Last entry for now in the series, although I'll do some posts comparing various predictions too.

Lessons from Past Climate Predictions: IPCC AR4

Where did you get the prediction data from?




AS said on other forum:

Regarding the observations being consistent (temperature wise). I don't think that they are to be honest, Santer et al should have plotted the 95% uncertainty range on the observations too, not just the models. Eventhough others (James Annan and others) have argued against using multi-model mean comparisons with observations and have advocated for the singular runs being the most important, it is an important characteristic that they're matched. If you consider it this way, there's a greater than 50% chance that the models do not match the observations, then you realize that by protecting ourselves with the 95% error bars we are still opening ourselves up for criticism.

For me, and this is purely personal, I think that the fact that the multi-model mean is too hot is proof that we know a lot about the climate because it would be expected considering they don't include the newly discovered negative forcing from aerosols over the last decade. If we get it this close without those aerosols then we're doing pretty good. If they matched perfectly then it would show that our understanding of radiative physics would be off.

Riccardo digitized the A2 multi-model projection for me, Robert.  It's explained briefly in the post.  I know technically I should show error bars, but the intent is just to show that models and data are roughly consistent, which indeed is the case.