 Give a warm applause to Dr. Jonathan Dongus. He is the co-lead of the flagship project, Copan at the Potsdam Institute for Climate Impact Research and all information about it. He will provide and he will introduce about more information for you. Thanks. I'm in a lucky position that my colleagues in two talks already gave quite a nice introduction to climate change and climate impacts and what can be done about it, maybe vegan becoming vegan. And so I can build on that and talk a bit about the what we see as a bit of the frontier of assistive modeling and some of the key challenges. And I try to create connections to things that are maybe of interest for people here at different places. So, well, again, to emphasize this again, this challenge of global warming here from a paleoclimate perspective. So this starts 20,000 years ago in the late, in the maximum of the last glaciation, where we had about four degree cooler climate than today, between three and four. And the mammoths were around there, right? And then there was a slow warming of about three degrees which took 10,000 years. It's actually quite slow. Then we have, there were 10,000 years nearly of a nearly very steady climate. That's the Holocene, the blue curve, right? That's where human civilization prospered, developed, where humans developed agriculture, technologies. The Industrial Revolution took place and then suddenly there's this kink, very comparably very fast warming there. That's the Industrial Revolution, driven by human anthropogenic emissions of greenhouse gases and land use change, mainly. This is all from paleoclimate data, from ice cores and other sources. Now these head crew data, these are measurements and they confirm the earlier sources here. There's still this quick warming and it's really important to mention that it's not only the amplitude of warming that's important here. We have about one degree of warming in this red curve already compared to these three degrees of warming since the maximum of the last ice age. But what is also very important is the rate of warming. You know, it's a hundred times faster than what happened from the last ice age to the Holocene. This anthropogenic warming, 100 times faster. And that's actually what creates a lot of problems for adaptation. Then if we go further into the future, this gets even more extreme, right? This is one of the quite extreme greenhouse gas emission scenarios which is actually a business as usual scenario. So that's what is projected to happen when we just continue with business as usual emitting. And this goes up to three degrees or four degrees. So we've seen that before as well. So again, 100 times faster, very extreme. So what is this doing? This is actually bringing us to a new geological age which has been termed the Anthropocene where social dynamics, social technological dynamics actually has become a dominant process in the earth system. The earth system is very complex. Lots of different processes are interacting on global scales, on local scales, regional scales, on different temporal scales. But humans have created these networks which we call also globalization, right? Networks of information flows, networks of trade, networks of resource flux. And it's actually these technological systems, these technological networks that have created, that have enabled this global warming, these huge greenhouse gas emissions. So it's really a social technological dynamics that has been driving this. So that's kind of relating to the title, right? That's the Anthropocene. There are some challenges. Then again, this curve here, the paleoclimate view on global warming, the last 20,000 years, but this also brings in some more information. One is the tipping elements, in so-called tipping elements in the climate system. These are parts of the climate system that have a non-linear response to global warming. So they can tip, they can tip like something that is standing on a set. They can tip over if a certain threshold is exceeded and the threshold is indicated by these uncertainty ranges there from yellow to red. So if, and a lot of them, actually a lot of these thresholds, they cluster in a range between 1.5 to two degrees. That's this cluster of tipping elements possibly switched within the Paris range. And these include important ice sheets on the earth which could create a huge sea level rise, but also a pink glaciers and coral reefs. And that's actually where this Paris agreement, temperature range comes from 1.5 to two degrees, has been agreed as a limit to global warming in the Paris Climate Agreement two years ago. But if you look at this curve, it shows that really most of the scenarios, these RCP scenarios that have been studies, they will shoot beyond that and they will hit even these upper clusters of tipping elements which are, for example, Amazon, rainforest, boreal forests, thermohyline circulation of the oceans. And only if you have a very nice sustainability transformation, this RCP 2.6 scenario, you might avoid these tipping elements to tip. So that's the picture of current climate science. And these might even interact and create tipping cascades. So this is from a paper that is just in press. But now there are not only climate changes an issue, but as Benny has pointed out in others as well, there are also other sustainability dimensions that humans have had an impact on. And they are summarized in the planetary boundaries concept and it's just important to mention that not only climate change is a problem, that all of these are a problem and these different dimensions like biosphere, integrity, destruction, climate change, but also chemical pollution or deforestation, they are all interacting and they are taking the earth system away from something that has been called the safe operating space for humanity. By