15 March 2022

Talking about modelling without modelling

Years ago, I was asked to teach a fourth year module on climate. It started in the second semester, so after the January exams. The school was feeling cheap, so they hired me by February 1. I had never taught on this module before. I had little choice but to take the materials I had inherited from my predecessors and just teach that. That year I fried my brain; it is difficult to keep up with a module at that level if it is not slap bang in the middle of your expertise, and if you don't get time to prepare.

I've been teaching on it ever since, and I now have had time to think about what it is I want to teach and teach that. Things changed, though, when the module organisation went to my colleague Mattias. I had to rethink my contribution to the module. Last year I taught on it, but I wasn't quite convinced I had found the final shape of my material yet. It basically was my job to talk about palaeoclimate; my colleagues will talk about the physics of climate. But there is quite a lot to say about palaeoclimate; which parts should I talk about? But thinking about palaeoclimate for my side gig gave me a new perspective. And I had an idea.

The reason we are also interested in climate is, obviously, that it is vitally important to our lives, and also changing rapidly. It is important to know what we are changing it into. The only way of doing that is modelling it, but the thing with models is, that if you put garbage in, you get garbage out. You need to be able to understand how climate works to be able to turn it into the equations of a climate model, you need to have good data from the past with which you can check the performance of your model. If it can recreate the climate of the past, then you have a good reason to believe it can also predict the climate of the future.

Decades ago, some people decided it would be a good idea to get as many research groups together who are engaged in climate modelling, and get them to model the same periods of time. Afterwards, they will be able to compare the various results, and used that to figure out which parts of the climate models robust, and which model might need tweaking where. This was the beginning of the Palaeoclimate Modelling Intercomparison Project. Initially, they didn't compare many periods, and these periods weren't a very long time ago; they restricted themselves to the present day, 6000 years ago in the early Holocene, when it was a bit warmer than it was in preindustrial times, and 21,000 years ago, at the end of the last ice age. And they stuck with that in the next two projects. Climate modelling progresses rapidly, so these intercomparison projects have to be done again and again. At the third iteration, the last millennium was added as a period to model. This is the period in which human influence becomes dominant over natural climate forcing. You can calculate how much effect, for instance, changes in the orbit of the Earth, or sunspots, or volcanic eruptions, or El NiƱo have on the energy budget of the Earth in W/m2. And you can do the same for greenhouse gases. The latter now have a bigger effect, and dominate everything. So if you can model the last thousand years, then you have clearly managed to put these factors successfully into your model.

Over time the people involved realised that maybe, they needed to model more periods than that; We are interested in what happens when climate gets considerably warmer than preindustrial times, and neither the last 1000 years, nor the early Holocene or the last glacial maximum are good times for that. In the early Holocene it was less than a degree warmer than in preindustrial times, so basically colder than it is when I write this. And, of course, the coldest part of a glacial period is not good analogy for our future. So decades after the first PMIP, the number of periods studied was greatly widened. They had to get back into the past a lot deeper!

Additional periods studied were: the last transition from ice age to current interglacial 26.000 years ago to the present day, a period of profound warming, but only to levels less warm than preindustrial times;  the previous interglacial at 127.000 years ago, when it was only about a degree warmer than preindustrial times, so similar to the present day. Sea level was some 6 to 8 m higher than it is today; that is something to think about. But they had to get back further in time. The next step was 3.2 million years ago. That is before the Quaternary; in this period it was a few degrees warmer than preindustrial times, but CO2 levels in the atmosphere lower than they are in the present day. The climate system has not caught up yet; we haven't reached the temperatures we had back then, but if we only model a time period like this we are still being very conservative in what sort of climate change we get ready for. So the project goes further back.

The next periods fall under the "deep time" umbrella; three periods from since the extinction of the dinosaurs are included in the current project. They are in the Miocene: 23 to 5 million years ago; the Eocene-Oligocene transition around 34 million years ago, and the warmest part of the Eocene 55-50 million years ago. If you go that far back in time you finally reach temperatures and greenhouse gas concentrations we haven't already reached, or might be reaching in no time. And the complicating factor is of course that if you look at this kind of timescale, a lot of parameters will be different. Ocean currents ran completely differently 50 million years ago, because of the different configuration of the continents; for instance, seawater could freely flow from the Indian Ocean through the Mediterranean Sea into the Atlantic, and could continue westward and flow between North and South America, as these two were not connected yet. But no one said it would be easy. And it is also the fact that things happened then that haven't happened in the periods earlier included into PMIP that makes this so valuable. We need to know what the ice sheets in Antarctica and on Greenland will do when we keep raising temperatures. We can get a bit of an idea from the last interglacial with its sea level metres higher than today, but we will want to know where this ends with our current atmospheric composition. And there are more feedback mechanisms we need to look out for; one I am worried about is methane release from seafloor sediments. They have been known to be associated with spectacular climate events; it is important we model periods in which this indeed happened, so we know are climate models can predict it.

I decided that I would follow these periods and discuss them with my students for the climate module I teach on. I am not a climate modeller and neither will I ever be one, but the modellers gave several periods in geological time extra significance, and I am rolling with it. The students need to know how these periods compare to the present they, and they need to know how good the data is we have about them. If they know that, then they have a better chance of being able to properly evaluate the robustness of the climate models. I think that makes them better climate scientists. So that is what I have been doing. I still need to do some polishing of these lectures for next year, but I am satisfied with the concept. I hope the students are too!

No comments: