Wednesday, December 28, 2011

On the observational assessment of climate model performance

I already blogged about this some time ago, but here is the final published version, with a free download for the first 50 who want it, by using "46672759" as both the userid and password. The title seems to have enticed an unusually large number of people to read it - it is currently GRL's most downloaded paper (at time of writing). I hope they were not all disappointed.

Incidentally, the first draft IPCC report is open for review, but presumably all who are interested already knew that. As per their request, I'll not be discussing it here.

7 comments:

William M. Connolley said...

You'll be disappointed to know that the password is still valid, so demand is not that keen :-)

crandles said...

It has reached 5th across all journals and with another 48 downloads might reach 4th ;o)

David B. Benson said...

Fandom is fleeting.

chris said...

I like the last sentence of your paper, viz:

"We have demonstrated that an ensemble can only be meaningfully challenged by observations which are both distant from it, and precise."

That makes perfect sense. But can't help thinking of those attempts to challenge tropospheric warming in model ensembles with satellite MSU tropospheric temperature reconstructions that were "oh-so-precise", but turned out to be inaccurate.

So whereas your sentence (and nice paper) is a succinct description of the requirements for meaningful challenging of model ensembles, it would also be nice to know in any particular case, that the observations are accurate as well as precise so that we could be confident that a meaningful challenge was in fact true as well as methodologically sound…

James Annan said...

Chris, you are right, I'm perhaps a bit optimistic in using the terms interchangeably, but I'm primarily addressing the situation where the obs are known to be quite inaccurate (at least when inverted to estimate hidden model parameters such as sensitivity) but even so have still been interpreted as casting doubt on the models...

Anyway, we are still number 1, presumably due to the reduced new publication rate over the holiday period :-)

Tom Fiddaman said...

Max downloads exceeded. :) for you. :( for me.

James Annan said...

Ah, that explains why it is no longer the most downloaded paper :-)

But it is freely available on my work web page (link at top).