# Outreach activity: How do we make climate predictions?

This text was written for GeoEd, the education column of EGU’s blog, and first appeared there on Nov 27th, 2015.

In my second year studying physical oceanography, I got a student job in an ocean modelling group. When I excitedly told my friends and family about said job, most of them did not have the slightest idea what I might be doing. Aside from the obvious and oh-so-funny “you are a model now?!”, another common reaction was “modelling – with clay?” and the picture in those people’s head was that of an ocean model resembling the landscape in a miniature train set, except under water. And while there are many groups seeking to understand the ocean by using simplified versions of the ocean or ocean regions, simplified geometries, selected forcings acting on it, etc – this is not the kind of model I was supposed to be working with.

Talking about climate models with the general public

Explaining to a laymen audience what a climate model is a daunting task. We have all seen the images of a region divided into smaller and smaller squares as a visualization of boxes which represent a grid on which a set of differential equations is solved, yielding a solution for each of the boxes (See Figure 1). But do we really expect everybody we show this to grasp the idea of how this might help to understand climate if they don’t have the background to understand what a differential equation is, let alone how it has been discretised and programmed and is now being solved? From my experience it is very difficult to keep people interested and captivated using this approach and, unless they already have a pretty solid background, it is unlikely they will actively engage in the topic and ask clarifying questions.

Figure 1: Modelled sea surface temperature of the ocean off Mauritania, North-West Africa. Depending on the model resolution, smaller and smaller features in the sea surface temperature are resolved by the model. Still, even the most complex model is still nowhere near as complex as reality.

A new approach: Let them experience the process of building a model!

I therefore suggest we use a different approach. Instead of concentrating on explaining the mechanics of an ocean model, let us focus on letting people experience the idea behind it by using a “mystery tube” to represent the climate (or whatever process we want to model) and have the audience build their own “models”.

The mystery tube is all over the internet. I have not been able to find the original source but let’s look at what it is:

Basically, we have a tube that is closed off at the top and at the bottom (See Figure 2). Four pieces of string come out of it. When you pull one out, another one gets pulled into the tube. So far, so good. But the pattern of which string gets pulled in when another one gets pulled out suggests that there is something more going on inside the tube than just two pieces of string going in on one side and coming out at the other. So, how do we figure out what is going on? Some of you may have already seen a possible solution to the problem. Others might find one as soon as they’ve gotten their hands on a mystery tube and pulled on the strings a couple of times. Others might need their own tube and pieces of string to play around with before they are reasonably confident that they have an idea of how the mystery tube works.

Figure 2: A very non-fancy mystery tube: A paper kitchen towel roll with two pieces of curly ribbon going through. But what goes on inside? Still a mystery!

If you were to use mystery tubes in outreach (or with your friends and family, or – always a hit – with your colleagues), it is in fact a good idea to have a couple of “blank” tubes and pieces of string ready and let everyone have a go at building their own mystery tube that reproduces the functionality of the original one. Ideally, as you will see below, you would have more than just the bare necessities ready and also offer flat washers, springs, paper clips or any other distracting material that might or might not be inside the mystery tube.

Why offer “distractor” materials? Because we are trying to understand how people come up with climate models, remember? The original mystery tube represents the process we want to model. We do not know for sure all the important components of that process, and therefore do not know what needs to be included in the model, either.

— SPOILER BELOW! If you want to solve the mystery tube mystery yourself, do not read on! —

Now, in the instructions on the internet the two pieces of string are connected inside the tube by way of a ring through which they are both fed. When I first build my own mystery tube, I was too lazy to search for a ring to connect the pieces of string, so I just crossed the two threads over. After all, the ring wouldn’t be visible in the final product, and the function would remain the same anyway!

