I’m heading home after giving a public talk, and have a bit of time to write about the recent Armour paper Projection and Prediction: Climate Sensitivity on the rise. It’s basically another attempt to reconcile energy balance estimates for climate sensitivity, with other estimates. Although the ranges for the different estimates do overlap, energy balance estimates (or observationally-based estimates) imply that equilibrium climate sensitivity (ECS) might be lower than the other estimates suggests.
The basic energy balance calculation is
where is the change in forcing after doubling atmospheric CO2, and is the feedback factor. As can be seen in the equation above, this can be determined from the observed the change in temperature , the measured change in system heat uptake rate , and an estimate for the change in radiative forcing, .
However, as Armour (2017) says, when using the above to estimate ECS a
key, but often unstated, assumption:[is that] that the global climate feedback in operation when equilibrium is reached, , will be equal to the feedback in operation at any given time, .
There are, however, indications that the feedback response may not be constant over time, possibly due to the pattern of the warming not being constant. This would imply that the above assumption is not correct and that the ECS inferred from a basic energy balance calculation will not necessarily accurately represent the actual ECS. This is illustrated in the figure below. It shows, as a function of time, (for both abrupt CO2 injections and 1% per year CO2 ramping) the ratio of the actual ECS to the ECS inferred using the energy balance approach. Essentially, the inferred ECS is typically smaller than the actual ECS.
A few additional points. We don’t know that these adjustments are correct. However, we do have a situation where there is a mismatch between different climate sensitivity estimates. We also have plausible arguments that can reconcile these estimates. This doesn’t make this reconciliation correct, but does at least provide arguments for why we shouldn’t dismiss the possibility of climate sensitivity being higher than the basic energy balance estimates suggest – especially as these estimates require assumptions that may not be true (constant feedbacks).
There was another point that I wanted to try and make, but may not do very well, as it is getting late. What we’re trying to do is produce a distribution that gives us some indication of what we might expect climate sensitivity to be. There is, however, essentially only one answer; we just don’t know what it is. We want to use these estimates to inform how our climate might responds to future changes in anthropogenic forcing. In some sense, we may never really know which estimates where right, as we would only really regard the energy balance estimates as having been wrong if climate sensitivity turns out to be very high (say > 4K), and the others as being wrong if it turns out to be very low (say < 1.5K).
However, trying to argue for a reduced probability in some region of parameter space, when we can't yet know that this region is actually less likely, seems – to me, at least – a poor way to inform ourselves. Given that these energy balance estimates do not actually rule out (with high confidence) much of the standard ECS range, and given that there are plausible reasons as to why they might be producing a range that is skewed to lower climate sensitivity values, would seem to suggest that we should be careful of using them to strongly influence our assessment of the expected range for climate sensitivity.
Anyway, my train is almost due in, so I'll probably stop there. Hopefully I've explained this fairly clearly (it's been a long day) and if others have comments, feel free to make them.
Nic Lewis already has a post on how inconstant are climate feedbacks, and does it matter? The answer to the question he poses (according to Nic, at least) is essentially that they are not inconstant and, even if they were, it wouldn’t matter.