The Blackboard

MarkR (Comment#1389)
March 27th, 2008 at 9:28 am

Hi Lucia

It appears at the very least that the IPCC use of the terms “likely” and “very likely”, in connection with their temperature forecast is false, and in fact the terms should now be in the area of “unlikely” or “very unlikely”. Would you be able to quantify this?

Also, I asked before about whether enough data was available to say whether there was no warming trend since 1998?, and you said no. But I wonder if there is a way to put a number on the likelyhood that a statistically significant non warming trend had occurred? ie very likely to very unlikely, in the IPCC’s terms. Hope this makes sense.

PS Perhaps you could put a link on Tamino for your analysis threads?

lucia (Comment#1390)
March 27th, 2008 at 10:09 am

Mark R– Are you suggesting I visit Tamino’s blog and post links to my articles? Just visiting other blogs and posting links is bad netiquette. I dislike the heavily moderated comments, and figure he’d likely just delete them. It’s his blog, so, his rules. But I prefer not to invest time writing comments over there.

It’s possible to do analyses since 1998. But there is no good justification for picking that year. I picked 2001 because that makes sense for the AR4. So, my evaluation of the IPCC projections related to the AR4, not the TAR.

My readers have since asked me to examine the TAR. I’m trying to figure out the correct year for evaluating the TAR projections, and I’m trying to figure out precisely what the TAR projections are. I’m doing so by reading the contents of the TAR and posting as I go.

It’s clear the correct year for hypothesis testing of the TAR is not 1990 — 11 years before the TAR was written, and well before the models used to make projections were either developed or tuned. (Not to mention before the SRES forcing scenarios used in the TAR were set forth.)

The argument for ‘verifying’ using data starting in 1990 is, simply, flimsy.

The correct year also can’t be after2001 when the TAR was published. The citation for the “tunings” is 2001– but those must have pre-existed the TAR otherwise, the TAR authors couldn’t have used them. So, accounting for the gap in publication times, starting comparisons sometime a bit before 2001 makes sense for the TAR. But how long before 2001? I don’t know yet.

It is highly unlikely that the “correct” year to start evaluations of the TAR, per se is 1998! It’s also not the correct year for the AR4.

But, David Stockwell looked at starting in 1998 on the principle of “multiples of 10 years.” You can read what he says here.