Richard Betts recently mentioned that he worked on the writing of section 2.5 of IPCC AR4 WG1. Looking at this reminded me of an error at the end of the previous section. This also seems timely in view of the IPCC's new protocol for addressing errors, discussed recently at Climate etc.

A key part of chapter 2 is the diagram, FAQ 2.1 Fig 2, showing the 'radiative forcing' effect of various human influences. The figure is quite important, so it also appears in the SPM as fig SPM2, as well as later in chapter 2 as Fig 2.20.
Some of these human influences are warming, and some are cooling.
One of the cooling ones is the 'cloud albedo effect', which is due to
aerosol particles making clouds more reflective, or more long-lasting.
Studies of this effect are listed in Table 2.7 (p 174) and Fig 2.14 (p 177).

The mistake was first noticed by a Climate Audit contributor, Michael
Smith, here back in January 2008. Here is his comment, slightly edited:

In chapter 2 of the WG1 section of AR4, section 2.4.5 discusses aerosol cloud albedo effect. Figure 2.14 shows graphically the cloud albedo forcing results of 28 GCM runs. See here: http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-chapter2.pdf (page 177)

After a lengthy discussion of the various issues involved in estimating this particular forcing, the authors summarize on page 180 as follows:

"Based on the results from all the modeling studies shown in Figure 2.14, compared to the TAR it is now possible to present a best estimate for the cloud albedo RF of -0.7 W/m2 as the median, with a 5 to 95% range of -0.3 to -1.8 W m2."

However, according to my Excel spreadsheet, the median of the data
shown in figure 2.14 is not -.7, the median is actually -.99. (The
values that were graphed in Figure 2.14 are shown in Table 2.7, pages
174-176. I used the values from that table. Note that only the values
shown in bold were included in Figure 2.14, so those were the ones I
used.)

In the following discussion, UC agrees with Michael, and it appears that the error was discussed and acknowledged at RealClimate (can anyone find the link to that?)
There is also a missing sign error in Table 2.7 in the number (-0.54) from the Suzuki et al study. Another slight confusion is that one number from Table 2.7 is inexplicably omitted from Figure 2.14 - if this one is included the median is -1.07.

One possible explanation might have been that different results from the same study (one study has 7) were lumped together as one number, but I checked doing it this way (taking the median within each study) and it gives a very similar answer, -1.11.

I am not sure whether this error was ever officially reported to the IPCC.
Of course, they would claim that it "doesn't matter", and it is in a field where scientific understanding is acknowledged to be 'low'.
Nevertheless, it means that the diagram that appears several times is misleading - the blue bar for cloud albedo effect should be considerably larger than shown. Also, the incorrect number -0.7 is quoted in the executive summary of the chapter on p 132.

Here are the numbers from Table 2.7:

[-0.45 -1.34 -1.37 -1.43 -1.39 -1.12 -0.85 -1.85 -1.55 -1.35 -0.54 -1.3 -0.77 -0.9 -0.5 -0.3 -0.22 -0.52 -0.75 -1.3 -0.86 -1.07 -1.1 -1.29 -1.79 -1.4 -0.65 -0.68 -0.74]

It is the -1.37 that is not included in Figure 2.14.

PaulM

Thanks for raising this and explaining in detail. I'll ask my co-authors about that. It may take a while due to people being on holiday etc but I will get back to you in due course.

Sorry Richard, but I cannot resist the temptation to ask how long it takes for a team of top climate scientists to calculate the median of 29 numbers.

I reported this to the IPCC secretariat today:

I write to report an error in the IPCC AR4 WG1 Report, Chaper 2, page 180. It is stated that the median cloud albedo RF from the results in Figure 2.14 is -0.7 W/m2. In fact the median is -0.985.

There are three further very minor errors in this section:
1. Table 2.7, Chen study, 0.54 should be -0.54.
2. Table 2.7 note d, reference to Fig 2.16 should be 2.14.
3. Fig 2.14 omits one of the bold values from Table 2.7 (Williams et al study).
If this is included the median is -1.07.

I also sent it to Piers Forster, Coordinating Lead Author of the Chapter, asking if he was aware of the error.

That reminds me,

I asked help/reported an error in a Nature Climate Change article to a prominent skeptic blog.

You know what the response was?

"Why should I care?"

Hi PaulM

I'm now back from holiday and have an answer for you, having consulted with Piers Forster.

