 Okay, so I'd like to motivate my talk today in development of the Coastal Community Landscape Evolution Model, also named Cascade. I did a bad job Googling the acronym before I named my model. I'd like to motivate the talk and development with some recent news from the North Carolina Outer Banks. So this is an image of Pea Island. It is a low-lying barrier islands in the North Carolina Outer Banks. At the south end of this photo or at the bottom is Rodanthe. I'm gonna use this mouse. So this is Rodanthe. There is no mouse. Okay. Just north of Rodanthe, there's a stretch of roadway that's really vulnerable to storm impacts. It's frequently overwashed and NCDOT has decided to bypass this region with a large bridge and connect farther up the barrier island where it's slightly thicker. But why this made headlines this week is for this reason. There was a home that fell into the ocean and this made global headlines. This is the second home that fell into the ocean this year. And it's a pretty gnarly video. We were looking at it yesterday and thinking, oh, we hope the door doesn't open and somebody's in there. And it was evacuated beforehand but it struck a lot of conversation. So these were two news articles that I found the first one. North Carolina house collapses. Climate change is a real thing. The Outer Banks Beach house collapse won't impact your summer vacation though. And on Twitter, there's been a lot of discussions. This one on the right. Why is it so hard to move or demolish oceanfront houses before the Atlantic finally comes for them? And Jessica Whitehead used to work for the North Carolina office of resilience and recovery. And she said, hint, it's a policy and economics problem. It's not a hazard risk deficit or information deficit problem. And what I would argue is that it's a policy economics and a coupled landscape change problem. So we know that coastal risks are increasing due to sea level rise, storms, shoreline erosion, but so are the housing prices. So this plot in the upper left shows the last 10 years of housing prices in Wrightsville Beach, North Carolina. We had two really large hurricanes, devastating hurricanes during this period, Hurricane Florence and Hurricane Dorian. They made little bumps in these trends but the prices are still going up. And just this last year, the most expensive home in the state listed for $13.9 million in Wrightsville Beach along the Intra Coastal Waterway. And that's what I'm showing on the right. And this is in an area that previously flooded. So why are housing prices increasing? It's because markets are dynamically linked to the coastal environment. So looking at this feedback loop, starting with the left, large tax revenues and high value of infrastructure necessitates that we manage the coast and necessitates that we remove, overwash from roadways to access communities and that we nourish speeches, that we build larger dunes and these environmental characteristics are what help maintain property values. With more properties at risk, we construct larger dunes, more dunes and continue to nourish and the cycle goes on perpetually forever. What we're interested in, my colleagues and I, is what will happen to this feedback loop when landforms change with climate impacts, sea level rise and storms? So to look at that question, we needed a new model, a new coupled model that could simulate coastal management actions to protect roadways, to protect communities and how those management actions and the changes to landforms impact coastal real estate markets. So that new coupled model is like I said, Cascade, the coastal community landscape evolution model. And it is really a series of coupled models. At the heart of Cascade is Barrier 3D. It's a quasi 3D exploratory model of barrier evolution. It was developed by Ian Reeves at UNC Chapel Hill and the domain on the right is the Barrier 3D domain. This model simulates the effects of sea level rise, dune growth, storm overwash, shoreline change and a dynamic short-based response on barrier evolution over decades. Barrier 3D is coupled with Bree. Bree is the Barrier Inland Environment Model. It was developed by Yopneen House and Jorge Lorenzo Hueba. And we use Bree specifically for a longshore sediment transport to connect our individual Barrier 3D models. So that's our natural Barrier coupled model framework as this top loop. At the bottom are our human dynamics modules. We manage the roadways and the roadway manager. We manage beaches and dunes and the beach and dune manager. And then we have Seahome, the coastal home ownership model. And this is an agent-based model and I'll describe each of these in turn in the next few slides. So first, roadways. We manage roadways by first initializing a roadway in the Barrier 3D domain. So this is just a cross-section of the 3D Barrier 3D domain. We initialize the roadway. And as the dune migrates towards the roadway, we continue to remove overwash. We place it back on the dune just like our DOT does in North Carolina. And we do this until the dune threatens the roadway and then we relocate the roadway into the interior at the elevation of the Barrier. So this is how we manage roads. We stop management when the road can't be relocated and that's shown on the image on the right where there's just wetlands. There's