The IPCC discards its models

Who will believe the CMIP5 models, after the IPCC plenary had to discount their temperature projections?

Guest essay by Barry Brill

Table SPM.2 Projected change in global mean surface temperature (°C) for the mid- and late 21st century relative to the reference period of 1986-2005. From the IPCC AR5 Summary for Policy Makers

Last August, I described the terrible bind which was about to face the IPCC Stockholm meeting http://wattsupwiththat.com/2013/08/31/can-the-ipcc-do-revolutionary-science/. Everybody knew the draft SPM was an embarrassment – the climate sensitivity range was far too high, the models were plainly wrong and the temperature ‘hiatus’ was left unexplained.

The IPCC had only three options – (i) re-run the models and re-draft the whole report, (ii) issue a string of caveats, or (iii) simply bluster on.

Pat Michaels and Chip Knappenburger ran a similar piece^[1] with three similar options. They predicted the IPCC would “do nothing and mislead policymakers and the rest of the world”.

After four days of intense debate and frantic wordsmithing amongst the Government representatives gathered at Stockholm, Thomas Stocker announced the final decisions at a media conference. There was no apparent backdown from the modeled temperature projections and no offer to re-run the models – just an assertion of increased certainty.

Months later, Lord Monckton has applied his aristocratic nose to the belatedly-published WG1 report, finding it to be both malodorous and “furtive”: http://wattsupwiththat.com/2014/01/01/ipcc-silently-slashes-its-global-warming-predictions-in-the-ar5-final-draft/. The Stockholm politicians were more wily than first met the eye – it turns out that the “95% certain” models were actually rejected by the meeting.

Method

The trick was to convey full oral confidence to the applauding media, whilst inserting into the written record a few key scraps of “deniability” regarding the rejected models. Scraps they certainly were:

(i) The plenary meeting failed to reach consensus on the sensitivity (ECS) estimate – the single most important figure in the report. No public comment was made regarding this unprecedented omission but a discreet footnote appeared on page 11:16 “No best estimate for equilibrium climate sensitivity can now be given because of a lack of agreement on values across assessed lines of evidence and studies.”

(ii) The meeting agreed that the lower bound of both sensitivity ranges (ECS and TCR) should be reduced in future. But those decisions were not retrospective, so the altered ranges were not reflected in the modeled (CMIP5) temperature projections circulated in the second-order draft.

(iii) The meeting reached a consensus that the modeled projections were flawed. Stocker admitted that a short-term allowance had been made because some models were “running hot” (ie sensitivity was too high). The SPM refers to the WG1 report {Box 10.2, 11.3} saying “…in some models, an overestimate of the response to increasing greenhouse gas and other anthropogenic forcing (dominated by the effects of aerosols)”.

(iv) SPM.2 (Note (c)) discloses coyly that the “assessed range for near-term (2016-2035) temperature change is lower than the 5-95% model range”. However, the discounted near-term figures were omitted from Table SPM.2.

(v) The Stockholm meeting directed that confidence in the mid-term (2046-65) future projections be marked down to “medium” – meaning about a 50:50 chance of being correct, or ‘as likely as not’ to be wrong. This nervousness about natural influences was not mentioned by Stocker.

(vi) SPM.2 (Note (c)) discloses that there is “insufficient scientific understanding” whether “the factors that lead to the assessed range for near-term projections”, might also apply to the mid-term.

In short, the Stockholm participants acknowledged the modeled future projections were based on over-estimates of sensitivity – and were therefore unacceptable. They then “assessed” ad hoc amendments to make the projections appear more plausible.

Discount Factors

The graphs reproduced in Christopher Monckton’s WUWT essay show that the final WG1 report has dropped the 2016-35 range to 0.3-0.7°C, while the models projected this at 0.4-1.0°C. At mid-points, then, the discount applied at Stockholm was approximately 33%.