The median given in the text (-0.7 W/m2) is of the numbers in the lower panel of figure 2.14, which gives the results from the more complete models - you will see on the left hand side that this panel considers more aerosol species than the top panel (which shows results from models which still followed the older practice of considering only mainly sulphate aerosols).

It's discussed in the text that as well as the more complete models giving a smaller -ve cloud albedo RF, observationally-based estimates are also for a smaller RF.

Admittedly this is not at all clear from the text in section 2.4.5.6 - it should have said "modelling studies in the lower panel of figure 2.14". Sorry about that.

Thanks very much for raising this, it is important that these things check out. I strongly encourage you to apply the same scrutiny to AR5 as an expert reviewer! (WG1 FOD comes out in December, WG2 and WG3 in first half of next year)

Richard, thanks, in fact I had a very prompt reply from Piers Forster saying the same thing. But using the values in the lower panel of fig 2.14 I get a median of -0.81, still not -0.7, so I think there must be another undeclared fudge.

This highlights one reason why I am not keen to be an IPCC reviewer. In the Second Order Draft, (the last version to be inspected by reviewers) the figure was given as -0.9. The mysterious fudges to bring it down to -0.7 for the final version were carried out after the expert scrutiny. Similar things happened elsewhere in AR4.

Hi PaulM

I think it is to do with the Chen and Penner values being from a sensitivity study so not all the values were used - just a single one.

Richard, c'mon!
If you are going to use a single value you are very quickly going to get accused of all sorts of things, of which 'cherry-picking' is likely to be the politest.
Surely you can understand why sceptics tear their hair out when you and PaulM have ploughed your way through what looks like a fudged estimate for cloud albedo and we then find out it was all down to one figure.
"Which figure do you fancy?"
"Oh! I don't know. How about that one?"
"Okay."
I'm not saying ....

Hi Mike,

Sorry, yes, that looked bad - a bit rushed!

The single value was not cherry-picked - it was (I believe - I didn't actually do this bit) the "control" value from the Chen and Penner (2005) study.

(Here is the figure BTW)

So, PaulM, see if you agree with this:

Using the lower panel from the above figure, take all values from all studies except the Chen & Penner (2005) sensitivity study, for which you use the UM_ctrl value only, and find the median. -0.7?

On the issue of the change from -0.9 in the SOD, this can be explained by looking at the chapter SOD page 2-150 - you can see that this version of the figure does not include the Chen & Penner (2005) or Penner (2006) studies, and without those it suggests -0.9.

If you look at the review comments on the SOD you can see that Chen and Penner (2005) was brought to the attention of the authors in review comment 2-830 by Joyce Penner, so the change is actually a result of the expert review process.

Incidentally, it's interesting looking back at the SOD and seeing that actually it is better explained than the final version (eg: it makes clear that only the bottom panel of the figure was used, and why). I do remember we had to shorten text in order to meet the word limit, so I expect the clarity was lost in the shortening.

Hope this clarifies!

Yes, this does seem to be the second fudge. But this explanation is contradicted by the caption to Fig 2.14 which explicitly says that the Chen and Penner results are regarded as independent.

Regarding the excuse about shortening, the (incorrect) wording in the final version is
"Based on the results from all the modelling studies shown in Figure 2.14".
The wording in the SOD was
"Based on the results from the 8 modelling studies shown in Figure 2.16: Bottom".
Can anyone really claim that the omission of the single important word 'bottom' was done in order to meet a word limit? If there were word limit problems, the irrelevant phrase "compared to the TAR" could easily have been dropped.

Well it wasn't deliberately done to confuse, if that's what you're implying! :-) Genuine oversights do happen - and also "compared to the TAR" is not irrelevant, as one of the important requirements of each Assessment Report is to be clear about how things have changed since the previous one.

But if we're going to get into specific details of wording, please note that section 2.4.5.6 says "Based on the results from all the modelling studies shown in Figure 2.14, compared to the TAR it is now possible to present a best estimate for the cloud albedo RF of -0.7Wm-2...." So this does not claim to actually be the median of the results in Figure 2.14 - it is merely based on them. An extremely important point is that the most recent observationally-based studies (using AVHRR and POLDER) point to smaller values of RF (-0.64 and -0.37 respectively) - see section 2.4.5.3. That is not in Fig 2.14 because the figure is about modelling studies and the AVHRR and POLDER are obs. Using only the results from the figure would have ignored the observational evidence.

