 Good morning, good afternoon, good evening everyone who are attending this webinar. First of all, I would really like to thank you all that you are attending the webinar and Maria Laura and our partners from the Waters China, Abraham and Long, for organizing this webinar. Today we talk about modeling human-fledged interactions, which is part of my PhD research work. I will give you an overview of the theoretical background behind this whole concept of modeling human-fledged interactions. And then I will present a short case study, just considering the time. And then I will reflect on the work. Then we will have a Q&A session. So, we start with FLIRS. FLIRS is the most frequent disaster. FLIRS is the most frequent disaster on Earth. They account from 1980 to 1999 in terms of total number of disasters. They were the second most frequent disasters. But from 2000 onwards, from the year 2000 onwards, the last two decades, FLIRS are the most common type of disasters. So, again, in terms of impacts, not only in frequency, but also in terms of impacts, FLIRS have the highest impact in terms of number of people affected per disaster in the last 20 years. And in the last 10 years, and then also the second highest economic loss in the world occurred because of FLIRS. And then these are reports from the credit and UNDRR. And then if we talk about FLIRS risk, then what's FLIRS risk? So, the impacts of FLIRS are attributed to the hazard, the physical component, and the vulnerability and exploited aspects, which are socioeconomic aspects. But in most cases, FLIRS impacts are mainly attributed to the extent and magnitude of the hazard. And if you see all kinds of, I mean, most, majority of FLIRS risk management models, they usually consider the hazard component alone. So, hydrodynamic modeling and quantifying the flooding. And because of that, actually, because of that, the climate issue is also very important that contributes to the hazard. So, the frequency and magnitude of the hazard is increasing so that the climate change is very important topic. However, I also strongly believe that we live in very exposed areas, on flat plains, humans settle in flat plains. The weather gap is increasing, poor people are living in very precarious conditions in exposed areas. So, and also the population increase, people living or settling in flood prone areas. So, the socioeconomic aspect is also quite important. So, the negative impacts of FLIRS are not just because of the hazard component, but also the socioeconomic component, the vulnerability and exposure. So, considering that, usually actually what we say is we need to take out nature from natural disasters because and there is a huge campaign nowadays actually, no natural disaster because actually, if you really think about it, the natural element is the rainfall, right? Or hurricanes, but it depends on how we manage our system, our urban system when it comes to the flooding itself. So, the flood is not a natural phenomenon. Flood is not a natural phenomenon. Flood is, well, runoff is a natural phenomenon, but if we manage that runoff, then floods will not happen. So, the natural aspects, we should be a bit careful on that because people think that, okay, it's natural, then it's beyond us. No, we can do something. One thing I mentioned in my thesis is I'm from Ethiopia and a rainfall event that occurred in the Netherlands, same magnitude may not cause any flooding, but it may cause flooding in Addis Ababa, the capital of Ethiopia because our storm drainage management system is not good enough. Our flood management is not good. So, these governance aspects, special economic aspects, the different adaptation mitigation aspects are very, very important. And if we talk about flood risk management, this definition of flood risk management from my PCC, which is very comprehensive also strengthen this idea. There are a lot of elements, several elements in flood risk management, designing, implementing, evaluating different strategies, policies, measures and understanding a reduction of flood risk, flood preparedness, response recovery, and then finally the goal is human security will be in quality of life and sustainable development. So, it encompasses all the flood, the physical aspect, but also the human aspect. And considering that, we need to study both the physical and in terms of modeling, we need to understand both the physical and the human element and the interaction between the two. And considering that the human flood interaction, which the flood risk management is the core, is seen as a complex adaptive system. So, just to give you a brief summary of a complex adaptive system, complex systems are systems with a large network of interconnected components with kind of no specific central control. And they have simple rules and using those simple rules, the components interact and then come up with a new collective behavior emerge. And this collective behavior has a pattern and can be with all kinds of information process can be studied. So, if you see here with the interaction of the societal different aspects of the society, right, elements of society, technological aspects and the natural condition, then all this interaction gives to some level of flood risk. And due to the different change in the society, technology and nature, the flood risk level also changed. So, this is the emergent phenomenon that we call in a system study. When the components of this complex system, when they adapt through learning or through evolution and adapt in a positive way, right, so that they reduce flood risk in this case, then this called a complex adaptive system. So, a human flood interaction, a flood risk management system is a complex adaptive system. And if we talk about complex adaptive systems, there are many different types of complex systems. We can talk about a couple human and natural systems. They call them chance or social ecological systems or social environmental systems. Social technical systems with the technical aspect, technical artifacts are, we give more weight in that case. And then we can narrow down and then we say human water systems or human flood systems even narrowing down. This is just to conceptualize the problem better, right, the system better, we need to narrow down. So in this case, we talk about human flood systems. And the system has components, subsystems, the human subsystem and flood subsystem. If we see this image, the natural part is the hydrometrological events, in this case, tropical cyclones or intense rainfalls that brings flooding. And if we talk about technology, then we have different kinds of adaptation measures, weight flood proofing, dry flood