From empty cardboard kitchen towel rolls to climate models

Which brings me to the main point of this blog post, first made by my friend and fellow outreach enthusiast, Dr. Kristin Richter (http://kristinrichter.info, currently University of Innsbruck, Austria), who is always my first stop when wanting to bounce ideas for demonstrations or experiments off: This is exactly why modelling climate is so difficult! We can build a perfectly working mystery tube but unless we cut open the original one we will never know whether our solution is the same as the one in the original mystery tube, i.e. whether there is a ring inside, or a paper clip, or the two pieces of string are just crossed.

You might argue we could find out what is inside the original mystery tube by other means, for example by shaking it and listening for rattling, by weighing it, or by many other methods. Yet, can we ever be sure we know exactly what is inside? And more importantly, would we even think of shaking or weighing the mystery tube if we weren’t specifically looking for what connects the two pieces of string? And are we really sure we are reproducing the full functionality of the original mystery tube? Maybe the original ring has a blade on the inside, so after a certain number of experiments one of the strings will be cut? Or maybe there is something else inside that will happen eventually, but that we cannot yet predict because our mystery tube, while reproducing what we observed from the original tube, just does not include that element.

The same goes for climate models, of course. We can reproduce what we observe reasonably well. Assuming we know of all “parts” of the climate and how they work together, we can make a prediction. But the climate is a lot more complex than a mystery tube. Of course, climate models are based on physical principles and laws and not just best fits to observations. Yet, in many places decisions have to be made for or against including details, or for representing them by one parameterisation and not another.

Can we ever know for sure what the future will bring?

So does that mean we should give up on making models of the climate because, while we might be able to reproduce the status quo, prediction is impossible? Absolutely not! But we need to be aware of the possibility of feedback mechanisms that might become important once a threshold has been crossed or tipping points (like when a hypothetical blade inside the ring will have cut through one of the pieces of string). If we are aware that there might be more to the mystery tube than just the pattern of how strings move which we observed at the beginning of this post, we can watch out for signs of other components. Like listen intently to the noise the string makes when gliding through the mystery tube, or listening for rattling when you shake the tube, or monitor the strings for wear indicating there might be a hidden sharp edge somewhere.

And the same obviously goes for climate. We need to monitor all observations and look closely at any deviation of the observations from our model. We need to come up with ideas of processes, which might become important under different conditions and look out for signs that they might already start to occur. We need to be aware that processes we haven’t seen evidence for yet might still be important at a different parameter range.

Once we have gone through all this with our audience, I bet they have a better idea of what a modeller does – even though they still might not have a clue what that means for the average day at work. But typically, people find the mystery tube intriguing, and you should definitely be prepared to answer a lot of questions about what your model does, how you know whether it is right, what processes are included and what are not, and voilà! We are talking about how to make climate predictions.

P.S.: This text originally appeared on my website as a page. Due to upcoming restructuring of this website, I am reposting it as a blog post. This is the original version last modified on November 27th, 2015.

I might write things differently if I was writing them now, but I still like to keep my blog as archive of my thoughts.

# I am missing institute seminars! Or: Why we should talk to people who use different methods

You probably know that I have recently changed my research focus quite dramatically, from physical oceanography to science communication research. What that means is that I am a total newbie (well, not total any more, but still on a very steep learning curve), and that I really appreciate listening to talks from a broad range of topics in my new field to get a feel for the lay of the land, so to speak. We do have institute seminars at my current work place, but they only take place like once a month, and I just realized how much I miss getting input on many different things on at least a weekly basis without having to explicitly seek them out. To be fair, it’s also summer vacation time and nobody seems to be around right now…

But anyway, I want to talk about why it is important that people not only of different disciplines talk, but also people from within the same discipline that use different approaches. I’ll use my first article (Simulated impact of double-diffusive mixing on physical and biogeochemical upper ocean properties by Glessmer, Oschlies, and Yool (2008)) to illustrate my point.

I don’t really know how it happened, but by my fourth year at university, I was absolutely determined to work on how this teeny tiny process, double-diffusive mixing (that I had seen in tank experiments in a class), would influence the results of an ocean model (as I was working as student research assistant in the modelling group). And luckily I found a supervisor who would not only let me do it, but excitedly supported me in doing it.