no land for the roadway to be relocated to or when 20% of the roadway touches water. Beach and dune management is a little bit more complex. We nourish the shore face in the beach. We filter overwash for the effect of homes and commercial properties, meaning we don't even allow it to be placed on top of the barrier. And for what is actually placed on the barrier, like I'm showing in this image on the right, we remove some of the overwash and put it back on the dunes. We also don't allow the dunes to migrate. We want this front line of homes shown on the image on the left to stay there. We're trying to protect those homes and we rebuild dunes after storms. We stop management of a community when the barrier becomes too narrow to sustain a community. And we use information from NAGSED, North Carolina as a minimum community width of 50 meters. This is just a home plus a roadway. So a barrier can get really narrow. Okay. And then our final coupled model, and I should mention, this is not just model coupling, this has also been a lot of model translation from MATLAB to Python. So a couple of slides ago, I had the systems logo on the bottom right. And we worked really closely with Eric Hutton's in the back of the room to translate these models in and couple them together. So real estate and policy interventions are included within Seahome, this agent-based model. And we supply agents in this model with environmental variables. And these agents can make decisions about purchasing real estate. So this is what Seahome looks like. It was developed by my colleagues, Zach Williams, Dylan McNamara and Marty Smith. And this is Zach and I debugging Seahome all last week. So we have results to show you today. So within this model, an agent is a potential homeowner that is trying to decide whether or not they want to rent or buy at the beach. Each of these agents have their own risk preference, their own wealth characteristics, desires to live at the coast and expectations about the market in the future. Mathematically, an agent is just this equation on the right, it's a bid function. It includes functions that encapsulate the effects of beach width and erosion rate on perceptions, sea level rise, barrier elevation, the frequency of storms, how often nourishment is going to happen and what the expected beach width might be in the future. Okay, so towards the results, the initial conditions for Cascade are drawn from simulations that were parameterized for the Virginia Coastal Reserve. We're going to use high and low topographies that I'm showing in the bottom left. And these were developed for different dune growth rates. Within Cascade, storms add stochasticity. So here in the slides that follow, I'm just going to show one set of storms over 200 years. And I'm going to apply two different rates of sea level rise. The first one is four millimeters per year. And the second is an accelerated case. And that's shown by the black line on the lower right. So Cascade is an exploratory model. It's not meant to simulate specific barrier dynamics, but rather look at general behaviors. So in motivation of the PI lane case, I wanted to make something that looks sort of similar. On the left, this is our initial Cascade domain. We have a community that has commercial properties. It's high and wide. It's connected to a roadway that's also high and wide on the right side of the domain by a really narrow and low barrier segment. So everything on the left is a community or the first 150 decimeters. So notice that says dam in the bottom, that's decimeters. And then on the right is all roadways. So here are two plan form views over time. The middle plot is after 190 years of simulation. The bottom is 199 years of simulation for the case of linear sea level rise, a really low rate. So in this scenario, the roadway wasn't abandoned for 190 years, which is a long time. And thereafter, there was a storm of sufficient strength or size that allowed the dune that remained to be overwashed and the barrier quickly widened. So some takeaways from this were mostly earth scientists were interested in landscape change. While both of these management actions, maintaining a community, maintaining roadways, act to narrow and lower the barrier. And we can see that through these plan form views over time. Something that's a little harder to see here without time series, but for the sake of time. If part of the barrier moves landward, for example, the roadway, the effects are felt on the lawn shore and you begin to see these curved shapes of the barrier landscape because of that. And for a community that results in a more frequent need to nourish. But we're interested in real estate markets today. So at what point in this simulation will real estate markets unravel? The first case that we're going to look at is if the government heavily subsidizes nourishment. A 90% subsidy. So the remaining 10% would be financed by local communities. Within SC Home, we assign an imaginary value to oceanfront homes and to non-oceanfront homes. So for oceanfront homes, it's 500,000. For non-oceanfront homes, it's 400,000. And anytime the price drops below that market price, the pool of agents becomes wealthier. They buy up the cheaper real estate. And each of these oscillations is linked to a different