Rather strangely, the recent Science Advances paper Nonlinear climate sensitivity and its implications for future greenhouse warming, that’s been getting some attention recently, made it into our Journal Club today (and it had nothing to do with me). It’s, consequently, given me some impetus to write about it. Fortunately, James Annan has already written about it, so I don’t need to say much.
Essentially it uses proxies and models to reconstruct temperatures and forcings over the past 784000 years. Their key result is probably that the climate sensitivity is state dependent; higher in warm climates than in cold. They conclude that in warm climates, the equilibrium climate sensitivity (ECS) is 4.88 ± 0.57 K, which only just overlaps with the IPCC’s likely range (1.5 – 4.5K). What’s maybe more interesting is that they also try to estimate the transient climate response (TCR) which, using paleo data, requires a model of ocean heat uptake. They conclude that it is 2.74K with a likely range from 2.23 to 3.43K. This, again, only just overlaps the IPCC’s likely range (1 – 2.5K).
So, their estimates for climate sensitivity are somewhat high and, I would argue, unrealistically so. A TCR of 2.74K would suggest that we should, today, have warmed by around 1.7K. We’ve only warmed by around 1K. It could be so non-linear that we will suddenly see accelerated warming as we approach a doubling of CO2, but we’re already about two-thirds of the way there, so this seems a bit unlikely. It could be that internal variability is masking a large amount of forced warming, but 0.7K of cooling is much higher than is regarded as reasonable. It is currently regarded as extremely unlikely that more than 50% of the observed warming since 1950 could be non-anthropogenic, so it seems similarly unlikely that non-anthropogenic influences could have induced substantial cooling over the same time period.
I should make clear that there are indications that natural influences have induced cooling, but it’s most likely at the 10% level than at the level of almost halving the forced warming. For example, the pattern of sea surface warming may have enhanced the cloud feedback so that we’ve ultimately warmed less than we might have done. Again, however, the impact is probably at the 10% level over the period of a few decades. It, therefore, seems very hard to reconcile a TCR of 2.74K with current observations.
I think that’s all I have to say. Although the ECS and TCR estimates do just overlap the IPCC likely ranges, it does seems as though these estimates are somewhat higher than is reasonable. Of course, we can’t necessarily rule out such high values, but I wouldn’t regard this paper as a reason to give them more weight than they currently have.