Warming rate probably accelerating

 Another in depth statistical analysis from Open Mind (Tamino).  I won't repeat the whole article (best to read it for yourself).   In essence, what he does is look at the change in the temperature anomaly over 30 years, the conventional measure of climate change (because shorter periods are too susceptible to fluctuations from natural climate cycles such as, for example, El Niño/La Niña).    This chart shows what that looks like:

Note that the change over the 30-year span has been centred at 15 years,
which is why it apparently ends at 2009.

He then adjusts the data for volcanic eruptions, El Niño, and the sunspot cycle to test whether the change in trend since 2000 is statistically significant, and concludes that it is.

But wait, there’s more. We know the cause of some of the fluctuation, and we can adjust the data by removing our best estimate of those known factors. In particular, we can remove an estimate of the influence of volanic eruptions, solar variations, and the el Niño southern oscillation, and I’ve done so by a modified version of the method of Foster & Rahmstorf.

Comparing this graph of adjusted data since 1946, to the graph of raw data over the same time (the third graph in this post), two things are obvious. First, the level of scatter (of random fluctuation) is quite a bit less for the adjusted data than for the raw. Second, the “obvious” simple model for the raw data (a linear spline with two straight-line setments) isn’t so obvious for the adjusted data.

The same analysis used on the raw data, returns different results on the adjusted. This is mainly because the uncertainty levels are so much reduced. The warming rate for NASA data, for example, now looks like this:

Again, the change has been centred midway on the 30-year span

The final 30-year estimate suggests the rate was definitely bigger then 0.02°C/year, while the earlier rate is definitely less. Also, the statistical tests (fitting a parabola and a linear spline) now are definitely significant at 95% confidence.

Using adjusted rather than raw data, the same is true for all five data sources. While the significance is weakest for HadCRU and strongest for NOAA, it’s over 95% confidence for all of them. My conclusion is that recent acceleration of global warming isn’t just likely, it’s confirmed.

There is still room for doubt, if you doubt that the adjusted data represent things correctly then the recent apparent acceleration may be just random accident. All told, I find that too unlikely.

See also this piece I did, based on Open Mind's analysis: 

Terrifying acceleration in global heating

We are heading towards climate catastrophe, and still we do not act, stumbling blindly to the precipice.