Saturday, April 22, 2006

Science on Hegerl et al (and Annan & Hargreaves)

Science has a review of Hegerl et al, which includes a brief nod to our work too, and a handful of other researchers endorsing the method and at least her results, if not explicitly ours :-)

While it all looks pretty encouraging, it is still some way from a clear confirmation of Gavin Schmidt's view that "Basically no one really believes that those really high sensitivities [measured in the past five years] are possible," and that even Hegerl et al's high estimate is unrealistic. Certainly I think Gavin is right - but as I said recently, I realise that our view is still somewhat of an outlier among people working in this sub-field. For example, not that long ago, Michael Schlesinger was actually claiming that climate sensitivity was "likely" (70%) greater than 4.5C, and he sounds unconvinced by Hegerl et al so far ("I'm not comfortable with the results."). It is possible that he had only had time to glance quickly over the paper before being asked for a quote - my experience is that it can take a few days for the full force of our arguments to sink in. Anyway, it remains to be seen whether, and how quickly, he modifies his views. I do sense some turning of the tide.


Anonymous said...

This CPDN thread seems to have developed into a discussion about Hegerl et al and Annan & Hargreaves.

Dave seems to have avoided answering the question of whether he was waiting for Julia but perhaps this is just because there is more interesting things to discuss.


James Annan said...

Thanks for that Chris - I don't think I'll bother starting on another front - at least, not there and not yet :-) There's still plenty of discussion to be had though - and in fact there is a workshop in the UK at the end of June (specifically on uncertainty and climate change) which should be useful.

Anonymous said...

Wonderful to read some actually reassuring news on the climate change front.

As a journalist, I have to ask the obvious-but-probably-dumb question:
how can hard numerical values be put on global temperature estimates with as vastly many variables as future climate change?

James Annan said...


I'm not sure that I understand the question. It's an estimate of a temperature change, which necessarily means it's a hard number. The uncertainty interval is basically a measure of our confidence.

Anonymous said...

Poor phrasing on my part. Of course temperature estimates have to be a hard number. My question was in reference to the probability estimates mentioned in the Scientific American review: "In this week's issue of Nature, they report a 5% probability that climate sensitivity is less than 1.5°C and a 95% chance that it's less than 6.2°C."

I ask because I've encountered skepticism about GCM forecasts, and not just from the general public, but from working climatologists as well.

Can you talk a little about how researchers arrive at those probability numbers? Thanks...

James Annan said...

Ah, that makes more sense :-)

These sort of estimates are tyically based on observed/estimated past changes in temperature and how they relate to the observed/estimated forcing at the time. For example, if a -1W reduction in solar output (during the maunder minimum) resulted in a 0.7+-0.2C temperature change (at 5-95% confidence), then +4W from doubled CO2 can be expected to give 2.8+-0.8C warming - ie an estimate (in this example) that 2 < S < 3.6 at the 90% level.

That's the basic idea, anyway - thermal inertia of the earth system makes it a bit more complex in practice, since it is rarely in a true equilibrium. For the most part, these calculations don't actually depend on complex GCMs at all, but rather simpler energy balance models - and the biggest uncertainties are in how well we can estimate the historical forcings and temperature changes, rather than the modelling itself.