proofing, different technologies, hydraulic structures like dye systems and early warning systems starting from monitoring up to dissemination of information. And then if we talk about the societal aspect, then we have the general public businesses, municipalities, emergency and disaster management and social institutions. In this case, we focus a lot on social institutions also. So what are institutions? Institutions are humanly devised constraints that shape human interaction. So these are devised by humans by us so that we constrain and behave, our behavior so that we can survive, right? So these institutions are very fundamental in the social, economic and political makeup of human beings and they define our interaction, right? For example, just to give you an example, COVID rules now. So if the government says we need to put masks when we enter in a closed space, shops, malls, whatever, that is an institution. There is a constraint that defines what somebody has to do and when, in what condition. And if not, what is the sanction? So if you don't follow that, you will be fine. For example, in the Netherlands, 95 euro. So these are humanly devised constraints that shape our interaction, how we behave. If you don't stop on a traffic light, red light, then you will be fine. So these are institutions. We can mention more examples like norms, different kinds of norms, how we behave and climate change agreements, the current agreements like taking place, the COP26 for example, these agreements between countries. So all these are institutions, sustainable development course institutions. These are not, they don't only affect individuals, but also between countries, between, so all these are institutions. And they are expressed through institutional statements and these are using the ADICO grammatical syntax. It's called ADICO refers to the actor, the A, the D is the deontic, what is allowed or not. I is the action, the aim, the action. C is the condition, in what condition, why. And O is the or else, if we don't follow that, what is the sanction. And considering this ADICO grammatical syntax, all of them don't have sanctioning, some of them have sanctioning, some of them not. For example, if you consider climate change agreements, there are usually handshakes and they agree, but some countries don't follow, nothing serious happens if you don't follow that. While, well, you can be kind of alienated politically or diplomatically somehow, but if you, if you steal, and if you are caught, then you will go to jail, or you will pay some kind of consequence, right. So, some like criminal course, these are rules, there are norms, rules, norms and strategies, some of them are strategies, people do things collectively when many people do it, then others follow, these are like a word of mouth kind of institutions. In flood risk management, these institutions, different institutions, we can have policies, ordinances that shape the hazard, vulnerability and exposure components of flood risk. What are the main gaps we identified in modeling chemo and flood interactions, there are two different types of modeling aspects, stylized models and system oriented models. These stylized models, they use differential equations, so they lump the whole area into one equation, and then there is no heterogeneity that my decision, my perception, flood risk perception, compared to my neighbors and area is not the same. So, but if you lump it, then you lose that heterogeneity, the different, these micro level interactions or micro level behaviors and institutions are not involved in this. And if you see system oriented models, current, most of those we identified, they consider flood as exogenous element of that system, so they don't consider it as part of really actual part of the system, but like an outsider. And because of that they simplify, use simplified flood models, and also in terms of institutions, they don't have this, this well defined concept of institutions, social science theories, while they just use very simplified set of rules. And for example, here what you see here is a system diagram. Now, what was the research? It was to develop a modeling framework, as much as possible, a very holistic one. And a methodology based on this framework, a methodology to build holistic human flood interaction model that provide the models that provide new insights into flood risk management, policy analysis and decision making. And what approach we followed, the complex system perspective, complex adaptive system perspective, so we use system oriented models. So, the benefit of this is that when you use system oriented models, you can have, you can connect different subsystems, right? Flood subsystem, in this case, and human subsystems so that you can address both subsystems with the right type of modeling. And then we define concepts, the importance of having a framework is to really define concepts, what are the most important elements that should be addressed in human flood model. And then finally, an integrated modeling approach was used because, or has been applied. That is because now you will say system oriented models will have the human model, will have the flood model. Then we need to integrate these two and see how the system fits back between each other, the feedback between each other. And then how the systems, both subsystems, evolve over time. So we came up with this coupled flood agent institution modeling framework called CLAIN. It has five elements, the first one agents, agents are representations. These are model representations of individuals or composite actors or including households or government organizations. These agents, they have states, internal states like age or if the gender matters. For example, for the model, then gender. If socioeconomic conditions are important, then these kinds of, these are the states of, or for example, if it is a government entity, then the budget, the staff. So these are the states of the agent. And then agents have behavior, they have actions, they have interactions, they interact with each other, they interact with another type of agent. So all these, and the second ones are institutions. The institutions are what we mentioned before, they are expressed using adico statement. And these are either written or verbal or formal or informal or we can also define them as, as I said, rules, norms and strategies. And strategies, actually, if we follow the institutional statement with the adico syntax, then we define institutions as rules, norms and strategies, shared strategies. And agents, they create, change or abandon institutions. While the institutions also influence agents behavior, right? Agent, and