Last week’s Royal Meteorological Society National Meeting took us on a whirlwind tour of the Earth’s climate in the deep past.
In his introduction, Alan Haywood described the timescales on which palaeo studies operate. Most of the Earth’s past has been spent in a ‘Greenhouse World‘ state, where there is no ice anywhere on the planet. The palaeo record takes us back to the Greenhouse World and the time of the dinosaurs, 40-80 million years ago. Then, 15 million years ago, ice began to grow at the poles and the Earth became an ‘Icehouse World’. Note this is not the same as an ‘Ice Age’. We are in the Icehouse World right now, because there is ice at the poles, but the ice extent is much lower than in an Ice Age.
Studying climate data from these periods can give us useful information about how the Earth’s climate system behaves. Crucially, as Alan Haywood pointed out, they provide an ‘out of sample test’ for climate models. Observations of the present-day and the recent past are used in the development of climate models, since a bare minimum requirement is that it should simulate the main recognisable features of the present-day Earth. One must be careful when comparing the results from climate models to these observations because the same observations could have been used in developing the model, generating a false sense of confidence in the model. Palaeo data are generally not used in model development.
There are some disadvantages to using palaeo data. The data themselves are often quite uncertain, and the coverage across the global can be very sparse indeed. These data aren’t really observations. They are based on proxies – indicators of previous climates such as remnants of tiny animals and bubbles trapped deep in ice sheets. Observations do not offer a gold-standard absolute truth. In reality, observations themselves are uncertain and this must be taken into account. The problem of observational uncertainty becomes even greater for the deep past. Several times during the meeting this theme emerged: if a model is shown to be different from proxy-derived observations, this isn’t necessarily evidence that the model is wrong. The best truth we can arrive at lies in a subtle synthesis of evidence from both models and proxies.
Aisling Dolan (University of Leeds) presented some multi-model simulations of the climate during the Pliocene (around 3 million years ago) from the PlioMIP project (Pliocene Model Intercomparison Project). She showed that there were substantial discrepancies between the models and proxies and among models themselves. The uncertainties in the models and the proxies were large enough to explain this discrepancy. This is one of those – irritatingly common – situations in climate science where little information can be gleaned. The large uncertainties don’t mean the models and proxies agree: merely that they cannot be said to be inconsistent with each other.
She mentioned something interesting about the PlioMIP experimental design. The point of a ‘MIP’ is the run different models under exactly the same setup so we can understand where models disagree and hopefully work out why. The PlioMIP setup involves using constant concentrations of greenhouse gases, solar activity and so on. These constant values represent an average over a long period. But, as Aisling pointed out, our observations of the Pliocene don’t represent an average over a long period of time. For example, the average carbon dioxide concentration used in PlioMIP could represent the concentration 3 million years ago, while the average solar activity could look like 3.2 million years ago. So we are not really giving models a fair chance here.
Matthew Pound (University of Northumbria) spoke about the Miocene, a very interesting period in the record where there were potentially low carbon dioxide concentrations but a very warm climate. He found that a model could not reproduce Miocene vegetation using low carbon dioxide. Once again, though, he pointed out the huge uncertainty in the measurements of carbon dioxide during this period. One reconstruction of global carbon dioxide concentrations in the Miocene puts them at 400 ppmv – at which level the model does a reasonable job.
‘Not inconsistent’…uncertainties strike again. But it is often the job of science to accurately determine the level of our ignorance rather than the level of our knowledge.
Palaeo data can help us understand the sensitivity of the Earth’s climate to changes in carbon dioxide. Mat Collins (University of Exeter) spoke about using these data to calculate this sensitivity. The IPCC express it as a global-mean temperature change for a doubling of carbon dioxide, and put the likely range at 2 to 4.5 degrees C. He noted an interesting conflict: palaeo data suggest climate sensitivity is rather high, whereas recent observations suggest it is on the low end of the IPCC range. There is still plenty to be done on the climate sensitivity issue.
The meeting closed with a panel discussion which summarised the day’s talks and gave the speakers a change to agree or disagree as they pleased. It was a fascinating discussion: informal, yet well-structured, engaging and informative. Probably the single best section of any RMetS meeting I have been to, in fact. Some of the discussion centred on the idea of finding, sometime in the past, an analogue for the climate of the present or the near-future. The idea is that, by studying the history of this period, we can gain some insights into current climate and how it may change. The problem here is that it is very unlikely that such an analogue exists. Humans have altered the Earth’s chemistry and biology in such novel ways that we leave quite a unique signature on the planet – leading some to call this age the ‘Anthropocene‘. Ed Hawkins, a scientist from my Department who was also at the meeting, added another point on Twitter.
— Ed Hawkins (@ed_hawkins) February 13, 2013
Most of the time, the climate of the past varied quite slowly, and so at any point in time we could say the climate was roughly in equilibrium. In reality the climate always changes slightly, but most of the time it exists in a relatively stable state. Compare this to the present-day. We know the present climate is very much not in equilibrium, and the changes are likely to become more rapid in the future. Studying equilibrium climates (which may remain stationary for long time periods) may not tell us what we need to know about climate changes over short timescales.