At Climate Audit, Nic Lewis^[2] shows that the IPCC’s own revised aerosol numbers unavoidably produce a TCR of around 1.3°C, while the CMIP5 models use a mean TCR of slightly over 1.8°C. The modeled projections therefore need to be discounted by about 25-30%. And this really matters –

“The key determinant of the range and mean level of projected increases in global temperature over the rest of this century is the transient climate response (TCR) exhibited by each CMIP5 model, and their mean TCR.”

Michaels & Knappenberger’s^[3] commonsense sanity check show that TCR can’t be more than 1.4°C based on observations. They call for a discount of 20%.

The Stockholm meeting would have been most reluctant to discard the modeled projections that have always been the mainstay of IPCC reports. They applied the discount only because it was obvious to all that the models were badly astray. But, if that was the case, why didn’t the discount apply to ALL the projections?

The SPM’s reference in Note (c) to “different levels of confidence in models” seemingly refers to the different opinions about TCR levels. It appears (from the full WG1 report) that the “additional uncertainties” relates to the causes and longevity of ‘the hiatus’.

Both of these factors are obviously major concerns to both the IPCC and the world in general. Yet both continue to be ignored in the assessed projections for 2046-2065 and 2081-2100 “due to insufficient scientific understanding”.

The mid-term and long-term ranges shown in SPM.2 are taken directly from the draft circulated to IPCC member governments on 2 August 2013. They take no account of the fact that temperatures during the preceding 2016-2035 period have since been assessed downwards by 33%. They must therefore be wrong.

In addition, the mid-term projections are expected to be affected by natural influences that cannot be simulated.

Implications

The table of temperature projections is the heart and soul of each voluminous IPCC Assessment Report. Climate alarm stands or falls on the credibility of that table.

Despite its customary obscurantism and spin, the IPCC has now admitted that:

• a number of its CMIP5 models seriously exaggerate future warming;

• the climate sensitivity range used for the modeled projections is too high;

• internal variability^[4] is expected to significantly offset warming (for some decades);

• scientists cannot quantify the influence of sensitivity or of internal variability beyond about 2035; and

• consequently, the modeled temperature projections are unreliable.

These admissions severely dent the authority of the IPCC. But their manner of reacting to this situation will do even greater damage to the Panel’s credibility. The Stockholm meeting decided against the obvious course of excluding the faulty models to obtain an ensemble of reliable simulations. Instead, it decided to:

• replace the table of modeled projections by an assessed table which met the “expert” opinion of participants (the majority of whom were non-climatologists);

• publish assessments/projections which have not been subject to any form of review or comment;

• apply arbitrary and swingeing (33%) adjustments to figures which pretend to tolerances of hundredths of a degree Celsius;

• tolerate non-robust (only 50:50 confidence^[5]) projections covering the next 60 years, even when they know those projections to be wrong;

• disguise its own puzzlement and internal disputation beneath a threadbare cloak of increased certainty and consensus.

Perhaps the most grievous fault of the AR5 “assessed” table is that it presents a very broad gamut of potential future temperatures – ranging all the way from the benign to the alarming. Then it confesses that it has no opinion (or even a leaning) as to which part of this spectrum is actually likely to occur.

The WG1 report (11-7) defines the term ‘climate projection’ as “a climate simulation that extends into the future based on a scenario of future external forcing”. Clearly, the near-term figures provided by the AR5 are neither simulations nor projections. They are simply guesses.

###

References:

^[1] www.cato.org/blog/ipcc-chooses-option-no-3

^[2] http://climateaudit.org/2013/12/09/does-the-observational-evidence-in-ar5-support-itsthe-cmip5-models-tcr-ranges

^[3] http://www.cato.org/blog/new-ipcc-report-will-be-internally-inconsistent-misleading

^[4] Which is excluded from the models because they “are not expected to reproduce the timing of natural internal variability” [D1]

^[5] Lower than any previous published projections. This huge uncertainty should decimate cost-beneft studies.