nourishment, which I'm not showing here, nourishment and erosion. So another way to look at this, and I think this is more intuitive, is looking at the investor market share. So that's essentially the amount of renters that are in the market. So for the non-oceanfront, for this landscape change scenario, we start with a lot of renters upwards of 70%. And after 125 years for non-oceanfront properties, we have zero renters. Everybody that lives on the island is one of the wealthier agents that owns their home and is more tolerant of risk. So let's add some accelerated sea level rise. So here I've added the red line, which is now the non-oceanfront property for accelerated sea level rise and the green line, which is the oceanfront properties. And you can see that it goes to zero a lot quicker after about 80 years. So this process of the wealthier agents coming in, driving out renters and investors is just, it happens faster with accelerated sea level rise. And you'll notice the red line stops at 80. And that's because the barrier actually drowns in the middle. So you'll see the timing now for the accelerated sea level rise scenario of roadway abandonment is 84 years as opposed to 190 years. And thereafter, there's a low on storms and sea level is rising so quickly. There's not a storm to come in and overwash the barrier that that middle barrier actually drowns almost five years later or less than five years later. So the barrier community in the left becomes isolated and we don't account for this in our real estate model. So I ended the simulation there. And I wanted to look at, okay, what happens if we choose a different type of subsidy? What if we as a government only subsidized nourishment by 50%, we put a lot of the burden on communities to pay for it themselves. I've added here the brown line, which is the non-oceanfront properties with a 50% subsidy and the purple line, which is the oceanfront properties. And with this lower subsidy, there are more wealthy agents in the system from the start and it's a much quicker drawdown to no renters in the system. So these lower subsidies counterintuitively exacerbate social inequalities. And so it's something, this was an interesting takeaway for us that there are social implications when coupled with these landscape changes and policies to consider. So my takeaways, climate-driven changes to coastal landforms can unravel real estate markets. And there are complexities with spatial variability, with management, lower subsidies, likely exacerbate social inequalities. And I think this is an interesting demonstration of how we as earth scientists and earth surface process modelers can incorporate humans into our frameworks. So I was on the geomorph crew of this project. It's 10 years in the making. It took a long time to get this model up and running. The economics crew consisted of Zach, Marty, Dylan and Satya, economists, non-linear dynamicists. So it was a really interdisciplinary crew and we had a lot of help from systems. So with that, I will take any questions. Thanks, that's really exciting work. I'm curious, right now you've got the homeowners making purely economic decisions with some level of risk aversion. Are you giving any thought to more psychologically complex decision processes like theory of planned behavior or protection motivation theory? Ooh, I have not given really much thought to what the agents are thinking because I'm on the geomorph crew, but that's a really interesting question. And I'm sure that Marty or Zach would have more thoughtful things to say, but in terms of the economic model and what goes into it, it's empirically based. We sent out surveys or other members of our crew created surveys to try to understand risk perceptions. I imagine a psychological component to actually mathematically put that into our bid function. We would need data on that from people. Yes, we have some funding dealers out there to try to improve this model to include other social effects makers. So one of the paradoxes of the whole situation is that homeowners with insurance can't use that insurance to move their house or they can only get the insurance house destroyed like your video. Is that something that could be added or could the cost of insurance be added to the economics? I know you don't know the economics, but is that something? Yeah, I'm sure it could. I think what is happening here, I think that would just drive out investors more quickly. Yeah, if we added that to the bid function, what's happening is that the risk tolerance of wealthy agents is so much so that they don't care. They're willing to live at the coast no matter what essentially within our framework. So I don't know if insurance would matter for them as much as investors. That was really interesting. With the coupled models, have you thought about keeping track of uncertainty of different elements on the geomorphic side and on the economics and human side, tried to match that so that you're not being too precise in one area and way? Yeah. Yeah, I think we can do that with more model runs. This is just one storm sequence, but if we do hundreds or 100 was the plan, storm sequences and then average, I think we can account for some uncertainties there and play with some of the different management variables. That was my thought.