The Cherry-Picking Climate Change Committee
By Paul Homewood
Dr David Whitehouse has responded to the Committee on Climate Change’s response to Owen Paterson’s GWPF lecture last night.
In its response to the 2014 Annual GWPF Lecture given last night by Owen Paterson, the Committee on Climate Change (CCC) has issued a statement to what Mr Paterson said. It was immediately taken up by the twittersphere, with some calling it “robust and thorough.” Unfortunately when it comes to Paterson’s claim that the earth’s surface temperature hasn’t risen in 18 years the CCC shows itself to be out of step with what has now become mainstream scientific opinion.
In any physical measurement of continuous variables in the real world, especially the various temperature measurements used as climate change metrics, it is meaningless to quote a measurement without its associated error. Thus the CCC in quoting a 1998-2013 annual average global surface temperature ‘trend’ of 0.04 deg C per decade (based on HadCrut4 data). According to the CCC this is proof that Paterson is wrong. In reality, the CCC is being misleading.
In any time series it is unlikely that over a selected period the gradient would be zero – even if the statistical properties of that time series are stationary. In other words, just finding a gradient doesn’t mean there is a real trend. The reason is simple: If the gradient falls within the errors of measurement then the standard scientific practice is to conclude that there is no statistical evidence for a trend. The 0.04 deg C per decade figure quoted by the CCC should have an error of +/- 0.15 deg C, a figure that is four times larger than that quoted by the CCC. The reason why the CCC finds a (statistically insignificant) positive gradient at all has nothing to do with global warming but the presence of two La Nina years – 1999 and 2000.
Here is HadCrut4 from 1998 to the present day. The data ranges from a temperature anomaly of 0.14 to 0.85 deg C with no statistically significant gradient. There is no trend. Click on image to enlarge.
The CCC should have also followed standard scientific practices. Firstly, starting any trend analysis with 1998, which was a strong El Nino year, is clearly nonsense. Trend analysis should also vary the start and end points to look at how inter-year variability might affect the outcome. Carry out the same analysis for 2001-2014 gives a cooling of 0.02 deg C per decade (+/- 0.17 deg C). Starting with 2003 gives a cooling of 0.047 +/- 0.22. Starting in later years yields with one exception a cooling “trend.” Note all errors are 2 standard deviations.
So the CCC has been unscientific, cherry-picking and misleading. If a thorough analysis of HadCrut4 is performed over the past 17 years there is no statistically significant trend, either positive from 1998 or negative from 2001 onwards. The claim that global temperature is rising in this data set is absolutely rejected. This is not a slower rate of temperature rise, the data show no temperature rise in this period – period.
Of course the world has warmed during the period in which global estimates could be made, post 1880. One can go back before the last 17 years and show there has been a rise and dilute the impact of the recent standstill. But there is no getting away from the fact that after 40 years of no change in global surface temperature (1940-1980), it started to rise again for 17 years (1980-1997) but has been constant since then. The rise and the subsequent standstill are now of equal length. It is curious then that the CCC say that the “temporary slowdown” has not been long enough to alter projections of future change when those very projections were based on a period of equal duration.
Only campaigners, committees and some commentators use misleading phrases like ‘slowdown’ or ‘warming trend.’ In the peer-reviewed scientific literature, most scientists call it a temperature “hiatus” or “pause.” In fact, the global warming pause has become, as the journal Nature acknowledges, the biggest problem in climate science. It is a real problem that the CCC think it doesn’t actually exist
Comments are closed.
NOT A LOT OF PEOPLE KNOW THAT 
Bunch of jobsworths – are there any real scientists amongst them?
No – AGW is just grant fodder; the “scientists” are however clever enough to know that if they admit the gig is up, the bank account goes down.
“There is no trend. Click on image to enlarge.”
I wonder why they didn’t just add a trend line to the HadCRUT4 chart, since they were using WfTs and the facility is there to do so: http://www.woodfortrees.org/plot/hadcrut4gl/from:1998/plot/hadcrut4gl/from:1998/trend
The fact is that there *is* a trend, currently +0.06C per decade (to August 2014; for some reason WfTs hasn’t updated yet). Anyone can use that link to switch between the other global data sets and find that a +0.05C/dec trend occurs in UAH (the other main satellite producer) and in GISS it’s +0.07 C/dec. WfTs doesn’t host NOAA, but it also has a +0.05C/dec trend since 1998. In fact, the only data set that doesn’t have a warming trend in its best estimate data is RSS.
It is true to say that the trends in HadCRUT4, UAH, GISS and NOAA aren’t statistically significant; but that’s not the same thing as saying “there is no trend”. Furthermore, if these data sets do start showing a statistically significant trend in the next year or so, does that mean that the GWPF will accept that there is global warming? They could be boxing themselves in.
He makes it absolutely clear.
“No statistical evidence of a trend”
That’s one quote Paul, the other is the one I used above: “There is no trend”. It’s fair enough to point out that there has been no statistically significant trend; but to say “there is no trend” is just incorrect.
It’s interesting to note this statement made by GWPF: “…after 40 years of no change in global surface temperature (1940-1980), it started to rise again for 17 years (1980-1997) but has been constant since then.”
In the case of UAH, the rate of temperature rise between 1980 and 1997 is actually slower than it was between 1998 and the present. What are we to make of that?
There is always going to be some sort of “trend”. If it was downwards, would we be arguing to put more GHG into the atmosphere to avoid an ice age?
Obviously not.
You have to remember the context to this.
Owen Paterson has argued that the Climate Change Act should be suspended, partly because global warming has not occurred “in 18 yrs”.
Gummer has argued that we need to carry on with the economy wrecking measures because the original projections of catastrophic warming are still valid.
OK, we have 0.04C /decade warming over 18 yrs. So who is nearer the truth?
I think you meant 0.04C/decade. Am I right? You will have noticed that none of the people who present plots of trends, such as WFT, include confidence intervals for that trend line, or for individual observations. Of course, with time series you cannot just produce a new “Year” within the series. It is historical and fixed, though some organisations are willing and able to change observational data, as we all know. However, one can extrapolate the trend line and its confidence intervals to a future date (that’s why people produce trend lines – trying to guess the future). But no-one in the climate sphere, as far as I am aware, comes forth with the confidence intervals for a future single observation, such as July 2015. My software does that and i wonder why others’ don’t. When you look at the confidence intervals for a single future observation you are sure to be impressed by their width, something that renders them effectively useless for practical purposes such as contingency planning.
There is another complication, that of autocorrelation in the observations. This is very common in climate data, and it has the effect of further widening the confidence intervals for a prediction. Quenouille’s correction is sometimes used to allow for this autocorrelation.
Of course, the underlying reason for being sceptical about trend lines is that they are inevitably based on a “model”, the simplest of these being the straight line model, which can clearly be observed to be inadequate for most data sets. Nevertheless, it has its advantages, primarily simplicity, but also it is perhaps the safest or least controversial model. Nevertheless, beware of linear models, but use them until something better presents itself.
Yes thanks, Robin
I meant 0.04C/decade. I have now amended
All statistics done on climate time-series data are mathematically unreliable due to the proofs of basic statistics being based on independent data rather than time-dependent data. So Confidence Intervals and Significance Levels should be extremely large. On top of that, you need the measurement and calculation error margins which includes things like not enough weather stations, broken or degraded equipment, poor ‘autocorrect’ algorithms (some may even have a warm bias!) etc.
So why is anyone bothering with a hundreths of degrees?