He suggested in his comment that Roger should use this Hansen data as a forecast verification, and one should compare like with like. Well…. here it is!

The fact is: unless we make a squished little graph and add the apples-to-oranges met data comparison, the Hansen projections don’t look particularly good. They aren’t horrible but they do indicate the warming is on the low end of what is projected, and they suggest that Hansen’s 1988 paper over-projected warming. Why did GISS Model II over project? Beats me!

Does Model E over-project? Beats me!

Of course, we still have experienced warming, and the models projected warming. So, qualitatively the models agree with the data. But, that doesn’t make them much better than people’s estimates based on 0D or 1D models. (I base my sense that warming is probable on simple models and the warming trend.)

Gavin’s reaction (as I said at CA this morning) to this whole thing has been most interesting, to say the least.

You and Dr. Pielke are right though, it’s a cherrypicker’s dream.

The black line connecting data does bend more toward the lower emissions scenario. Since the IPCC updates these every 5-7 years, we always have a new prediction before there is enough data to do any meaningful statistical tests on the old prediction.

I also still haven’t done any meaningful statistical tests on comparing the Hansen projections to the data– it would require dealing with the inter-annual autocorrelation that remains even after one removes the trend.

But, what’s worse, I’m not sure quite how to deal with the problem of putting everything on the same basis with regard to defining the zero line for temperature anomalies.

It certainly doesn’t make sense to have scenarios and experimental data all have different baselines and then compare. After all, in that case, the crappy agreement between the Scenarios and the data might just be due to different baselines!

Statistical comparison would be greatly simplified if these predictions provided real temperatures. Then, the temperatures would just be temperatures, and I would need to worry about the meaning or import of shifting the — as far as I can tell– arbitrary zero point for the baselines!

I find the IR4 Figure TS.26 most interesting.

Note how the 10 year average temperature trends since 2000 seem to following the ‘orange/no CO2 increase’ line?

It is way too soon to draw any conclusion about the trend but the divergance is interesting because a similar divergance was seen by 1990 when you look at Hansen’s 1988 predictions.

Someone who is predisposed to believe in warming will probably assume that the last 10 years are just an inflection point in the graph like we seen in 1992. Time will tell.