then we have the third one is the physical, physical processes. These are hydrologic and hydrodynamic processes, rainfall, runoff, infiltration, whatever, coastal processes, pipe flows, like 1D2D kind of models, surface flow. All these that generate the flood, any type of fluvial, fluvial, any type of flood, flash, whatever. And these are the physical processes. And they would have the urban environment. The urban environment in this case is this kind of splitted the physical process from the urban environment because first thing we talk about urban environment because we talk about urban flood risk management. So specifically urban environment. And then in an urban environment, of course, all these physical processes happen on that environment, but any other kind of hydromethrological or geotechnical event can happen on an urban environment, landslide or earthquake, whatever. But we don't address those ones in this case, specifically floods. So that's why we kind of speak them. In fact, if you think about the agents also, they live on that urban environment. So the urban environment is a link between the human aspect that is agents, institutions, and then the physical processes in this case, those processes that generate the flood. And then finally, we have the external factors. Two types of factors we identify here. One is the source of flood. This is actually connected to what I said before the natural part of natural disasters. The rain, the rainfall, the hurricanes, these are so just first let me back up. So if you see this outer boundary, this is the system boundary. That means everything within that rectangle is part of the system. And that system can change or components of the system can change or do something within their power. In terms of policies or in terms of policy implementations measures and whatever. Whereas everything, the external factors are those outside or beyond the limit of that system. So the natural part in this case, the rainfall or this, I mean, there are some technologies that alter rain pattern at least. But normally rainfall and hurricanes, we consider them as natural phenomena. So because we don't have that much influence on them, that's why we put them as source of the source of flood as external factors, including climate change and scenarios. These are included there. And then you have the external economic and political factors. These are like institutions actually, but external institutions. So, for example, a financial crisis that happens anywhere would have an effect on a specific place. But that specific study area cannot have that much influence on that global crisis, right? But because of the global crisis, then budget cuts and this and that, all these things affect the work flood risk management related work. So national level policies affect city level or regional flood risk management activities while if the regional level system components, in this case agents, even if they don't want to follow this rule or if they want to change this, they cannot because it comes from, let's say, the parliament. So it goes beyond that system boundary. So that's why these are like external economic and political factors that have an effect on the system, but the system cannot really directly, at least, cannot affect or change the factors. And then if we talk about modeling, then if you see this, the physical processes, the source of flood, the external factors and part of the urban environment, for example, the rivers, the channels, the topography. These are part of the flood subsystem and they can be studied in hydrodynamic models, whereas the agents, institutions and the external economic and political factors and including where the agents live and what they do on the urban environment. That aspect is the human subsystem and is studied using agent based models. So agent based models are computational models that simulate agents interaction using set of rules. In our case, the set of rules are the institutions. Okay, so from that what we do, how can we develop a couple ABM flood model. Then first thing is we conceptualize the system using the framework. So we define what are the agents, the institutions, how the urban environment looks like, what are the physical process, the important or the relevant physical processes. If it is a coastal flood, we don't talk about pipes, for example, or urban drainage channels, maybe because we have only coastal flooding or vice versa. And then if we have any external factors, what are the external factors that affect the system. Then we build the agent based model and the flood model and then we couple the two models. And then see the feedback between the change in the agent based model, what is the feedback on the or the effect on the flood model and then whatever happens in the flood model, how does it change the human behavior and how does this continue over time. And considering flood risk management phases, this study addresses the recovery and prevention and mitigation phases, the long term flood risk management aspects, not immediately after an event. That is most operational level events and they require some level of tweaking of this framework to work or to develop human flood interaction model for this short term or operational level models. We need to somehow adjust the framework, but that framework is for long term flood risk management. And I will briefly give you an example, a case study, effects of formal and informal institutions on flood risk management in St. Martin. St. Martin is a Caribbean island, it's located in the hurricane belt and floods are the most frequent with hurricanes and then after hurricanes or related to hurricanes also sometimes or isolated events. Hurricanes are the most, floods are the most important, the most frequent disasters of St. Martin. They have densely populated centers and they are located near, which are located near the coastline. It's a very small island, there is a lack of sufficient stormwater infrastructure and if you see here, these are hurricanes that pass within 100 kilometers radius of St. Martin until 2017 when we did this work. And then after major, St. Martin is a Caribbean island and the economy is based on tourism, it's very, very, very important. And if you see here, after Hurricane Luis in 1995, the number of tourists dropped significantly and then, you see, after, and then it only starts to go up after three years, it's the same level it reached there. And then another hurricane, Hurricane Lainey, which was a big category five hurricane happened, and then again goes down. So it is very important, hurricanes, storms, but not only the wind, but also the flooding aspect. And also, if you see