Since my holiday is now over, I thought I might briefly comment on a recent paper by Cheng et al., called Observed and simulated full-depth ocean heat-content
changes for 1970–2005. John Abraham, one of the authors, has already discussed it in his Guardian article, so I won’t say too much. The basic result is that
What’s more interesting, though, is that they compare this with the results from climate models. Although there is a spread in model results, they do find that
The full-depth OHC has increased by 0.74 [0.68, 0.80] x 1022 J yr-1 (0.46Wm-2) and 1.22 [1.16–1.29] x 1022 J yr-1 (0.75Wm-2) for 1970–2005 and 1992–2005, respectively….
the ensemble median has excellent agreement with our observational estimate: 0.68 [0.54–0.82] x 1022 J yr-1 (0.42Wm-2) from 1970 to 2005 and 1.25 [1.10–1.41] x 1022 J yr-1 (0.77Wm-2) from 1992 to 2005.
and, consequently, suggest that
OHC be a fundamental metric for climate model validation and evaluation, especially for forced changes (decadal timescales).
which seems related to something Roger Pielke Sr and I wrote about a while ago.
I think this paper is very interesting, especially as Anthropogenic Global Warming (AGW) is fundamentally about energy; our emission of greenhouse gases (GHGs) drives the system further and further out of energy balance, with most (~93%) of the excess energy going into the oceans. Therefore, as the paper suggests, using Ocean Heat Content (OHC) to assess climate models makes a great deal of sense.
Also, since the energy imbalance is largely driven by the change in forcing and the resulting feedbacks, if the models are doing a reasonable job of representing the change in total energy, then it provides some confidence that estimates for equilibrium climate sensitivity (ECS) are reasonable. The transient climate response (TCR) is maybe less clear, but if the models do a reasonable job of representing the distribution of energy within the system, then it should also provide confidence in TCR estimates. As the table below shows, the models quite closely match the full-depth OHC change, as well as the 0-700m OHC changes.
To be clear, though, some models suggest a larger OHC change than has been observed, and some suggest less. As the paper itself says, [t]he CMIP5 models show a large spread in OHC changes, suggesting that some models are not state-of-the-art and require further improvements. However, that the ensemble median is in quite good agreement with the observations provides some confidence that some models are doing pretty well, and that you can at least use the ensemble to assess AGW. Also, this whole process of comparing observations and models is a crucial part of doing research. If we had multiple Earths and a time machine, it could be pretty easy to develop the ideal model. Since we don’t, it’s not easy and takes time, as we both collect more data and develop new methods for analysing the data.