slowly to warming and had poor representation of the ice/ocean physics. There were also other evidence which strengthened the view that the ice sheet contribution generally was underestimated both in the past and in the projections.

I severely doubt that ice sheet models will be mature enough to be able to give credible mass loss projections for AR5. There is simply so much still unknown concerning key-processes. E.g. how important is the Zwally-effect vs. the Holland-effect? Both processes increase boundary flow velocities and dynamical discharge.

Hans von Storch said...

Dear Aslak,

we are talking ab out two different issues:
a) was the account given by IPCC AR4 an accurate description of the knowledge to be assessed at the time?
b) How will the assessment of AR5 be?

These are obviously very different questions - and I would invite you (and others) to start new threads on these issues. Would you like to write a short text on your position on (a)? -- and/or, if you wish another, independent one also on (b)? Maybe, we could entice someone like Peter Lemke or John Church to write about (a) as well?

-- Hans

Hans von Storch said...

_Flin_/98, multitude of threats?
I meant that we are confronted with various threats, climate being one of them. There is the certainty that many people will die of hunger in the next half hour, and the next ...; new diseases will show up; some political adventurers will begin wars; strange ideas will lead people to suppress or kill other people.
Your narrative is assuming that there is one dominant threat, and we have often enough heard claims that climate would be the most important of all threats, but I doubt that. Your narrative is also assuming that there is only one possible reaction; and a solution which undoubtely will solve this problem, without severe repercussions. Maybe the car would explode, when you stop? Or somebody is chasing you to kill the child sitting next to you? Then you can not stop, but maybe you can slow down, without stopping, and turn your car; or you have enough time to add wings?

Your narrative is really not a good model of the situation we are in. Life is not that simple.

PaulM said...


1 False analogy. Better to say you are driving a car, stop for ten years, then some people tell you you are about to crash unless you do what they tell you to do.

3 It was an IPCC special committee (but on their website under AR5), see Pielke's blog

5 Your comment is irrelevant. The point is that Houghton's statement is false.

6 An example is the dishonest comparison of 25 yr 50 yr and 100 yr trends in chapter 3 page 253. This would never have got through expert review. It was not in the earlier drafts.

7 There are dozens of examples, including the above, and the antarctic sea ice story discussed on various blogs.

8 p 249 "Clearly, the world’s surface temperature has continued to increase since the TAR". False statement by IPCC, showing Houghton's statement also false.

Hans von Storch said...

PaulM and _Flin_: May I suggest that you try to determine what the issues are you really disagree on, and after having discriminated the significant issues from the minor ones, discussing the significant one by one in some depth? Maybe, you will avoid terms like "clearly"; reference to blogs is usually also not helpful. -- Hans

Anonymous said...

Hans, the word 'clearly' appears twice here. Once it is written by Aslak Grinsted, and the second occasion it is a direct quote from the IPCC :)

It would be interesting to hear your opinion of this quote from the IPCC?

And also your opinion of the IPCC statements on Antarctic sea ice?

Kooiti MASUDA said...

I looked at the last item mentioned by PaulM (#13), Page 249 of IPCC AR4 WG1 around the sentence "Clearly, the world’s surface temperature has continued to increase since the TAR" (the second occurrence of the word "Clearly" on that page).

If the sentence is taken out like PaulM did, readers likely think that it means the trend from 2001 to 2007, or if they learn about the editing process of IPCC, from 2000 to 2005. The trend in this 5 or 6 year time frame is certainly not "clear" (though I do think it is more likely positive than not). But this is not what the sentence actually meant. (Scientists do not want to discuss trends in such a short term.)

The context is comparison of knowledge available for TAR and for AR4 about the trend since the latter half of the 19th century to the present. The likely amount of observed warming in this sense was revised to a larger value. This is a result of various factors combined: in addition to the shift of "present" time, some past data have been recovered, and the analysis methods are elaborated.

I do not agree with the choice of the word "clearly", and also I can say that this sentence is ambiguous.

Let us not accuse IPCC and its authors about past minor flaws. Let us not use such ambiguous parts of the reports as evidence for anything. Let us discuss how to make better documents by IPCC and by others in the near future.

In addition to expert reviewers, IPCC can have reviewers who check expressions which can be misunderstood. But it would be too much burden for IPCC or for the reviewers if they are deemed punished when any faults would be found in the reports. We should encourage improvement, but we should not demand perfectness.

Also I hope, in this particular case for example, that the change due to the extended time frame and due to revision of the past time series can be discussed separately. But according to the current scheme, authors of the IPCC reports should not do original analyses but just discuss results already published. And the authors of the individual papers do not have obligation to IPCC. Should IPCC commission studies to answer their questions? Or should it ask national governments for voluntary commission on behalf ot it? It seems to be a difficult problem related to the balance between neutrality and effectiveness of IPCC.

ghost said...

I would say, the IPCC report should be as correct as possible and should summarize the current scientific knowledge about climate change and climate change caused by the mankind.

That does not mean, it should show a consensus, it should illustrate the range of scientific opinions and the corresponding underlying evidences. And maybe should give a possible "grade" of the opinions.

For example:
* Co2 increase is not caused by the mankind (EG Beck, Spencer, Schrumm): special category
* temperature observations are biased because of "warming stations" and stations dropout (Watts, Pielke Sr, Spencer): special category(*)

you see the pattern, just kiddin'.

Of course, real findings and real controversies and uncertainties should be handled like that but in a thoughtful way.

I would say: show the controversy and the underlying evidences of the different opinions. What does really support the GCR hypothesis, and what not? What are the differences in the "hockeysticks"? What does support a low sensitivity regarding Co2? What does speak against it? I know most of this stuff is already in the report, but it shut be explicit. However, I could imagine, it could be tedious for scientists.

(*)special category is the nicest term I found

Kooiti MASUDA said...

Addendum to my previous comment (#16):

The sentence
"Clearly, the world’s surface temperature has continued to increase since the TAR"
should have been worded like this, for example:

"The Centennial increasing trend in the world's surface temperature which appeared in the TAR continues to appear in the AR4 where records are extended forward by five years."