Note it: POST

The Parliamentary Office of Science and Technology has published a briefing paper on weather and climate. I've had a quick glance, and this caught my attention.

Natural forms of climate variability are likely to be the main influence on the UK’s climate over the next few decades.

Who knew?

This seems really interesting to me in the context of the ongoing scrap over the plateauing of temperatures. If the claim in the POST paper is correct (the citation is to this paper by Hawkins et al), then that presumably means the plateauing could continue for decades, or even morph into a decline, and we would still be faced with arguments that temperatures are rising and that we only have 24 hours left to save the Earth.

Update on Feb 3, 2012 by Bishop Hill

Having had a look at the Hawkins paper, I think maybe POST has got the wrong end of the stick. Here's what he says:

Using data from a suite of climate models we have carried out a quantitative assessment of the contributions to uncertainty in predictions of regional temperature change. We have estimated the contributions to the total prediction uncertainty from internal variability, model uncertainty and scenario uncertainty. For lead times of the next few decades the dominant contributions are internal variability and model uncertainty.

Update on Feb 3, 2012 by Bishop Hill

No, I'm wrong. Ed tweets that POST are interpreting him correctly. A few decades coming in which natural variability swamps AGW.

Comments

[Richard C (NZ)]

The POST quote is this:-

Natural forms of climate variability are likely to be the main influence on the UK’s climate over the next few decades.2

But following that is this:-

Then, as the century progresses, the influence of increasing greenhouse gas (GHG) concentrations is likely to be of growing importance.2

The paper referenced is:-

2 Hawkins and Sutton, Bulletin of the American Meteorological Society, 2009.

This is what Roger Pielke Sr says about Hawkins and Sutton 2009 (abbreviated):-

Comments On A Scientifically Flawed Paper “The Potential To Narrow Uncertainty In Regional Climate Predictions” by Hawkins and Sutton, 2009

The authors are making the scientifically unsupported claim that there is less uncertainty in predicting the global average surface temperature several decades from now, as compared with shorter time forecasts of this metric. That is, unlike weather forecasts which deteriorate in skill with time, they are concluding that after a few years of less accurate skill, climate prediction starts to increase after a couple of decades.

There is no scientific basis for this claim of the improvement in model prediction skill and reduction in internal variability of any climate metric for time periods of several decades into the future.

This paper should never have passed peer reviewed.

http://pielkeclimatesci.wordpress.com/2009/10/07/comments-on-a-scientifically-flawed-paper-the-potential-to-narrow-uncertainty-in-regional-climate-predictions-by-hawkins-and-sutton-2009/

Cross-posted to Climate Conversation Group (NZ) 'Sceptics query our truth – we shall besmirch and slander them'

http://www.climateconversation.wordshine.co.nz/2012/01/sceptics-query-our-truth-we-shall-besmirch-and-slander-them/#comment-79330

[Ed Hawkins]

Firstly, hello everyone! Secondly, I was having a twitter discussion with Bishop about this, but didn't realise he had written a blog article on it, else I would have contributed earlier.

Where to start? Perhaps a quick look at the UK temperature timeseries will help with some perspective:

Central England Temperature timeseries

There is a lot of natural variability! 30 year linear trends can be up to +1C and -1C in 30 years:

Blog article

caused by variability alone.

The projected AGW trend for the UK from IPCC AR4 is roughly +1C in 30 years (with a range, depending on which model you use). So, it is entirely feasible, and consistent with model simulations, that the UK experiences no warming in the next 30 years, or that it experiences +2C warming in the next 30 years. The most likely warming rate, according to IPCC is +1C in 30 years, but the natural variability can mask or enhance the projected long-term trend.

Globally, the variability in temperatures is far smaller, and so it is not consistent with the model simulations to see no warming over 30 years.

cheers,
Ed.

[Bishop Hill]

Thanks Ed

I think POST article could be clearer - I read it as saying something about the next few decades in particular. It's more about the relative sizes of AGW and natural variability I think.

[Richard C (NZ)]

Ed Hawkins at 9:55 AM

I note that HadCET temps are now lower than they were in 1659

http://junksciencearchive.com/MSU_Temps/HadCET_an.html

[Ed Hawkins]

To Richard C:

1. Roger Pielke's criticsm is completely incorrect. We clearly state in the paper (which is freely available) that we are NOT considering skill, but uncertainty, which is entirely different.

"A final important point to emphasize is that the discussion of prediction uncertainty in this study is
based on the variance of model predictions (“spread”) rather than the variance of prediction errors (“skill")"

2. If you think that one year of CET is enough to make a claim such as 'it is colder than 1659', then you will be interested to know that the 2011 CET data (10.70 deg C) was the second warmest ever (behind 2006).

cheers,
Ed.

[HaroldW]

Dr Hawkins -
In your paper's abstract, you wrote

Our findings have implications for managing adaptation to a changing climate. Because the costs of adaptation are very large, and greater uncertainty about future climate is likely to be associated with more expensive adaptation, reducing uncertainty in climate predictions is potentially of enormous economic value.

Uncertainty in predictions is defined as composed of model uncertainty (as you write above, the spread of model predictions, given identical inputs), scenario uncertainty (our inability to predict the anthropogenic inputs to climate models), and natural variations (unpredictability of non-anthropogenic inputs).

It seems that the accuracy of models ("skill") is of no relevance to your analysis. It would seem from the paper that you would be satisfied if all climate models converged on a single answer -- thereby producing minimum model uncertainty -- even if that answer does not correspond to the actual climate response.

Could you comment please?