here in November 2014, we've had huge rainfall around 250 something millimeter in 24 hours, a lot of flooding happened. And if you see here, so flooding in St. Martin and many people suffer because of that, and it's a very common problem. So related to flooding, there are different policies. For example, existing policies, the St. Martin Beach policy, which if you see here, says the strip of sand with the wheels of at most 50 meters of which the surface consequences should not have any building. But if you go to St. Martin, there are a lot of buildings just next to the shoreline. So this is, for example, an institution, some policies and institution. And then if you consider also, for example, this is in Dutch, building ordinance, the building ordinance says that every house built in St. Martin should be elevated by 20 something. And then some proposed policies based on previous studies we met only hydrodynamic models. And what they did is that now they want to do is they want to, they want to elevate everyone who builds house in certain areas has to elevate their house by certain level. So in some areas, so if you see this part, so the dark red ones up to 1.5 meters and then the light red ones are in the regions in those light red ones up to 0.5 meters. So these are institutions. So now we put them in an institutional statement using the ID programmer. So if you say consider the beach policy, households must not build house any building within 50 meters of the coastline. If they do, they have to, they should enter some kind of fine sanction, but in St. Martin, usually they don't enforce that. So there is the sanctioning part is empty here, but it is considered as a rule because there is a sanction, but it doesn't happen. That's why we don't write it. And households must elevate houses regardless of their location by 20 centimeters. Flood from policy is from 0.5 to 1.5 meter depending on where they are situated. And then we include also this flood hazard reduction strategy. So governments implement flood hazard reduction measures if, for example, a number of houses in an area is greater than some kind of territory. Then we develop once we using the claim framework, we identified the components, the agents, the institutions, the physical process and everything. Then we develop a flow chart, a claim implementation flow chart to develop the models. So if you see here, for example, first we initialize agents, then estimate urban housing. So in this case also what we added is there will be urban expansion. So based on proposed plans, we kind of know which area they plan to expand in terms of urban development. So we include that and that means whenever you put, in principle, when you put any house, you are creating an impervious layer or imperviousness should increase within that catchment. So we change the imperviousness of that catchment in this case and every time a new construction happens, then we do that and then that changes the input file of the flood model. And then the flood model fits back. So that means whenever we run the simulation, then every time the flood model will run like a new model because it incorporates something more now. And the other element is if you see here, if they implement any structural measure that will also be implemented in the hydrodynamic model. So the next time the hydrodynamic model runs, it runs with all these aspects. And then if we analyze, so what we did is we run such kind of simulation is the 30 times step. Each time step is a year because most of the long term infrastructure change or buildings, whatever, they happen within years. It's not just the short time. This is the difference between the operational level and this is strategic level policies. And also activities. So if you consider reconstruction or recovery and mitigation plans, if you build a dive for something, you don't build a dive in a day. It takes quite some time, right? So that's why we use a time step of a year. So this is a 30 year time step. But what we did is that there may not be flood every year. So, for example, in this case, there is no flood in the first year, second year, there is a flood of 100 year recurrence interval, then five, zero, then in the fourth year. So a five year recurrence interval flood. So it continues like that. So some years there is nothing and then boom, a 50 year event happens. So then these are reflected here. If you see the beach policy, what we did, we play, we exercised some scenarios. We created scenarios to see what is the impact of the beach policy, for example. In this case, a beach policy, there is no beach policy. That means distance from the sea, the shoreline is zero. So if you remember the policy says that no one can build with a 50 meter of the shoreline. But in this case, we say, okay, what if it's zero? Then this is the impact, the number of flooded houses, total number of flooded houses looks like this up to 500, for example, in the time step 19. And then building, this is the actual one, the current condition, it stands from sea 50 meters. And if we increase that up to 100 meters, no one can build up to 100 meters. But then if you see these flooded houses, the total number of flooded houses, the effect is not very significant. Because there are already many houses built along the cost. So unless those houses are demolished, then the impact is not really allowed in terms of the beach policy, even after this saying, okay, no one can build. Of course, the number of houses that will be affected by the policy will increase, or the number of potential houses. But in terms of impact, in terms of number of flooded houses, then it is not significant. And then if you see the flood zoning and the building ordinance, so this is flood zones, this is for the building ordinance. Normally, the flood zone is a very small area, so the cumulative number of households that follow or not follow the flood zone, they are very small, up to 250, because of the spatial extent of the building ordinance, because it covers the whole area, the whole St. Martin, then the number of people affected by that is up to 3,000, right? And then if you see households that are flooded, but followed the flood zone policy, then we are talking about very small number. Here we talk about huge number, number of houses flooded, but that didn't follow. And if you talk about houses falls that are not flooded, but because they followed the building ordinance, then huge number. Here, maximum up to 60 houses in terms of the flood zoning, but in terms of the building ordinance, it reached up to 450. That means we are saving a lot of, maybe in terms of money, I'm not sure because this can be also poor areas or can