2011-11-23 02:05:49Separating signal and noise in climate warming
John Hartz
John Hartz
john.hartz@hotmail...
98.122.98.161

Click here.

2011-11-23 18:30:33
Glenn Tamblyn

glenn@thefoodgallery.com...
124.179.95.162

Lwarence Livermore Mational Laboratory

Lawrence Livermore National Laboratory

2011-11-24 00:39:55
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

A caveat to consider when we reference this article:

The TLT temps are noisier than surface temps, owing to stronger ENSO impact. Therefore extending this to say 'all temperature trends less that 17 years are meaningless' is risky. It may be true, but it's not what the study looked at.

2011-12-04 16:35:40
John Hartz
John Hartz
john.hartz@hotmail...
98.122.98.161

Given that deniers keep hammering on the "cooling trend" meme, it is important that this article be posted.

2011-12-05 17:02:29signal from noise
Larry Wade

lwade@caltech...
66.215.92.124

Another way to look at this paper:

We are studying climate change.  'Climate' describes a 30 year average of weather.

It looks like the defined period is about correct for it to be meaningful.

Climate change:  aptly named, aptly defined.

2011-12-06 22:25:04Do we need devils advocacy?
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

Somethimes I don't know whether I should post a comment on a story or not. Here's one I didn't post.

Yes, scaddenp, that is my understanding too. Indeed the Real Climate FAQ has a question on this: for the handful of processes which have to be parameterised rather than being derived from basic physics, the parameters are determined from control runs to produce a stable climate, not the 20th century record.

However, if I may play devil's advocate for a moment, the use of hindcasts to eliminate mistakes in the physical model is itself a weak form of training, and one which is hard to quantify. Suppose you introduce a better model of a process into the model, and find it no longer hindcasts correctly. Your first assumption is that your modelling of the new process is wrong, when in fact you could have had two errors which cancelled out. It's very easy to fool yourself when dealing with complex systems.

I think there is some evidence that this is the case. The fact that apart from some of the Canadian models, the CMIP3 models all reproduce 20thC climate well, but with different forcing and sensitivities, is at least suggestive. Hansen talks about this in his recent paper, and there is more in <a href=http://www.agu.org/pubs/crossref/2011/2011JD015924.shtml>this paper</a>.

On the other hand, the fact that recent models start from basic physics and the topology of the earth and can deduce the existence of the El Nino cycle is very compelling, as are the IPCC 2007 predictions of regional extreme weather impacts. Thus knowing what models predictions we can rely upon appears to be a rather subtle question - one which is beyond my ability to answer.