 Okay, so to begin the day, I just want to say these kinds of systems meetings, you know, when you have breakout groups, you have clinics, you have posters, you have keynote talks, plenary talks, it's complicated and I don't do anything. So, I think, join with me in thanking the integration facility team for putting this on. They work for months and months trying to get this together, so please join me in thanking them. And Lynn's not here, but, you know, 99% of what she does is, I mean, 99% of this meeting is because of what she does. Okay, so we have three keynote talks this morning. We have clinics, we have three more in the afternoon. We have breakouts for the other working groups, and then Greg will wrap it up. Oh, there he is. He's looking on like he's master in control. Okay, our first speaker, Jonathan Gilligan, is going to give us a great talk on human and natural systems and their role of agent-based bonds. Okay, so I'd just like to thank you very much for the chance to be here to be talking, and I'm going to be talking about the role of agent-based models and connecting human and natural systems. And I'm really going to be trying to kind of pitch this in a way about how to kind of take some baby steps into this if you're not already doing it. And in that light, I'd like to talk about the value of applying simple models to complex problems. And there's a lot of reasons why you might want to do this. And I don't want to be disrespectful to the incredibly cool work I've been seeing here about people doing really complex models that I find just so inspiring and even a little bit intimidating. But I'm also going to try to talk about why simple models can be really good, especially as we start trying to couple the human and natural system. And one aspect, I spend a lot of time thinking about use of models as a way of communicating about what we know, and we can have a dichotomy or a continuum between black box and glass box models. We can have large complex models that give incredibly faithful representation of what's going on in the world, but that are really hard to understand. And even if the code is open, they can still be very inaccessible to really understand what is going on. And sometimes simple models, you can trade off some fidelity to what the world is doing, but gain from that the ability to have some simplicity to understand and to communicate to others, particularly stakeholders, policymakers, you can see what's in here, you can see the assumptions I've made, you can question and change them. And that can have some real value. In addition, simple models, when you're starting out on something you don't understand, can be really important ways to identify important dynamics in the system. When I'm going to go out and launch a very expensive household survey, for instance, it can help to know what I'm looking for, and models can be useful early on in helping to say, where does this research go next? Where should we concentrate our resources as we develop the models? Where should we concentrate on building in complexity? And so I'm going to talk a bit about this and illustrate with a couple of case studies. One's a project that's pretty mature, and the other is one that's just really underway, and you're going to see the rough edges on it. So simple models for identifying important questions. One, I'm not going to give a lot of details about this, but this is a model of crop choice by farmers in Sri Lanka facing water scarcity. And one of their options is choosing do they grow rice, which needs lots of water, or green vegetables, which need less water, potentially produce a lot of profit, but have a lot of uncertainty associated with them. And looking at how this varies according to the conditions of the reservoir that they are irrigating, and whether they have private wells. So what we see here is this model, I'm not going to go into the details, but we've got these different columns here are representing different psychological theories and economic theories of how people think about risks and make decision under risk and uncertainty. On the x-axis is the fraction of farmers who have a private well, and on the y-axis is the fraction growing green vegetables. And what really stands out here is one of these decision models called prospect theory has really just radically different results from these other two models here. And so seeing that which psychological model of thinking about risk and making decisions under uncertainty the farmers use may have a really big impact on crop choice. That helps us focus the psychological part of the research on saying we'd better find out about how the farmers think about risk. So that's one aspect of how we can use a simple model as part of the design and development of the research project. I'm going to talk for a while now about another aspect of using agent-based models. This is really looking at education and public outreach, developing a participatory simulation that both has automated agents and also human activity to teach and inform about impacts of flood control. And it's just sort of wonderful to be talking about this here where Gilbert White set up the Natural Hazards Center. Gilbert White all the way back in the 1940s was warning us that we've really got to pay attention to the human element in thinking about floods. That floods are acts of God, but flood losses are largely acts of man. Going back all the way to Ph.D. dissertation at Chicago he was warning that building structural flood defenses can actually increase flood losses because as people lose the signal of high-frequency low-magnitude floods they get an illusion of safety and start developing much more valuable property in the low-elevation flood plain. So this is something that's been known, but often there's still a policy reflex to put in flood defenses. So we developed a participatory simulation, interactive simulations that provide users with trial and error. They can try things out, get prompt feedback. They can learn from experience. And in participatory simulations, multiple players interact with the model at once and then they interact socially with each other. And so as researchers we can look at how the interactions between them play out. And we integrate participatory and agent-based