Its probably a measure of how accepted the bulk of the AR4 SPM is, that the most interesting discussion about it seems to center around the sea level rise uncertainty ranges. There does indeed seem to be some confusion here… RP Jr explores this, and points out that it would have been nice had the IPCC made a comparison easier, with which I agree (and complains about the take on this from the most authoritative source, of course RC; though as far as I can see he is wrong to say that we assert that the range isn’t lower).
But (before venturing onto the minutiae) I’ll say that as far as I can tell the main message is For each scenario, the midpoint of the range in Table SPM-2 is within 10% of the TAR model average for 2090-2099. So whatever disagreement is small; and likely to be due to estimates of the range rather than the central value.
The other point to make is what we’re discussing here is *not* what the septics have been pushing, which is based on a simple misreading of the report (thus RPs Thus, I conclude that the top end estimate has indeed come down from the TAR to the AR4, and those making this observation are accurately representing the AR4 is a touch disingenuous, because there are two schools here: those correctly noting a lowing of the range and those incorrectly touting nonsense that the value is halved: the latter sadly including the WSJ).
The AR4 itself says The ranges are narrower than in the TAR mainly because of improved information about some uncertainties in the projected contributions (which on reflection probably suggests that the narrowing of the range is real, unless of course they are wrong :-). And the converse of this is that the *minimum* rise has increased, from 0.09 TAR to 0.18 AR4 (I don’t see the skeptics trumpeting this; nor anyone else really as its too small to be exciting).
I can’t claim to fully understand the differences between the two sets of numbers. I agree it would be ratther nice for them to be comparable, but making things back-comparable can be tedious and (from the point of view of the science) largely pointless.
But there are two other points of tedia that may affect the comparison: As Gavin points out to RP in a comment (and which RP uses in his picture, but not within his text)they change to 2090-2099 from 2100 for the headline changes, and the reduction to 90% from 95% in the uncertainties. Switching to a 2090-2100 average knocks off about 5 y of SLR off the top end, which is about 0.04m (although there is a smaller opposite adjustment for the other end). And the change to a 90% range reduces the values somewhat (but by how much?).
So… bottom lines: This is a storm in a teacup. But we should get it right. Comparison isn’t easy. But since IPCC themselves say the ranges have narrowed, they probably have.