Lying With Charts, Global Warming Edition

[Nederland]

72. Actually it is not

I email Lucia to ask her about the differences between her graph and Gavin's, in particular I asked her about why her CI range was smaller than his. This was her response:

Generally, one can chose 95% confidence intervals that are:

1) The 95% spread of all runs in all models
2) The 95% spread in model means. (Spread in model means was what the IPCC used when they actually published their forecast.)
3) A 95% spread based on the variability of model weather in a typical model.
4) A 95% spread based on a statistical model (like ARIMA) and the residuals to a linear fit in models.

I choose #2 because that is what the IPCC used. Gavin used #1 which tends to give the largest spread and has now become popular with defenders of models.


I guess it all comes down to what method you believe is the correct one--the one chosen by the IPCC or the one that Gavin Schmidt cherry picked because it gave him the best results.