approaches. The players play high-level policymakers who are making policy for a city and automated agents simulate a low-level response by the population. So we simulate a flood plain here. We have three cities here, here, and here. And the colors represent the flood risk. We thought we initially started thinking about this with realistic real rivers and Hecaraz, and then we realized people get lost in the detail when they're looking at that. And for a pedagogical and communication purpose, a much simpler stylized straight channel looks a lot better for just getting at the essence of the problem. We use particle hydrodynamics for the water flow, both overland here and in the channel. And then we have agent-based land markets that control the development of property along this river. One challenge was nonlinear time. If we run the model at a constant rate of time, either a flood goes by so fast that the participants can't see it or we're sitting there for a long time with the participants getting bored because nothing's happened. So we have this run at a fairly fast clip, but as soon as a high-magnitude flood happens, everything slows down so the participants get to see this happen and then pauses for discussion. The players are the planners for these neighboring cities. They receive tax revenue and decide on flood wall construction and all the codes open source on GitHub. It runs in NetLogo. So the master interface that the teacher or the person running the simulation would be seeing is this, which shows the floods. It has controls of what they can do. And then the individual players will see a segment of the river like this and they will have a chance here, this players, considering whether to build a levy in the region indicated in red. This runs in sort of sequences where the model runs for about 20 years with the slowdown in case there's a large magnitude flood. Then it pauses. There's time for discussion, policymaking, putting a plan in place to build new flood defenses. And that repeats five times. And then after 100 years, we generate manually a 200-year flood so the players can see what a really large magnitude event does on the cities that they have prepared over the previous century. So this would be what it looks like during a flood event. They can see a somewhat 3D illustration of the flood wave. They can see shading in of the map for the flood inundation and then later the damages that are produced. And the experiments were done with pre-service middle school social studies teachers, basically in their teacher education in groups of two or three and we gave them pre-questionnaire to know what they knew coming in and briefed them on this random simulation and then debriefed and gave them a post-questionnaire to see what they had learned. And indeed we found the subjects really did learn a lot of the impacts of putting flood defenses that blocking a flood in one place can increase the flood wave another place that building flood defenses can encourage property development in high-risk areas. But one unexpected result we saw with students became very emotionally engaged with this. Students would say, oh, this is terrifying. We had one student who just gas put her hands over her mouth and said, oh, this is terrifying during one of the simulated flood events, one of those things probably lower fidelity visuals than an 8-bit video game from the 1990s. And emotional engagement. There's a lot of theories of learning and cognition. Antonio DiMascio and Paul Slovic in particular has done a lot on the role of feeling and affect in understanding and thinking about risks. And so we feel this is potentially a really interesting approach for education and public engagement. And of course, all the way back to Joseph Weizenbaum's work with the AI Eliza back in the 1960s, it's well known that interactive simulations can really facilitate emotional engagement. So we're continuing to work on this as potentially a pedagogical tool and a tool for public outreach and information. So that's one thing we're doing now to change gears a lot and talk about some work in Bangladesh is a coastal embankment project put in in the 1960s and 70s in response to some really catastrophic floods in the mid-1950s enclosed lots of areas along the coast of Bangladesh in earthen embankments creating polders named after the Dutch polders. And there's 123 of them. I'm going to focus on some problems here in what's called the Beals. Low-lying areas in backwater of these tidal channels. And in these areas, building the polders, they look something like this. There we go. They look something like this. You've got this embankment. You've got some slew states that are supposed to allow the interior to drain. But by avoiding flooding, the interior is starved of sediment and there's accelerated subsidence. Others in our group in collaboration with Steve Goodbred and others have measured that in some islands, over the last 50 years, there's been about a meter and a half of accelerated subsidence due to the sediment starvation and then the compaction of the soil. And that leaves you with, in some of these places, the interior is below the mean level of the river. So it's really hard to drain. The monsoon rains fall and they just gather in there and you have water logging. The place turns from a fertile area for farming into something like a swamp where nobody can grow anything. And that's a really big problem up in these VL areas and it's actually, people warned about that back in the 20s and 30s. An engineer called Mahalanobis warned that embankments for flood control are likely to make the situation worse in the long run. And a farmer interviewed by our research team in 2011 said that if the river could flow properly, then everything will be all right. So in response to the water logging, local grassroots efforts sprung up in the 1990s to try to ameliorate this. Engineers had been working on it throughout the 1980s and failed. All the engineering projects had really failed to relieve the water logging and some locals in grassroots efforts in a couple of places called Bielbacatea and Bielpina instituted a old indigenous practice of knocking big breaches into the embankments during the rainy season. Let the rains and the tidal flow wash over the land and