Email: m.prosser@pgr.reading.ac.uk
On the recommendation of my supervisor, I along with Javier Amezcua, Vicky Lucas and Benedict Hyland represented Reading Meteorology at the Institute of Physics (IOP) “Studying the Climate: A Challenge of Complexity” conference on February 6th, 2020. The programme and speaker list can be viewed here. It was a fantastic set of speakers delivering many a killer point in front of an engaged audience.
While some may consider the philosophy of science a complicating distraction, I think I ignore it at my peril. Certainly climate science is not without its philosophical issues; one might even say it is riddled with them…
David Stainforth (LSE), the keynote speaker stated it thus:
“The study of anthropogenic climate change presents a range of fundamental challenges for scientific and wider academic inquiry. The essential nature of these challenges are often not well appreciated.”
So how does climate science compare with other natural sciences? Opinions abounded, but here are just some I can recall:
1 – We can’t really conduct controlled experiments in the way that other natural scientists can, as we have just the one Earth and can’t turn back time (we have to beware of post-hoc explanations, and some of our predictions may never be verifiable/falsifiable).
2 – We therefore rely heavily on numerical models.
3 – We’re also doing our science while the climate is changing around us, and thus there is a strong sense of urgency.
4 – There is therefore a pressure to be multidisciplinary.
On a more practical side, David’s talk left me with a novel way of thinking about ‘climate’. Thinking about a climate metric, such as temperature, I would have thought hitherto of simply a mean and a standard deviation (a very Gaussian way of looking at it!). But David argued that climate is often best conceived of as a more generalised distribution. While a bell curve is symmetric, unimodal, a distribution need not be (and this can be true in the climate system). Studying and predicting a stable climate distribution may already be difficult but studying and predicting a changing one is even harder!
Now for a bit of a whirlwind tour of other arguments/points. There was Reading’s own Ted Shepherd arguing that in climate science we often over focus on avoiding false positives (type I errors) at the expense of incurring false negatives (type II). In other words, we get reliability at the price of informativeness, especially at the regional level where policy makers are somewhat eager to be informed.
Then there was Geoff Vallis (University of Exeter) who posed the question “If models were perfect, would we care how they worked?”. Perhaps a pertinent question, as there appears to be a trade-off, an inverse correlation between complexity of models and our ability to understand them. If the models became so complex that they were beyond the abilities of any human past or future to comprehend, what would we do then? If they become as complicated as the Earth system itself, surely we would have long since lost any grasp on them? Indeed, models already appear to be predicting phenomena without us understanding why. Complexity is not necessarily accuracy (How do we assess accuracy in climate science?) and Erica Thompson (LSE) highlighted the importance of ‘getting out of model land’, and staying with the real world, something some of us may need occasional reminding of.
What even are models? Two expressions given were ‘book-keeping devices’ (Wendy Parker) and ‘Prosthesis of your brain (Erica Thompson). No doubt there were others.
Marina Baldissera Pacchetti (University of Leeds) talked about her work on climate information for adaptation that gives us: “guidelines on when quantitative statements about future climate are warranted and potentially decision-relevant, when these statements would be more valuable taking other forms (for example, qualitative statements), and when statements about future climate are not warranted at all.”
In the afternoon, there were breakout ‘lightning’ discussions. We could choose to join 1 of the following 8 groups:
1. Should we aim to estimate the mean/expectation behaviour of the climate or focus on the worst-case?
2. Is the way we go about climate science now the only way of doing it?
3. If our computers were infinitely fast, what science would we do with them?
4. If our models were infinitely good, what science would be left to do?
5. What fact, if only we knew it, would have the biggest impact on climate change?
6. How should climate science approach the question of geoengineering?
7. What is the benefit to society of general circulation models?
8. What is the public needing to know, and are we working enough on these questions?
My group was 3, but we ended up accidentally merging with 4 and made for a very interesting and varied discussion!
Which group would you have been most pulled towards had you been there? What philosophical thoughts on climate science have you had? What do you think is the most under-appreciated? I would be interested to hear your thoughts.
Many thanks to the event organiser Goodwin Gibbins (Imperial) and all involved for a thoroughly enjoyable and stimulating day.
If anyone would like to get more into the Philosophy of Science, I would recommend this thoroughly engaging 10-hour course of lecture by the Uni of Toronto on YouTube, the trailer of which can be viewed here.
The Social Metwork 