Expert judgement does have to come into it, using all available evidence including obs as well as models alone (I'm sure you'd agree with that?!) and I don't see a problem with that.

So while I completely agree with you that the explanation of where the -0.7 Wm-2 came from is not as clear as it should have been (again, sorry about that) I still think it is the best possible estimate which is backed up by the text. And it is still made clear that the Level Of Scientific Understanding is "Low".

What sort of QC/QA or any type of checking was done on these documents? It reminds me of days in academia many decades ago when the figures didn't matter so much it was the guru's words that mattered.

I wrote to the Dept of Energy and Climate Change about the carbon trading scam in the EU. I paraphrase their answer. The vast majority of climate scientists... The IPCC 90% probability .... Stern says it will cost much less .... In any case it makes sense to reduce dependency on hydrocarbons ... Hope this is helpful. Yours.

Yeah, yeah, heard it all before but what about Ms Duggan and the scam?

http://blogs.telegraph.co.uk/news/jamesdelingpole/100102296/sun-causes-climate-change-shock/

Reading around this it's shocking how many people don't quite follow how hypothesis testing works. We have AGW and 90% (although I never saw any statistics apart from top of the head stuff). Here we have a credible partial explanation (with no top of the head probability of course). Seems disproving an alternative (Ha1:AGW done it) now depends on supporting another alternative (Ha2: Factor X done it). Of course the null for each is different (H01 and H02 resp) and the alternative for the other is potentially part of it but not all of it.

It's also shocking that the mainstream media are not reporting it. Nature and CERN not good enough research then?

Camp David

Nature and CERN not good enough research then?

No. Not if they're going to cast doubt on CO2 as the main driver of global warming. Too much political and scientific capital, not to mention the credibility of MSM "environmental correspondents", and the livelihoods of half the staff and hangers on of Greenpeace, WWF, Friends of the Earth and several hundred other NGOs you could mention — Environmental Investigations Agency was another that I first heard about today (see Unthreaded) — is predicated on CO2 being the big baddie.
Anything that casts doubt on that is likely to get spiked. With the result, if we are not very careful, that we will never really know to what extent CO2 is a driver because we refuse point blank to investigate, or if we can get away with it even permit, research into alternatives.
That's post-modern science, man!

Mike,

Of course the paper doesn't claim much, CERN were very careful not to upset anyone who it may not have been prudent to do so.

I'm afraid the goose only had the chance to lay one egg before someone saw an enconomic opportunity. The rest of the clutch were consequently not seen as golden and hence discarded. Depends what is 'gold' to one. Some would have the goose shot in case it lays any more inconvenient ones. That's post-modernism for you.

Thanks for your comment.

My own thoughts were that the paper released from CERN empirically validates part of Svensmark's work with the intention of releasing more data as it is confirmed over a period of time.

I think this is happening gradually for a number of reasons. Mainly the indications are that Svensmark is correct which is going to step on the toes of a lot of influential people that need to come around to the idea.

LB et al

Without doubt it supports Svensmark's hypothesis in which case it should be considered in the 'modelling' instead of discarded apriori as not supported. Now let's for the sake of it say that this factor is a good predictor of global temp. Then accounting for the variability it is able to explain may not leave very much in the system that CO2 can explain.

It at least should not be dispelled out of hand, and any counter suggestion that it does not affect temps should not be used as evidence that CO2 done it. The null is simply that CO2 didn't do it, for that we don't need any other alternative in proper science.

Of course in subsequent discussion one must bear in mind that temps may be causing CO2, whilst temps are never going to cause solar flare activity. Ones a chicken or egg and the other isn't, yay for solar flares the only true X we have.

The cosmic ray idea was discussed in this section of our IPCC AR4 Radiative Forcing chapter. Evidence for how it might actually translate into a quantifiable radiative forcing wasn't clear at the time (and there are still many gaps). I'm not involved in the equivalent chapter for AR5 (I'm in a different chapter and working group this time) but natural radiative forcings such as this are in scope for that chapter, and I imagine it will be covered.

As I mentioned on a previous thread on this, the latest conclusion by someone attempting to actually quantify the influence on cloud cover (and hence RF), by Pierce and Adams, was that the effect is small. The CLOUD paper doesn't say anything about whether that conclusion should be altered, and I wrote to the lead author (Jasper Kirkby) to ask about this but have not yet had a response.