bring in new sediment. And indeed, here we're seeing, after a couple of years, after building a breach right here and right here, you can see this sort of slay forming here, new land, within a couple of years, they'd raised the level of the land by about two meters where it was flooded, scoured out a silted up river here to a depth of about 10 meters and added about 600 hectares of new land. A few years earlier over here, a place called Bielbacatea, a similar thing produced over 1,000 hectares of new arable land that had used to be waterlogged. So these were really successful efforts. Governments saw this and thought this is great. We're going to impose a program to go everywhere that there's waterlogging and we're going to cut breaches in the embankments, let the tides flow in, bring sediment and it will make people better off. So up here is a place called Bielcacatea and here's what it looks like in 2010, about six years after they cut a breach down here. And you can see there's some new land forming, but much of this is still really waterlogged, underwater. People here were really unhappy about this. Several years later, it actually started to improve. A new breach was put in here. You start to see it's drier. People reported, hey, this is actually starting to work, but boy, it took an awful long time compared to what we were promised. And the unpopularity of this when it was forced on people by the government was really dramatic. In 2012, a motorcade of a member of parliament coming to talk to people in this area about what was called tidal river management here was met with a protest that turned violent. Nine vehicles were burned, about 50 people were injured and in a kind of post-mortem report, the Asian Development Bank evaluating this project which they had funded said that part of the problem was the engineers got no buy-in from the local communities. There was a failure to compensate farmers for the losses that they suffered when their land was temporarily flooded after the breach was put into the embankment and there was a lack of understanding of the indigenous knowledge base. Local farmer interviewed by my research team in 2012 said the government engineer would not accept our suggestion because he thought it was given by the non-experts. So bringing to this big problem that couples engineering technology, the natural system, and the human system, we have some challenges. We have to think, how are we going to represent local decision-making? The information we've got is some history about the local grassroots activism versus the imposed government program. We want to explore the role of voting and negotiation for building popular support as a first step towards developing a really coupled model of the sediment transport with local decision-making. Inequality is a really important thing. There's a lot of inequality in wealth, in land holdings, and so forth, and even within the land, some interviews I did a few years ago revealed that the people within a holder, there's actually hierarchies because the land's not level, it's like a saucer, and the people with the higher elevation land are less affected by a lot of the environmental degradation, and even if they support title river management opening the holder, they're going to want to close it sooner because their land will gain enough elevation it won't be waterlogged anymore. People at lower elevations in the middle want the breach to remain open longer so there's conflict within the community, and then we have very complex land tenure. Insecure land tenure, tenant farmers, sharecroppers, land that is nominally owned by poor people but actually owned by a rich person with a dummy owner to evade government regulations, and all of this becomes a very complex thing to try to model, especially because we've got a lot of limited data. We have a bunch of qualitative social science field work describing what's happening. We don't have any real quantitative social science work from these particular areas. We have a lot of physical measurements from elsewhere in the system but not right up here where we're working. And so in designing a model, another reason for simplicity and avoiding too much complexity at the beginning is if you don't have enough data to constrain parameters, too many parameters is a real curse. So we try to keep it simple at the beginning with the thought we're going to eventually build this out and put more complexity in, but only where we need it. And on the first day of this conference, Marco Janssen talked about the challenge of trying to develop agent-based models when you're working with qualitative social scientists who give you narratives but not numbers. And I want to very gently push back against that and say that I think the narrative is in many ways the most important thing. The narrative, the qualitative social science work that tells you what the meaning is, what is going on in the community. That tells you what you want to model. Quantitative work becomes really important later when I know what I want to model and then I need to calibrate parameters. But I really want to emphasize the value of narrative and the value of qualitative social science work in helping us figure out what to model when we're modeling the human side of the system. So in modeling tidal river management, we have a simple model of the coupled human natural system. We start with a simple zero-dimensional model of sediment deposition, which we adapt to a two-dimensional landscape by simply applying a semi-empirical scaling law to sediment concentration. We then develop a stylized representation of the agricultural impact of how water logging or flooding affects crop yields and add a human dimension of community decision-making. So in this work, we're using stylized models. They're informed by empirical data, but they're not strictly calibrated. We have not attempted to qualitatively validate this work in this area. Chris's sediment model was validated in a region much farther down the river, but we don't know that this is totally accurate for where we are. So we're working with a very stylized simple model to try to get at the dynamics as a first step towards developing new things. The basic frameworks in place, and now we're working on adding more