[Ed Hawkins]

Hi HaroldW,
I certainly do not want to give the impression that we would like zero model uncertainty and the 'wrong' answer. We absolutely want the 'truth' to be within the uncertainties we estimate, but the smaller that uncertainty is the better.
cheers,
Ed.

[Richard C (NZ)]

Ed Hawkins at 10:36 AM

So does prediction become more certain as the century progresses? Or less?

If we can preclude disagreement over definitions of "skill", the certainty (or uncertainty) gives us an indication as to the usefulness of prediction in regard to the statement "Then, as the century progresses, the influence of increasing greenhouse gas (GHG) concentrations is likely to be of growing importance".

The statement can only be valid if certainty increases as the century progresses, because if it doesn't increase, you're just making stuff up.

[Richard C (NZ)]

Ed Hawkins at 10:36 AM

In regard to CET 2011. If we look at 1978 -2011:-

http://junksciencearchive.com/MSU_Temps/HadCET_mon.html

At Max it is equal lowest this century and at min it is was coolest since 1985.

Since 2006, minimums have become progressively cooler.

There's no correlation to CO2.

I was just making the observation that the latest measurement is cooler than 1659 - what's 2011 got to do with that?

[Ed Hawkins]

Hi Richard C,

Fig. 2 in the paper we are discussing clearly shows that the uncertainty increases with time! However, for a few decades the signal increases faster than the uncertainty, so the signal-to-noise increases initially.

As for CET, many blog commenters have happily claimed that a cold 2010 means that there can't be any AGW, which is why I raised 2011. For the correlation, why not take annual mean CET and CO2 back to 1659 or 1850 and see if there is a correlation - my guess is that there is. The seasonal cycle plot you link to is highly misleading.

I don't have any more time to respond here.

cheers,
Ed.

[Richard C (NZ)]

Ed Hawkins at 10:37 AM

So your sole support for the statement (cited by the UK Parliamentary Office of Science and Technology):-

"Then, as the century progresses, the influence of increasing greenhouse gas (GHG) concentrations is likely to be of growing importance"

is,

"...'for a few decades the signal increases faster than the uncertainty, so the signal-to-noise increases initially"

Don't you think that you've led the UK govt astray? I say this because Figure 5: (A) Signal to noise ratio for decadal mean surface air temperature predictions, shows that the Global signal to noise ratio DECREASES from about 2040. Furthermore, we've already had a decade since 2000 where "the signal-to-noise increases initially" but that has not helped the early predictions as this up-to-date plot of IPCC estimates vs observations by Nir Shaviv shows:-

http://www.sciencebits.com/files/pictures/climate/IPCC-Prediction.jpg

BTW, why didn't you show up-to-date observations on Figure 1: Global mean, annual mean, surface air temperature predictions?

At this juncture, for global temperatures to get on the IPCC 'Best Estimate' track there will have to be an abrupt rise in averages in excess of 0.2 C. The only hope for that is a large El Nino similar to 1998 but what are the prospects of that? The JAMSTEC AL prediction (using verified IOD-based methodology) is “A La Nina condition will continue in the [NH] winter and spring, and begin to dissipate in the [NH] summer”. The problem is that the current La Nina is not recharging the Pacific's heat sufficient for an El Nino of any significance (if it even eventuates late 2012).

So what we have is basically continued stasis until some factor forces a climate shift. Your prognosis (by implication) is GHG forced warming but the astrophysicists are predicting that astronomical cycles will have an influence c. 2014. A simple responsible medium-term climate risk assessment by assigning probabilities to the possibilities until the direction of the shift becomes clear is then:-

0.33 warming (GHG forced)
0.33 no change
0.33 cooling (astronomical cycles)

Not to factor in the full range of alternatives would be negligent, I'm sure you agree Ed but that is exactly what the UK Climate Change Risk Assessment (CCRA) by The Department for Environment, Food and Rural Affairs (Defra) has done:-

http://www.defra.gov.uk/environment/climate/government/risk-assessment/

Their basis? UK Climate Projections (UKCP09)

http://ukclimateprojections.defra.gov.uk/

Their basis? HadCM3 climate model and therefore IPCC Special Report on Emissions Scenarios (SRES) i.e. probability factors from the outset:-

1.0 warming (IPCC version)
0 no change
0 cooling

Even though they go to great length to obtain probabilistic projections AFTER their initial assumption:-

3 The construction of probabilistic climate change projections

http://ukclimateprojections.defra.gov.uk/content/view/2017/500/

Negligence, do you agree Ed?

[Richard C (NZ)]

Ed Hawkins at 10:37 AM

The long-term (back to 1659) CET vs CO2 is here:-

http://junksciencearchive.com/MSU_Temps/HadCET_an.html

I have no idea what the statistical correlation is but I suspect it is extremely weak and inconclusive. A simple comparison coveys what could be considered a compelling correlation up to 1949 but it breaks down after that. And this at a time when CO2 forcing was supposed to have kicked in.

Meantime, CET has plummeted back down to 1659 levels since 2006 with little prospect of regaining any correlation to CO2.

[Richard Betts]

Feb 3, 2012 at 5:27 PM | Phillip Bratby

I don't recall UKCIP saing anything like this (or UKCP09). Perhaps Rchard Betts can clarify.

Hi Phillip

UKCP09 did include natural variability and indeed, like the CMIP3 results analysed by Ed, indicated that natural variability is more important than the emissions scenario for the next few decades. Please read the report, e.g.: here.

Having said that, Ed has done a great job at explaining it rather more clearly!!

Cheers

Richard