Double-diffusive mixing, for those of you who don’t recall, looks something like this when done in a tank experiment:

And yep, that’s me in the reflection right there :-)

Why should anyone care about something so tiny?

Obviously, there is a lot of value in doing research to satisfy curiosity. But for a lot of climate sciences, one important motivation for the research is that ultimately, we want to be able to predict climate, and that means that we need good climate models. Climate models are used as basis for policy decisions and therefore should represent the past as well as the present and future (under given forcing scenarios) as accurately as possible.

Why do we need to know about double-diffusive mixing if we want to model climate?

Many processes are not actually resolved in the model, but rather “parameterized”, i.e. represented by functions that estimate the influence of the process. And one process that is parameterized is double-diffusive mixing, because its scale (even though in the ocean the scale is typically larger than in the picture above) is too small to be represented.

Mixing, both in ocean models and in the real world, influences many things:

• By mixing temperature and salinity (not with each other, obviously, but warmer waters with colder, and at the same time more salty waters with less salty), we change density of the water, which is a function of both temperature and salinity. By changing density, we are possibly changing ocean currents.
• At the same, other tracers are influenced: Waters with more nutrients mix with waters with less, for example. Also changed currents might now supply nutrient-rich waters to other regions than they did before. This has an impact on biogeochemistry — stuff (yes, I am a physical oceanographer) grows in other regions than before, or gets remineralized in different places and at different rates, etc.
• A change in biogeochemistry combined with a changed circulation can lead to changed air-sea fluxes of, for example, oxygen, CO2, nitrous oxide, or other trace gases, and then you have your influence on the atmosphere right there.

What are the benefits of including tiny processes in climate models?

Obviously, studying the influence of individual processes leads to a better understanding of ocean physics, which is a great goal in itself. But that can also ultimately lead to better models, better predictions, better foundation for policies. But my main point here isn’t even what exactly we need to include or not, it is that we need a better flow of information, and a better culture of exchange.

Talk to each other!

And this is where this tale connects to me missing institute seminars: I feel like there are too few opportunities for exchange of ideas across research groups, for learning about stuff that doesn’t seem to have a direct relevance to my own research (so I wouldn’t know that I should be reading up on it) but that I should still be aware of in case it suddenly becomes relevant.

What we need is that, staying in the example of my double-diffusive mixing article, is that modellers keep exploring the impact of seemingly irrelevant changes to parameterizations or even the way things are coded. And if you aren’t doing it yourself, still keep it in the back of your head that really small changes might have a big influence, and listen to people working on all kinds of stuff that doesn’t seem to have a direct impact on your own research. In case of including the parameterization of double-diffusive mixing, oceanic CO2 uptake is enhanced by approximately 7% of the anthropogenic CO2 signal compared to a control run! And then there might be a climate sensitivity of processes, i.e. double-diffusive mixing happening in many ore places under a climate that has lead to a different oceanic stratification. If we aren’t even aware of this process, how can we possibly hope that our model will produce at least semi-sensible results? And what we also need are that the sea going and/or experimental oceanographers keep pushing their research to the attention of modellers. Or, if we want less pushing: more opportunities for and interest in exchanging with people from slightly different niches than our own!

One opportunity just like that is coming up soon, when I and others will be writing from Grenoble about Elin Darelius and her team’s research on Antarctic stuff in a 12-m-diameter rotating tank. Imagine that. A water tank of that size, rotating! To simulate the influence of Earth’s rotation on ocean current. And we’ll be putting topography in that! Stay tuned, it will get really exciting for all of us, and all of you! :-)

P.S.: My #COMPASSMessageBox for this blogpost below. I really like working with this tool! Read more about the #COMPASSMessageBox.

