Wednesday, October 31, 2012

David Rose, Not so Sweet

A few weeks ago David Rose of The Telegraph made a stir, claiming that global warming had been nil over the last 16 years.

It was complete crap, not worthy of a so-called journalist.

What was that claim based on? Only this:

This is a plot of the slope (linear trend) of the HadCRUT4 surface data, from the indicated date to the present.

So, from January 1990 to August 2012, the linear trend is 0.144 C/decade, with a 95% confidence level of 0.024 C/decade due to uncertainty of the statistics of linear regression. (HadCRUT4 does give the measurement uncertainty of each data point, but they quickly become less than the statistical error -- for the 20-yr slope the measurement uncertainty is about 0.002 C/decade, while the statistical uncertainty is 0.015 C/decade.)

The "true uncertainty" is the square root of the sum of the squares of these two numbers, which is essentially just the statistical uncertainty (plotted above).

So David Rose's claim of no global warming is dependant on this small, tiny, temporary period between 1997 and 1998 where the 95% CL error bars reach barely into the x-axis.

That's all. That's it.

That's why Tamino calls Rose a cherry-picker -- justifiably.

You really have to wonder how some people sleep at night. You really do.


charlesH said...

Do I understand your graph?

Starting from x to present one gets a trend increase/decrease [With one or two exceptions?]?

typical trend increase

typical trend decrease

DocRichard said...

You are of course right, David Rose is trying to make a silk purse out of a sow's ear, total warming continues when ocean heating is taken into account - and yet... the models do not predict the recent slowing in the rate of warming. Because they do not, the contrarians have been able to make hay, and the more illusory hay they make, the less effective the response of international policy makers becomes.

Therefore it will be better when the models are able to predict the kind of variations in the rate of warming that we have seen since 1997.

It looks as if the fluctuations are related to ENSO.

It might be a good idea if the modellers turn up the gain on ENSO effects. Although it is impossible to predict exactly what ENSO will do, by putting in an estimated ENSO wave, the models will be more likely to anticipate increases and decreases in rate of global temperature increase. Which would be helpful. What do you think?