be said there are some very, very fancy houses in St. Martin, there are also slum areas. But if you see in terms of number of flooded houses, that's from the social justice perspective, if we consider just not monetary values, but everybody affected should be accounted in that sense. Then number of flooded houses, the building ordinance affects huge number. So if people actually follow the building ordinance, then many houses will not be flooded. So that is the message. So our conclusion, one of our conclusions is that the building ordinance is an existing policy. So if the St. Martin government enforces this policy properly, then they can reduce the flood risk significantly up to 450 people sent here. While introducing a new policy, a flood zone policy, that adds people potential new builders to elevate the house by 1.5 meters, which is quite hot. That is, that doesn't work. So what we say is that focus on the building ordinance, which is a very, just elevating 20 centimeters, you save significantly instead of asking people to elevate 1.5 meters from 0.5 to 1.5 meters. So this shows that with these kinds of models, we can have a new kind of policy analysis system, decision support system for flood risk management. So if you need more examples, then I will point you to my thesis, maybe I will show that at the end. We also did a research in Hamburg, Germany. In that case, what we did was that we coupled also human flood models, but mainly focused on adaptation measures like flood proofing measures, elevating houses, also part of it, but also weight flood proofing or just moving valuables to upper floors and these kinds of things. Finally, the advantage of the claim model and the methodology to develop the hydrodynamic and the coupled modeling system. The whole thing provides a holistic and explicit conceptualization, holistic as much as possible including most components of the human and the flood subsystem. It's very explicit in a sense that we address both subsystems directly and it is designed to be very degenerate. That means in terms of, especially in terms of scale, right? We can have any, we can develop this model, this framework and the model for any area. In terms of technical aspect, we can have any kind of agent based model or any agent based modeling platform. We can have any type of flood modeling method and then we coupled them and then we have this. So the framework provides all the necessary concept and it provides flexibility in terms of model development and then it provides also a very interdisciplinary approach. For example, in case of Hamburg, we also added some psychological theory, protection motivation theory, how people develop this kind of protection motivation towards flooding. And you can have different disciplines and then, yeah. Excuse me, Yaren. Yes. Sorry to interrupt. Just wanted to point out that we have around like nine minutes left. So if you could kind of. This is the last slide. Okay, fantastic. We'll get to Q&A after that. And what are the limitations of the claim and the methodology? So because of, because it doesn't provide particular theory or scale or method, then modeling human flood, to model human flood interaction. So anybody can use any type of theory scale. That's a little bit difficult to manage. But that is the flexibility of it because it's very generic, but also that can be one of the limitations. In terms of complexity, yeah, level of complexity. I don't define how complex this model should be or the model representation in terms of being holistic. What is the level of being holistic? That is, can be questionable, but that's very highly subjective. If we talk about conceptualizing and modeling operational level institutions, as I said before, we need to adjust the framework somehow so that both the hydrodynamic model and the agent based model run simultaneously because you have to see the propagation, flood propagation and how people move with the warning. So it requires a little bit different tweaking of the framework. It requires a large amount of data that is quite a disadvantage because you have two subsystems, the human, the flood, and we need to have a lot of data. And in terms of also computational resource, the simulations, they require quite some time. That's it. If you want to learn more, then please check my dissertation. You can find it in the IHE repository or in the Theodore repository. Thank you very much. Thank you, Yaret. Thanks a lot for that great presentation. We'll now begin with questions. I'd like to request my colleague Long to please share the questions that we have so far on the screen. In the meantime, I would like to begin with a question of my own, Yaret, if you don't mind. What scope there is within the claim framework to take into account certain operational realities, such as let's say corruption, such as the quality of governance, municipal governance and all that. Would it be possible to factor in these kind of variables so as to say in future iterations and future work using this model? Actually, we wanted to add that. It's a little bit politically also sensitive aspect. So that's why we didn't include it because we work closely with the government of St. Martin, also part of the ministry. But yeah, there are other types of models, not human flat models, but from the social science domain, there are models showing a level of corruption. It can include it because it's an important institution actually. And there is a way to also parameterize it so that you can model it. And yes, it is an important aspect. Okay, look forward to talking to you more about that at a later point. Let's get to the audience question. The first one is from Melissa Mellison. How can the model address informal institutions like unwritten agreements, agreements that change depending upon specific context and actors? How big is the risk of oversimplification of institutions or human behavior using statements in modeling exercises? So kind of similar to the question that I had, I suppose. Yes, informal institutions. So for example, this word of mouse kind of institutions, these are informal institutions like, yeah, many people do it, I follow and then I do it because others are doing it. It is possible to implement it as long as we know, I mean, so agent-based models work like this. You have agents, you have these institutions in other agent-based models, you see them as rules of interaction or set of rules. So the set of rules, you define them in such a way. For example, one is if there is a statement. So if something, then this, if not, then that, you know, like that. So you can have all kinds of, let's say, possible scenarios to cover that. And then agents, they