Interview with LML Fellow Davide Faranda

As a new feature of the LML blog, I will be running a series of occasional interviews with some of the LML Fellows and other individuals linked to the laboratory. This is a first interview with Davide Faranda, an expert in atmospheric dynamics and dynamical systems theory at the Laboratory of Climate and Environmental Science in Saclay, near Paris. Davide’s research ranges over a number of areas in atmospheric science and climatology, and focuses on applications of dynamical systems theory and extreme value theory. A list of his publications is available here.
Mark Buchanan
 
Mark Buchanan: From perusing your research papers, I see you do a lot of work with atmospheric flows and dynamics, working to understand weather extremes, atmospheric jets, blocking events and related phenomena. How did you get interested in this field?
Davide Faranda: Well, this was actually very, very long ago. When I was a child I was always watching thunderstorms, snowstorms and other interesting weather and I thought I wanted to be a meteorologist. And the best option for doing that was to study physics. So I did that in Bologna, and, as it turns out, in Bologna there was a good school of dynamical systems, and so I ended up doing my master’s thesis in dynamical systems.
As it happens, there were also many scientists there with expertise in climatology, and we became interested in finding out if the methods of dynamical systems theory could really work to get inside atmospheric flows. And, fortunately, it worked out. We’ve been able to produce some interesting results this way.
Mark Buchanan: Indeed, the methods you use seems to be a unique aspect of your research. I was under the impression – from following dynamical systems theory fairly closely many year ago – that the standard methods don’t work well for high dimensional systems. I’m thinking of the embedding theorem which can be used to reconstruct attractors. The atmosphere is certainly a high dimensional system, and yet you’ve found that these methods are very useful. What’ changed in the past decade or so to make this happen?
Davide Faranda: I think there are two things. One is the availability of data. The embedding tools were available in the 1980s. But if you think of the climate time series we had back in the 1980s,  they were very limited. It was essentially impossible to have daily data on pressures and temperatures. Now this is all changed. For the North Atlantic, for example, we now have data on daily pressures going back some 150 years. This is a huge data set. The second point is that we’ve changed the technique. We’ve moved from using the embedding results to instead using extreme value theory, and to study the recurrences of particular states. This allows us to build an asymptotic framework where the theory predicts the precise statistical distribution you should expect to find in the data.
This way you can check your analysis to make sure you’ve achieved some kind of convergence to reliable numbers. Then we can then read out useful information such as Lyapunov exponents or the dimension of the attractor. This is very nice.
Mark Buchanan: One particular topic in your work I find interesting is so-called blocking events — periods of stasis in the atmosphere which seem to be linked to weather extremes. What are these events? And is there any potential we might be able to predict them?
Davide Faranda: The idea of these blocking events is quite simple. We have different forcing of the atmosphere at the equator and the poles, as the Sun is stronger at the equator. So the temperature and pressure are higher at the equator. The result is that air tends to flow from the equator toward the poles, just like water flows toward the hole in a sink. But because the planet is rotating, the air can’t just go straight to the pole. The Coriolis force makes the flow curve and deviate in the mid-latitudes where it forms a jet — the famous jet stream. This jet would just flow smoothly around the globe at mid-latitudes, except that it’s unstable. Like many instabilities in physics, this drives oscillations in the jet. It all happens naturally, without the need for mountains or any geographic features.
Now, sometimes these oscillations get so large that they tear off from the jet stream, and you get a loop of flow enclosing a zone of high pressure, which moves toward the North pole. Or you get a zone of lower pressure moving toward the south. Once these big structures are moving around, they can occasionally become blocked and stay in one region to quite a while. One example is the anti-cyclone we had in the UK and France this past winter when we had fog and stable weather for many weeks.
The question is: what causes the blocking? We understand the basic theory, and the properties of the average blocking event, but why they occasionally get so extreme, we don’t know that. This might actually be explained by geographic details, such as the fact that we have continents and oceans, and that the difference in temperature over oceans and continents is increasing with climate change. These differences may contribute to building up these long blocking events.
You also asked about predictability: right now with weather models we can predict blocking events, but we can’t predict how long they last. We can’t make predictions beyond 7 or 8 days.  because of the fundamental chaotic dynamics of the atmosphere, and in this case the instability of the jet.
Mark Buchanan: You’re also involved in some work on the problem of so-called climate attribution – efforts to link specific extreme events as being caused by climate change. Can you tell me about this?
Davide Faranda: I think this was one of the most important lines of work of the group I’ve been coordinating at the Laboratory of Climate and Environmental Science at Saclay. We focus on extremes of climate change, and the basic question is: to what extent is climate change going to alter the probabilities of certain kinds of extreme events? To do this we need tools to analyse single events and our approach is to look at the dynamical pattern of the entire atmosphere associated with the extreme event. We can try to determine the likelihood for this pattern to recur in the current atmosphere, with climate change, and compare this to what it was in the “normal” atmosphere in which here was no climate change.
This is a little tricky as we have to construct in some arbitrary way this atmosphere without climate change. It’s an issue which arises a lot in social science — the needs to do counterfactual analysis, as we only live in one world. Essentially what we do is, if we have a long data set we assume the beginning part of the record is less influenced by climate change and the near part is influenced more strongly. But there are many subtle issues here. Even defining an events isn’t so easy. If we look at temperatures, we can consider the peak temperature in Paris, or the peak temperature as an average over France, for example. There are many possibilities and depending on how you define the event, you get different results. This is difficult to explain to the general public who want just a simple answer: yes or no.
Mark Buchanan: One final question. Is there are particular field you would really like to move into, which you haven’t been able to study yet? A topic where you think “I’d love to work on that, but I just don’t have the time right now.”
Davide Faranda: Yes, very much. I’m interested in applying all these techniques in neuroscience. Neuroscientists now have amazing measures of brain activity, and it’s possible to define states of the brain much as you can for the atmosphere. We did have a project a few years ago with some colleagues in Marseilles to apply these ideas to neuroscience, but, as I’m in a climate lab, I’ve never had time to fully commit to this analysis. But I’m now working with some other colleagues in the UK and we’re trying to push this analysis forward. Our dynamical indicators gives some interesting insights into how the brain is organized, and which states are more or less organized.
For example, one preliminary result is to understand so-called “mind wandering” states. This is what happens in the brain when you think in an unfocused way, not connected to the immediate environment. These seem to be the equivalent of atmospheric blocking in the brain. When you’re daydreaming or doing nothing, the brain is a high dimensional manifold where it is ready to move in many directions and jump into many other possible states. Then when you perform an action, all the dynamics collapse into a low dimensional manifold. So, we’ve started to do something in this area, but I would like to do a lot more of this in  the future.

Leave a Reply

Your email address will not be published. Required fields are marked *