people like you and Rockstream with Stefan. And the challenge for sustainability transformation is to basically return to the safe operating space, which is the only state of the earth system that is known to be a good state for human societies to prosper. And you can then bring the social dimension in and ask, well, for what earth system states are these planetary boundaries not violated, but where also certain social foundations are met. So they are basic needs of humans like food, water, income, but also jobs, voice, gender equality. So, and the safe and just operating space proposed by Kate Rawers is a hypothetical space of earth system states that allow for these two things, the environmental ceiling and the social foundation to be met at the same time. And that's of course what some people call earth resilience. So the question is in which what are future trajectories of the earth system where the earth system and human societies in it are resilient. And we are currently on a business as future trajectory, which takes us out of such a resilient state clearly and might take us even to hot house states similar to those where the dinosaurs lived. That's why the T-Rex was there on earlier slide. But then the question is are there actually non-linear social transformations that can maintain the earth system in a manageable state in a state of higher resilience? So that would be the curve that goes down, right? Or are there intrinsic feedbacks and tipping test kits that take the earth system up to such a hot ice free state? And now the problem is really that current models of global change cannot really answer such questions in a satisfactory degree. And why this is so, I will try to explain in the next slide. So this is the challenge of integrated earth system modeling that through macroscopes of computer simulations that Benny has already talked about. And the challenge that Chen Huber has already outlined in 1999 and even before that. And the idea is to look at the earth system as a co-evolution space of where there are environmental dimensions of the biophysical climate system but also social dimensions of human societies. And you can look at trajectories of the earth system in this space. There might be catastrophic domains where you don't wanna go where the tipping elements are tipping. There are inaccessible domains where you cannot go because they violate for example, basic physical laws, energy conservation or something like that. And then, but then there might also be parts of that space that are something like a safe operating space in the sense of the planetary boundaries or even a safe and just operating space in the sense of Kate Rower's in the sense of the so-called Oxfam donut. And the challenge for such an integrated whole earth system analysis is really to ask for example, are there even, is there even something like a safe operating space in the sense what is the size of it? What is its shape? What is the resilience of societal states within this space? Under which conditions does it exist? These are really systems questions where you need a complex systems analysis to answer them. So you have to run huge ensembles of complex models. You have to do uncertainty analysis, probabilistic things and so on and so forth. And even becomes a bit more complex so they are different dimensions of such an analysis which are very important and just to mention them very briefly, such new, such a kind of integrated earth system analysis we think needs to take into account really the dimension of human agencies or models have to represent agents and they cannot, and their macro dynamics that emerges from agent behavior. For example, people making decisions, sharing information on social networks, what macroscopic dynamics emerges from that that might be relevant, for example, for climate policy. Then actually taking these networks into account that make up globalization, modeling them explicitly, that is where complex network theory provides a lot of tools for doing that. That's the second point, representing these networks in our macroscopes. And the third point is really to capture the core evolution of nature and societies in these models and to really go beyond what just doing optimization calculations like economists often do with actually good reasons but if you wanna look at resilience, if you want to look at these questions like does a safe operating space exist at all? You cannot do it with optimization models. So a novel type of models is needed for doing this type of analysis which we call world earth models. Such models currently don't exist. And so we are doing some modest efforts in the coupon project to create such models. And for example, again, these models should allow to represent social network dynamics, opinion formation, social tipping points, tipping interactions, to address questions like how do actually these climate tipping elements like the green land ice sheet and Arctic ice sheet interact with potential social tipping elements? What are potential social tipping elements? For example, opinion formation is known to, has the potential to be a tipping element. Climate policy can be a tipping element. The divestment movement, the social movement on divestment from fossil fuels can be a tipping element. There can be tipping cascades in this network. And this is something we have to, so we have to tuvat. And this is just showing the types of global change models that are out there. The two dominant types of models that are out there right now are so-called earth system models and integrated assessment models. Both are focusing respectively on the biophysical dimensions of the earth system. These are the earth system models. They focus on the climate system. And the integrated assessment models, they focus mainly on the socio-metabolic