And here is the full citation: Glessmer, M. S., Oschlies, A., & Yool, A. (2008). Simulated impact of double‐diffusive mixing on physical and biogeochemical upper ocean properties. Journal of Geophysical Research: Oceans, 113(C8).

# My workshop at MeerKlima.de

Today I ran a workshop at the MeerKlima.de congress in Hamburg: A congress for high school students, organised by a student committee. The large lecture theatre of the chemistry department at the University of Hamburg was crowded for the opening lecture by Mojib Latif:

For my workshop, however, we set a limit of 40 participants due to the size of the room (and the amount of stuff that I had lugged in from Kiel. Yesterday’s ice cubes did very well, btw!). And there were two TV crews and a photographer documenting the awesome ice cube experiment.

You can watch documentaries of the workshop here and here (both in german).

Sneak peak of those two documentaries, obviously only of the tiny little sequences featuring me:

And thanks to Johanna and Dirk for their support before, during and after the workshop!

I also got to watch another workshop by a colleague, who used the Monash Simple Climate Model (which I have talked about here) and I have got to say: That is such an awesome tool for teaching about models and/or the climate system! You will definitely hear more about it in the future as I incorporate it into my own teaching.

And last not least we had a phone call to the Meteor off Peru which rounded off a day full of bumping into people I hadn’t seen in a while. Always great to reconnect with old friends and colleagues!

It was great fun to be part of this congress, and it was a great way to experience first hand how science outreach can work in such a format. Since the congress was curated by the students themselves, many students were very interested and asked great questions. Also, the topics of the workshops corresponded closely to what students really wanted to see and hear. It would be amazing to see this scaled up next year, maybe over several days and with more parallel sessions, so that participating students really get to pick and choose exactly what topic they are interested in and that even more students get the opportunity to experience such an amazing congress!

# Using the Monash Simple Climate Model as first exposure to “real” climate models

When talking to the “general public” (which sometimes just means friends or relatives) about working in climate sciences, it is sometimes really difficult to explain what it is we do every day. I have described a very simple way of explaining how climate models work before. But while this might help provide a general idea of what a model does, it does not show us what climate models actually do. But there is a great tool out there that does exactly that!

The Monash simple climate model is a real climate model. When I was still in Kiel, almost 10 years ago, my sailing buddy Janine was working on implementing the first version of that model! And now the DKRZ (the German Climate Computing Center) hosts an web-based interface that lets anyone access the model.

You can build up the climate model step by step, adding representations of processes like ice albedo, clouds, or many other and then compare model runs including those processes with those runs without. You are even shown the difference between those two runs to see how properties like surface temperatures are affected by the process under investigation! And really awesome feature? The visualization of which processes are switched on and off. See below: On the left, in experiment A, all processes are switched on (and therefore shown in the picture on the top left). In Experiment B, on the right, almost all processes have been switched off, only incoming solar radiation and outgoing radiation are active. Looking at the temperatures below, this shows how Experiment B is only influenced by the sun and temperatures are the same along lines of constant latitude. In Experiment A, though, the temperatures are modified by many more processes, and therefore the distribution is a lot more messy.

Screenshot from http://mscm.dkrz.de, shared under CC BY-NC-SA

You can also look at different climate change scenarios, and you always get to see the CO2 forcing of the respective scenario. You can also compare scenarios with each other (see below). Doing this, you can vary parameters, too, to investigate their impact. You can always look at different model fields like surface and subsurface ocean temperatures, atmospheric temperatures, atmospheric water vapor or snow/ice cover.

Screenshot from http://mscm.dkrz.de, shared under CC BY-NC-SA

There are very nice video tutorials for a quick start, and puzzles where you can test how well you understand the model.

I absolutely love this tool, and I wish I was teaching anything related to ocean and climate so I could use it in my teaching. This opens up so many possibilities for inquiry-based learning. Or basically just interest-driven exploration, which would be so fun to initiate and then support! You should definitely check it out! http://mscm.dkrz.de/