have this, we introduce randomness to just simulate the heterogeneity of agents. So they can choose whatever they want. When we parameterize it, that's how it goes. Then they say, okay, if this agent with that randomness selects one, then it goes to that then statement. So if then, right? So it goes to that then statement. So this is a bit more technical aspect, but that's how you implement it. And then how big is the risk of oversimplification? Well, this is very difficult. This is what I said is very subjective. Also, it depends on data availability. It depends on computational resource. As I said, like this is Martin model. It took some three, four months to run the whole simulation. Just the simulation part. I mean, not the model conceptualization, building the model and whatever. Once all this is completed, just run simulations in around 40 or 50 machines. I run these simulations. It took about three months. So how all these things matter, but then you need to consider always sensitivity analysis and certain analysis. These are important just to show also that, okay, I simplified something here, but the implication of this oversimplification or whatever, how you put it is this. So you use this sensitivity analysis and certain analysis to tackle that aspect. Thank you. The next question is from Maya. I hope I have pronounced your name properly. Who asks if you have taken into account the modern concept of living with floods and living with water, such as floating villages in Amsterdam and floating houses in Vietnam. Well, so agent based models you developed so that you in this case the physical model that is physically based model and the hydrodynamic model is based on some kind of physics, right? The Navier-Stokes equations, 1D, 2D equations, that doesn't change wherever you apply this. Whereas the agent based model is based on the local situation. Who are the agents? What are the rules of interaction? So it really quite depends on this locality. So if you talk about Vietnam or Amsterdam that policies are different. I'm sure people way of living different economic social economic situations different. So all these things will be in fact, so the agent based model will be different. You know, even if you apply exactly the same type of hydrodynamic simulation. Of course you cannot have because the topography and the system also can be different. But even let's say these two are exactly the same you replicate. But because of the change in the agent based model that's the human subsystem. People way of thinking flood perception and this and that. By the way, in terms of flood perception, for example, maybe in Vietnam, people may have better flood perception than in the Netherlands. They don't know what to do when it happens. They just leave everything for the government. So the government of course spends a lot of money through the Ministry of Infrastructure and other institutions. They invest a lot of money and they do something. But yeah, local people usually they don't know a lot. So all these human elements, they are important. Gopal Kumar asks if there are any independent outputs from ABM or is it or is it that based on the scenario of humans subsystem human subsystem modeled ABN. The couple modeling framework produce outputs of the resulting impact on the flooding scenario. I'm not exactly sure the question but if I understand properly so you have the agent based model. You have the flood model. So whatever happens in the agent based model. For example, if some houses decide to build in a flood prone area and then when we run the simulation, there is a flood. Then that house will be flooded obviously, right? But imagine they elevate their house by one meter and then that decision. And then when you run the flood model and then that time at that time the flood generates, they say a 30 centimeter flood. Then that house will not be flooded. So that house is not. But if there is up to 1.5 meters, then that house will be flooded. So this depends on the tool how the agents decide or what do they decide and how the flood system looks like. That's if I understand the question correctly. Thanks a lot here. Mr. Gopakumar, if you would like to rephrase your question and put it to you to hear it again, please do that in the chat box and we'll try to get back to it. If we have time. Chef Shivalking asks, in the case of St. Martin, to what extent does the model take into account developmental developmental activities that do not adhere to the rules? Yeah, we put a lot of different types of so what we did is that some of the box plots that I showed in the in the results. These are different levels of enforcements and different levels of they say adherence or really following the rules. So you put different thresholds. So what if people follow they say around 50% of them follow the rules? How does it? What what is what will happen if around 30% only? Follow what will happen, but you don't specifically mention or point out which ones. This is a random process in the model. And randomly around 30% of them, they don't follow and you see the output so but you run this let's say 1000 times so. 1000 times different random behaviors right and then you show the collective results in a in that. In the chat. In the box plot. So yes, we considered a lot actually. Thank you. The next question is from Alexis who asks what can be strategies or or measures to take for the management of the consequences of floods in underdevelopment in underdeveloped countries like Rwanda where poverty is the highest. Well, this. Yeah, this depends on. What you what you have, what kind of flood, the type of flood is important. If you have costal place you. You implement certain type of measures if you have flash flash implement a different type of measure so this is a big to generate question. Difficult to answer because all these things. I mean the type of flurry is very important and then the what kind of policies to have and those policies are dependent. You, you develop policies based on the type of flood you have so. Yeah. Thanks, I will just share the different screen for the next few few questions so please bear with me for a second. Yeah, do you see my screen and do you see some questions on the screen. Yes, like before. Okay, fantastic. So, okay, the next question is from Alexander who asks what is a mechanism to integrate the urban expansion and watershed as it is shown in slide number 21. What is the mechanism to use the agent based the agent based parameter as input variable in modeling. Mechanisms to integrate urban expansion. So this you can do it in different ways. We kind of simplified that part we know from urban planning department in