or economic dimensions of the earth system. And then these new types of models that we envision should focus on a process, detailed representation of things going on in the socio-cultural domain. So really, again, what is socio-cultural? This is, again, opinion formation, creation of institutions, dynamics of organizations, and such things. And why do we even think that we can do that? Because it seems to be a huge, very difficult task. And we've seen that, even though we've come a long way since the Limits to Growth Report, 1972, I mean, there have been a lot of developments since then, 40 years past, or even more, 45. And there has been a huge progress in computing power since then, of course. So comprehensive earth system modeling is advancing fast already, and we can exploit this progress further. There are advances in complex systems theory, social ecological systems modeling, social simulation, and similar fields that allow really computer simulations of certain aspects of social macro-dynamics. So for example, again, this opinion formation example, we've done a study on how smokers have been marginalized in social networks in the US through the past 40 years or so, and we can do, at least on a qualitative level, quite nice projections of social network dynamics, and we can understand why certain things are happening the way they are doing. And this is actually possible through mostly agent-based modeling approaches, where you simulate the dynamics of thousands of simple agents and study the macro-properties that emerge from that. For example, the fraction of people that have a certain opinion and how it changes over time. Then the third point is that big data on social dynamics is increasingly available. And of course, this is valuable for science, but you have to be, of course, very careful about issues that can happen when you have too much data like that. And in most cases, you don't have it. That's one of our main problems. I mean, the data are there, but we cannot get it as scientists. And the fourth point is there are emerging research networks that aim at fostering such inter- and transdisciplinary research. So this is really about integrating natural and social sciences to a degree that is much closer than has been done in the past. And this is, of course, very difficult because everyone speaks different languages. But there are these networks, progress is slow, but there's some hope. And now this is a bit more concrete on what we are doing. In the COMPAN project, we are developing a software framework which is called COMPAN Core because it's our core activity at the moment. So it doesn't really have a meaning as an acronym so far. Maybe it has at some point. And the idea is really to provide a framework to build such models that has the potential to represent these different spheres of the earth system that I talked about before. So there's the environmental sphere than nature, which we call the biophysical taxon, which where, and then there's this metabolic sphere and the cultural sphere. And there are different entities in the model. So different things that are. And one thing that is very simple to an easy to understand are cells. So these are basically grid cells where typically a lot of biophysical dynamics are going on, but also, for example, agents, individuals, which is a different type of entity lives, can live in a certain cell. And then there's another type of entity that we highlight here. These are social systems. These are things like aggregate systems of parts of society like a nation state, for example, or a city that are described more on a macroscopic level by some aggregate variables, for example, like production, harvesting, wealth, capital stocks, and so on. And these different entities are of course interacting. They are somehow related to each other. And they are different modeling approaches like agent-based modeling, adaptive networks, differential equations, stochastic equations that we can use to model the dynamics of these entities. So they are. This is another type of another look at that, which is more like an UML diagram. So this shows how the different entities in the model are represented by classes, by objects, and how they can be related to each other. For example, an individual can live on a cell or a social system owns a number of individuals. So individuals can belong to social systems and so on. Then, OK, software design. We have a reference implementation in Python. It's object-oriented. We put a lot of effort in providing a good documentation test framework and to allow for parallel simulations on our cluster, because that's super important. And this is, of course, all work in progress in a way, because we have relatively low resources for doing that. But it's basically made in a way that people can really plug and play their models together. So it's a role-based modularization that allows for different levels of involvement of people. So they can be model users that are just using models in a scripting kind of way. They can be component developers that develop different parts to represent different types of system processes, for example. And they can be framework developers like us who are really providing the sourcing and trying to make it more performant, for example, or to port it to a different programming language or something like that. And we hosted on GitHub. The link is down there, but unfortunately, we are not allowed yet to put it open source because of our administration, but it will be, hopefully, open source early next year. And we designed this to be very, very flexible and modular so we can connect it to other models, for example, like these land use and vegetation models that Benny just talked about. And we try to follow these standards of open source open science here and to enable people