To be able to see if 3>2>1 we'll need to restrict to individual months for temperature. Let's say we try this with a couple of temperature variables - mean monthly T and highest Tx. I expect 2 to be > 1 and 3>>2, but from looking at the model variants so far climate model uncertainty dwarfs WG uncertainty. Problem is much of this is down to wacky data. I'm beginning to think that the modelling has lost the real-world agreement between variables - which is what the WG assumes to still be the case! Colin has sent revised code to Vaz - it now runs through without hanging. This was only achievable by stopping the checks for silly numbers!
Hat tip: AJ