St. Martin. We know the areas they want to expand so we know the areas but we don't exactly tell agents okay you're going to take the next this lot or that lot but agents kind of choose whatever lot they want to develop so they. They do that but you can also have another layer of model that you can have because the flexibility of this is that this the framework is that in the human element you can have economic models for example econometric models that define some income and socio economic aspects. Also urban expansion. If you have for example cellular automata models that shows how urban development will happen. Then you can also integrate all these kinds of things is possible. Of course it will increase the complexity of the model. And that can be a challenge or not because sometimes this the same models usually they don't fast but it creates a layer of uncertainty on all this. The already existing uncertainty so that's possible but yeah in our case in the St. Martin case we knew the area where they can develop so we just agents kind of select those area from yeah I'm kind of already existing. Fine. Thanks. Time OSHA Durrani asks if this model can be applied in a watershed in an Arab region where the area is highly populated or urbanized and it's it by urban floods or torrents of flash floods. I don't see any reason why it will not be at a word. Any population but I don't see it by urban floods. Yes. Yeah you can you can apply this in any. In any area. Maybe there can sometimes be. I mean the framework I'm talking about the frame of coming developing the model is another second layer right. First is the framework so all these aspects are in any urban area. And then if you go to for example a rural area maybe there are other things important maybe you need to add other concepts in the framework then if those concepts can be included in the. Either in the hydrodynamic or in the field model that's perfect. Sorry in the hydrodynamic or the agent base model that's perfect you are done. You still use two models but that is will not be covered in any of these. For example if you consider irrigation or something then you need to have a maybe a different order that. Care of the irrigation component then you have three different models and then the framework is also kind of evolved in this case as a different framework. Because it has additional layers and you should exactly know how those elements link with each other so that when you model that or when you start to conceptualize. Then it's easier for you to understand so but yeah it's possible to do it. Thank you. We'll skip a few questions in the interest of time so we'll skip a few questions from our direct colleagues I hope they understand such as Maria. Sylvia asks besides the elevation building rule do you take do you take into account the construction material I suppose in the modeling the material. No not in this case because it was not important for the. For the problem we had so you conceptualize the model for a problem so first thing the most important thing is what is the problem we find the problem and then you go to model conceptualization. Right. Question from Margie Cerega raise would it be possible to factor in the community awareness on the policy or institutional arrangement also what about the land user change in the urban settlement. Is that also factored in. Yeah the land use that's what I say like when you in this case specifically only when you when you change. Let's say previous area into impervious area like by building houses then we consider that. But if there are other types of change also you can do that as possible that's also what we say in terms of expansion and this you can have other types of models. Let's say for example like several automatic kind of models. Community level awareness I mean we have the in the humble model for example we have we considered perception but individual level perception. But when you talk if you talk about community level awareness. Yeah maybe that's a little bit different. It can be included in the agent based model. But I'm not exactly sure. Technically I'm not exactly sure how. Okay enough. Tom Lou asks did you use the protection motivation theory to generate the behavior of local people towards flooding. If so what are the challenges using this theory and what are your recommendations to improve it. Yeah it's. I really recommend to talk to. A psychologist not a psychologist sociologist. Well it comes from the psychological science. Actually from behavioral science so if you know anyone in that is very good to talk to them especially those who work it on the protection motivation theory will be good. Because it started with. How to stop cigarette smoking for example like that. How so developing a protection motivation right that's how it is so. But then there are a couple of papers actually related to flood using protection motivation theory. How to how people develop that and you can do it in different ways. You can see what we did. But I know people also use. If you have the resource. This way is based on the protection motivation theory can have a survey and then you. Can't parameterize most of the elements of the protection motivation theory. And then you can use it. You know in the model because anyway when you model you quantify things so you cannot just use. Qualitative stuff in terms of usually you kind of. Change to numbers parameterizing. To do the more. So it's good to include. People who have experience on this. Or domain expertise I would recommend that. Thank you. Sorry everyone we have exceeded the time that we had promised but if you don't mind we would just like to hang back and. And address the questions that have come in should take just a few a few more minutes I hope that's okay with you yet it. Yes. So the next question is how to determine the boundaries of the system, especially the institutional ones. And well that for example. If you talk about parts of a city then like a municipality then you are talking about that specific municipality then. If you talk about then what are the important. Policies that affect that specific municipality. But if there is a regional which is a higher layer now there is a regional rule. That affects the. I mean that that's directly related to the problem you want to address then you have to consider that but that's an external factor. Whatever you do within your agent base model will not affect. Or change. That rule because that's beyond. The power. The mandate of that local setting. So. It is possible. But you should know exactly based on your problem which ones affect. Your system. And then yeah, then you. Continue with your money if you consider the whole