from these different networks, but also maybe people from this community here to contribute to this. And yeah, this is an example how a typical script looks like that model run. So this is just in 30 lines of code here because this thing is highly modular. And this is just showing how simple this can be, actually, for a model user. And this is now a bit even more concrete. This is an illustrative example model that we put together to show that this whole thing works. And so the white boxes here, they are model components that a model component developer would provide. For example, a global carbon cycle. So this determines how carbon is, for example, diffusing between the atmosphere and the ocean, how carbon is going from the vegetation into the atmosphere. For example, entities involved in this carbon cycle component would be the world. So the whole planet, which is one entity, and then also cells because photosynthesis and, in general, biology happens on cells. And entities are, for example, also represented. Individual social systems are there. And then also, each of these entities is involved in different processes. So just to give one example, if you look into the social learning component, this is showing how people imitate each other. So it's a very typical human behavior to look what your friends are doing, and then you want to do it as well, many times, if you think it makes sense. And then the individual has this learn environmentally friendly process that it participates in. So this is, if you see your friend is environmentally friendly, maybe you want to be environmentally friendly too. And then we put this all together, and then this gets a bit complicated. Again, this looks more and more like the limits to growth model. And yeah, you can look into the paper for details here. But we have, of course, some results here. This is just really for illustration. So these are not predictions of any kind. They are not even projections, but they are just illustrations of the space of possibilities of Earth system dynamics, if you assume all the things that we have assumed. And this is a very legitimate thing that models have to do as well, or that we think models are important for as well, to explore the space of possibilities, to really understand how the world Earth system works. And you don't have to do predictions to have a good model. And this is, for example, we show here trajectories of the Earth system in the next 100 years without social processes, where, for example, the environmental friendliness of individuals is fixed, where certain subsidies, emission taxes, and fossil fuel bands are fixed. So here we have, then these things are not changing. We put them into the sociocultural dimension. Then there are, on the sociometabolic dimension, on the economic dimension, these are the shares of different energy sources that the societies use. So in this example, we have five social systems that are running at the same time, each has 100 individuals. And in this example, for example, you have a strong increase of biomass in the beginning, even later increase of renewable energy sources and a decrease of fossil resources, which is not very fast. And then the Earth system here on the biophysical dimension responds accordingly. And then if you switch the social processes on, so these are now really, people can do social learning, so they can learn to be environmentally friendly from their peers. There can be renewable subsidies can be implemented under certain conditions. Emissions taxes can be implemented and the fossil fuel band can be implemented basically by certain voting processes among the individuals. So we emulate the democratic decision-making process here. And then we see that all these things, they are switched on quite fast. And this leads to a rapid decrease here of the fossil energy sources and to a faster increase of the renewables compared to the case without social dynamics. So we get a completely different dynamics, which is of course what you would expect because we changed the model to a large degree. But this is what this type of modeling is about to really understand what is happening when you make plausible assumptions. And now, okay, what's the status of this? Copancore, it is already under operational use. It is ready for community integration and we have a description paper submitted actually today. And the open source lounge of the software is planned for early 2018. Now, okay, two, I want to give two outlook slides for types of research questions that I would find really interesting, important to look at. So in the context of this conference here, so one is really to look more into the competition trade of those synergies between these really, the two maybe main transformations going on in the 21st century. And well, one is that the digital transformation is ongoing rapidly and is just happening. And the other one is this necessary, maybe we think necessary, social ecological sustainability transformation. But there can be a competition even between them. For example, we've seen that before, there's a large and strongly increasing energy consumption of digital technology. So the internet in general is consuming huge amounts of energy. Then the blockchain and Bitcoin mining now is becoming, is mentioned a lot as an example because Bitcoin mining consumes the same amount of energy as a small country already. And it's just increasing exponentially. Then a bit more on a more subtle level, the influence of digital communication and online social networks and public opinion formation and governance processes that maybe relevant, that are relevant for sustainability transformation. So this is really the issue with debate about fake news and echo chambers and all these things, whether they are really there or not, is of course up to