social, economic, and political dynamics to this. So what we've got is I'm showing three maps here. An elevation rice yield with the embankments closed. This is the normal water logging condition. And then this is what it looks like if the embankment is breached so that river tides can flow over the land. So we start with a saucer-like elevation profile that's typical for a polder, but working with a stylized rectangular one. Here's where we're going to put the breach in. So this is the initial conditions. This is water logging in the first year is really depressing the rice yield. And if the embankment is breached here, then basically nobody can grow rice. If the embankment is breached for one year and then closed, this is what the elevation would look like. You're bringing up the elevation preferentially towards the breach. Rice yields go up. There's still water logging, but there's less. And if you left that breach open, some people could grow some rice around the perimeter at high elevation, but the people in the middle are still out of luck. After two years of the breach, this is what the elevation looks like, much less water logging. And again, here's what it would look like if you kept that breach open during that year. So we're going to look at decision-making about whether people decide if people had the boat or were able to negotiate among each other, would they stick with this, the water logging with no remediation, or would they be willing to undergo this or this and this for a couple of years so that they could end up, once they closed the breach, with something like this or this. So it's a trade-off of short-term losses against potential long-term gains. And the gains and benefits are not distributed evenly. So we look at simplified decision-making, starting somewhere and not getting too complicated at the beginning. We're looking at what the key informant interviews told us were a lot of the important considerations that things people worry about in deciding do they like title river management or not. And then we implemented some unrealistic but not unreasonable decision-making processes running just a simple boat or running an economic trade where people, not everybody agrees on this, so people can say, hey, I'm going to be better off, you'll be worse off, so I will pay you some compensation. And that way, I take a little bit of my gains, give them to you so that we're both better off. And so here I'm showing the discounted net present value of n years of breaches, where they're looking on to the total value of the rice they'll get over the next five years, if they open the embankment for one year and then close it, two years and then close it, three years and then close it. And these are showing the farmer's individual plots. We start with a lot of blue. People are better off, but some people at higher elevation are worse off because of the losses during the TRM. Two years, a few people are still better off, but more people are worse off. And by the time you get to three years, most people are worse off with TRM. A few people would be better off letting it go to three years instead of two or one. And we find there's no majority for any option. People disagree, some people want longer breaches, some people want shorter, some want no breaches. But if the winners can compensate the losers and we can have an effective market, then a majority emerges in support of a one-year breach here with many fewer unhappy farmers. So that is sort of where we are. We're going to look forward to developing more complex decision processes, but I hope it's useful for you to see some of what we can do with some fairly simple models. So thank you and I'd be happy to take some questions. So this is not so much a question, it's a comment. Right. And so over the course of the last several days, we've heard a number of talks and seen posters on Delta formation, Delta Dynamics. And in the world today, water still flows according to the physics of fluid dynamics. Sediment is still entrained according to the speed of water and its turbulence and the grain size. Things get moved down slope just like they always did. But Delta's big Delta's, the ones that are really important, whether it's the Ganges or the Mekong or the Nile or the Mississippi, heard in the talk yesterday with Kim and today's John, Delta's are formed by people who are actually altering the flow of water. They're the ones that are affecting where the water goes, where the sediment goes, when it goes. And so, I mean, what John's talking about is Delta formation in today's world, which is real different than Delta formation in the world of, say, 10,000 years ago. And so this is, I think this is really important to keep in mind that, you know, the physics haven't changed, but how they get implemented have changed significantly. If I could reply to that, I think that's exactly right. And so one of the things I didn't talk about that's an implication of this is with Steve and others in the project, we've measured that in a lot of these places with these embankments, the embankments have contributed probably 10 times as much to relative sea level rise as you static sea level rise has in the last 50 years. So better land management may be a really important part of adaptation to climate change. Yes. I'm going to have to run. The example that you showed of when you allow the flooding to happen and the decision makers with the social studies teachers that seemed like it would be an awesome thing to do with undergraduates. Have you, do you have a lab on that? Can I have it? Have you tried it? Okay. What's the story? Yeah, we, so this one, we did that with small numbers of undergraduates and I would be, as I said, the code is open source. I'd be happy to support you and do you know, I haven't actually designed an undergraduate lab, but I'd love to see that happen and it's something that my colleagues in STEM education and I are working on developing and I'd be more than happy to follow up with you on that. Yeah. Or follow up with the CSDMS educational. Okay. Absolutely. Yeah. And make that available because that seems like it would be really hit home, especially where I live in New Orleans. That would be an awesome thing to do. Right. Yeah. That's definitely on our, on our plans. Okay.