The Science Media Centre has put out a response to the GWPF report here. I was struck by Myles' Allen's contribution:
Their [TCR] prediction of 1.35 degrees C is, even if correct, only 25% lower than the average of the general circulation models used in the IPCC 5th Assessment. A 25% reduction in TCR means the warming we might have expected by 2050 might take until the early 2060s instead. Their 5-95% range of uncertainty in TCR (kindly provided by Nic Lewis) is 0.9-2.5 degrees C, almost exactly in line with the range of the models shown in their figure (1.1-2.6 degrees C).
Compare this with what is said in the Lewis/Crok report (long version, p. 49):
CMIP5 climate model TCRs are on average 35% higher than 1.35◦C, at 1.8◦Cor so,with the TCR for particularly sensitive models substantially higher than that (the UK Met Office HadGEM2-ES model has a TCR of 2.5◦C).
So is it 25% or 35%? Here's Lewis/Crok's Figure 6 (which is based on Table 9.5 of AR5), which provides some perspective on the figures:
;)
But this is not the only problem with Myles contribution. As is pointed out in the Lewis/Crok report (long version, p.49), the amount of warming produced by the CMIP5 models is actually much higher than you would immediately expect from their TCR figures.
Using data on simulated warming over similar periods for all the CMIP5models analysed in Forster et al. (2013),model average effective TCRs of 2.0◦C over the instrumental period, and 2.2◦C from the 2000s to the 2090s, can be estimated.
So in practical terms, the model TCRs are of the order of 50% higher (2.1/1.35=1.5) than the observational studies.