Netherlands then you know it's a different thing. But if I consider I live in depth then. This is a small city then that's something different. Next question sounds like it is straight out of your PhD defense. Mr Farhalen asked. If there is any way to validate the parameters or the architecture of the. In the case of fish behavior we can see the emergence of schooling behavior. Is there a similar way to see if the flood model correctly results in the emergent phenomenon that is the sum of flood damage. Yeah. With agent based models actually these questions related to the agent based model because usually with blood models you have historical data and then. You calibrate your model and validate your model doesn't kind of straightforward and that's actually usually questions from hydrologists about these kinds of models like. But in terms of the ABM what usually we do is because people's perception change and isn't that. We use expert opinions how. For example in the same model. After I run simulations. I showed that to the disaster management people in Saint Martin and then they say oh this is a bit exaggerated. So. Then what I had to do is that I had to change the parameters. In talking to them and then okay let's fix the parameters in certain way change the parameters and then run simulations and then they say okay this seems acceptable. So but then of course I have done a lot of sensitivity analysis with the parameters if they're actually very important then you need to know more about that parameter. Maybe if you need to collect more data then you need to collect more data some parameters. They're not. Some reason some parameters do not cause any sensitive sensitivity on the reason. But yeah that's that's actually how we did in this in this case. Yurma Paul Romero asks that in his previous study Dr. D. Balthasare at all included the people's collective memory in his modeling of human blood interactions is the claim framework capable of integrating such factors such things as collective memory. This is what I we said about this collective actions collective actions are a little bit different from individual levels by no agent based models on that. But we didn't consider collective actions in collective memory like collective recollection of past events those kind of. Yurma if but that is I would say that is perception actually like if you have the perception of flooding. If you were previously affected then you kind of know what you remember but people also forget for example in St. Martin the in 2017 there was huge event. Maybe if you followed the last webinar Dr. Zoran Vojnovic he mentioned that. Hurricane Irma but the the previous big event was 20 years ago almost the same day almost like one day difference actually in terms of 20 years ago many people kind of relaxed you know it doesn't happen and these because in 20 years nothing happened. So the first years you are very anxious and this and that but then slowly you kind of get used to the norm what they the normal situation and then no after 20 years extremely huge. Beyond category five I don't know because one of the comments I have with this suffice and scale category one two three four five that's related only the wind speed doesn't tell about anything else. So that is extremely strong hurricane actually and it's it destroyed the. The island so this this memory related to flood. It is very dynamic actually. But if you want to model it you can have it in the in the case of. Hamburg we have as perception but you can have some kind of graph or I don't know there are studies you can find and then see how memories fade and then you can include it from empirical studies you can include that in the morning is possible. And the very last question that we have and also the very last question we can possibly address given the time is again from your ma which is if the claim framework is capable of explaining how communities could be flood resilient. If communities. If communities could become flood resilient like does it explain what constitutes flood resilience the framework. It doesn't explicitly define resilience because we address. Hazard for not everything exposure we use that framework that's the concept of the risk. But if you have adaptation measures and many people don't get flooded or the risk reduces in terms of monetary value or in terms of number of flooded houses. Then that shows. There is some kind of resilience and especially if. Over time. You have five 50 years run or 100 year run and then you see that decline then that shows the resilience so I mean resilience. These concepts are quite complex because everybody defines them in certain way in their own way. And becomes a little bit. To be honest. Complex so but somehow resilience is related to also vulnerability also right so because we did a vulnerability analysis in St. Martin quite comprehensive I mean you cannot imagine how many things we add in that. And that is beyond any kind of resilience definition or something because we added so many things. And you can see if we say this is whenever then it includes the resilience concept also behind me so this depends on how we define these things. Thank you we'll have to stop there. So we will. Long we don't have any more questions to be. No we don't ask the last question. Okay, thank you. So with that we have come we seem to have come to the end of the proceedings. Thank you yet for your presentations and for your patient questioning, answering of the questions. And thanks to everyone for turning up in good numbers and for your questions and comments thanks. Thanks everybody for your patience because we run over time a good 20 minutes over time. So, as Maria Laura has has already has already put in the chat box, she has put certain links and those are the links on which recording of this of the session will be should be available by tomorrow or latest by Monday next week. So, these are links on the water channel website, and on the it website and the it YouTube channel. We'll see you at the, at the next webinar, which will be on December 9. And the speaker will be Dr. Rosh Rana Senge from IHE who's the coordinating lead author of the IPCC report that we are all familiar with and in this webinar Dr. Rana Senge will summarize the key findings of the report on sea level rise and how 33 different climatic hazards are projected to change in different regions of the world. So that will be on December 9. And until then, thanks again, everyone, and goodbye. Thank you very much. Thank you for you, for you to Abraham Long and the water channel and Maria Laura and it for organizing this. Thank you very much. And thank you for the audience. Have a nice day everyone.