debate. But this is really relevant to study this more also maybe using these types of models. And then the last point is things that are a bit more far off maybe, but that people are seriously concerned about like emerging general artificial intelligence and then asking really what would be the earth system consequences of such events. And this is of course a bit, yeah, then maybe a bit even more philosophical idea, but there's this interesting debate about the so-called techno sphere. So this is looking at basically the complex of global technological networks and connected to human societies and what this actually has to say about sustainability. And Peter Huff has written a lot about that and we've made a small contribution to that debate and maybe this type of modeling that I talked about might also shed some light on how human societies are actually interrelated with these technological macro structures that we've created and what this means for really the viability of sustainable development in the future and also this actually highlights some deep ethical questions again about what sustainability actually is and how this depends on basic humanistic principles. So that sustainability is really mostly about humans and what do you do with that concept if it's true that humans don't really have agency anymore as some people claim like Peter Huff to some degree, right? So these are all open questions and now the take home messages, computer simulation models that are earth system macroscopes and the big social ecological data analytics they are really essential tools for understanding sustainable development in the Anthropocene. We think a new class of world earth models is really needed to capture key aspects of this dynamics and really one central challenge of this century I think is really how to reconcile the sustainability and digital transformations and yes, please contact us, maybe contribute and to that. You used your time perfectly, which means that we have just two minutes for two questions. If there are any, please go to the microphone or via internet. I see one question here and one perhaps there. So we have two questions. Amazing work, really interesting talk, thanks a lot. So I'm working more in the earth system modeling community and there we also have, we have the tendency to make our models more and more complex and what we've observed is that as we make them more complex the less we understand them, we do less understand what's going on and I've seen that a lot of your work you try to do the same, right? You try to integrate all the processes that you're interested in and then we've observed that a lot of the uncertainties we know that we have are coming from the parameters. So each time you integrate a new module into your model, at some point you need to parameterize some values you don't know and for us it's been very hard to measure some of the quantities and my question would be I guess for you it's even harder, right? What are your parameters and how do you try to estimate them? Thanks a lot. Right, so I mean there are of course, this is always the challenge with modeling and for example, some parameters that are typical for these types of social dynamics modeling they are concerning rates. For example, what is the rate of imitation of behaviors or opinions between people compared to the rate of the network dynamics? So how fast the network changes due to processes like homophily for example which is the process that typically people are more likely to interact with other people that are similar to them than with people that are different and so these different timescales, they can be parameters and of course, you can address that by doing simulations with on parameter spaces, model robustness tests. One thing that we do a lot in the coupon project is actually to do these huge agent based simulations that I've shown you today but we also do conceptual modeling and careful theoretical work where we try to really understand what the models are doing in a more simplified way. So actually most of our work so far has been working on very simplified models of such things that we can still understand mathematically sometimes even derive analytical mathematical approximations and this works quite well but of course this is always a challenge. I saw that there is one question from the internet so I asked you please go afterwards and speak with him, we don't have time because of the next speaker. So internet, what is your question? Yeah, it's somehow connected. How does your group especially and your scientific community in general validate models and how do you convince people that predictions using this model actually have some validity? So I tried to address that during the talk already but very briefly I think the one thing is that in this social dynamics modeling particularly it's very difficult to do quantitative predictions and that's also why what we try to do more is to so we validate the models doing reproducing qualitatively what past developments. For example in this study on the smoking change in smoking behavior and how people were marginalized in social networks we were able to do that quite well. So these are types of and if you can then look at different properties of a social network at the same time and you see that okay my model can reproduce the change, the decrease in eigenvector centrality and the decrease in clustering or the increase of clustering of smokers and the decrease of centrality at the same time under a lot of different parameter settings then this is a good validation in a qualitative sense for that model and then again I would like to highlight that we really don't try to predict things but we want to explore the space of possibilities really. Okay then a big applause for our speaker, Dr. Jonathan Donges. Thanks.