 Hello, good afternoon. Good morning. Welcome everybody to today's webinar. It is 11.45 here in the Netherlands. So I would like to start with the webinar today called Managing Flood Risk in Semi-Arid Data Scars Regions. My name is Leneke Knope. I'm from the Water Channel and this webinar is part of the webinar series IHE Delft Online Seminars for Alumni and Partners in Cooperation with the Water Channel. So especially if I will warm welcome to all IHE alumni and partners. Now before handing over today to the speaker I would like to mention a few things. So today we have another interactive webinar and interactivity can be expressed two ways. So at the bottom right corner we have a chat box and in the chat box you are invited to pose your questions to the speaker. So we will collect them throughout the presentation and then at the end we will pose question by question to the presenter. And also during the presentation some questions will be asked to you through multiple choice questions and you can simply select the answers in the pop-up screen that you will see. Thirdly I would like to ask you one thing. Just to have an idea on who is in this room would like to ask if you can write down your name and the organization where you're from. So our speaker today is a fresh alumni of IHE from the April branch and her name is Adele Young. She graduated in water science and engineering with a specialization in hydraulic engineering and river basin development and her research focused on early warning systems of urban flooding in Alexandria in Egypt using hydrological data. Adele herself is from Trini.n Tobago and she has worked there for eight years as civil engineer. And in this webinar she will present an introduction to the concepts of flood risk management and also an overview of some suitable strategies to cope with flash floods in semi-arid regions. And having said that I am very happy to hand over to Adele and wish you a nice upcoming hour. Thank you so much line okay. Okay and welcome to everyone. Thank you for joining. I know it's quite early for some people so those who made the effort to get up early this morning thank you so much for being here. The topic it's a pretty broad topic in the sense of managing floods but it's sort of specific in that regard that is looking at semi-arid and arid regions and looking at managing floods in those data scarce regions. So we have a lot to cover today so I'll try to do this as quickly as possible and try to be as clear. So given overview of the presentation we'll look at the characteristics of arid and semi-arid regions. Why are they so interesting? Why do we want to study floods there? I'll give an overview of flood risk management concepts. This might be new to some people or it might be something that you've heard before just presented in a different way. And then what is involved in flood risk management, flood risk assessments, what data sources can be used. We'll briefly look at some strategies that could be applied to flood flood management and then finally we'll look at a case study in Alexandria. So the first question as Lenike mentioned, just if everyone could put their names of their countries I would greatly appreciate it. But the first question that I have today is really what proportion of the world's land surface is classified as arid or semi-arid? So if you can just put that information just fill in the pool at the bottom so we can see what everyone thinks. Yes, does everyone put their answers in? Okay, so just to move on with the presentation at this point, the answer is actually 33% of the world's surface is covered or classified as arid, which really seems as a lot if you think about it and with the trends in desertification expected to increase there's the risk that that number might actually increase as well. Now if you look at this map that I have presented here, you'll see that a lot of these areas is located on pretty much every continent. So it's anywhere that there's a desert, there's the north of Africa, the Middle East, down in Australia as well parts of China and also in north and south America. Now these areas, what makes them so interesting is the fact that there's really not a lot of precipitation. At times of the year they really go through droughts, but it also means that there's a lot of, they are also at risk of flooding. So it's a very unique situation, well, in a sense. And if you look at this table here, it shows that there really is not a lot of precipitation with less than 100 in some regions in hyper arid regions and with a range of 100 to 300 in arid and 200 to 800 in semi-arid regions. Also to notice that this aridity index is also a factor of precipitation as well as a potential evapotranspiration which means that it's not only about the amounts of rainfall that these areas receive, but it's also about that potential evaporation. So how much water they're also losing. So why are they so interesting? Really they're quite unique and different from other areas because of their climate and natural geographic features. As mentioned they have infrequent rainfall, they're subjected to droughts, there's poor vegetation in some areas, a lot of erosion, steep slopes in some of their catchments, and also high sediment loads in their rivers because there's a lot of deposition, use of high transmission losses and that sometimes allows heavy debris to stay in the river once the next flood comes. Re-precipitation because the storms that they have are usually of smaller frequencies, a low magnitude frequency of storms and there's a huge variation in when this happens. So you may have a huge plant one year or a huge rainfall one year and then no rainfall for two years and then another huge rainfall after that. These are usually short duration events with high intensity and the way in which precipitation is caused is also very important because this can either be caused by convection clouds which is induced by high temperatures or because of frontal rains which is the case in a lot of those North African regions along the Mediterranean. Now because of these characteristics when combined it's pretty much a recipe for flash floods which is classified as flooding that occurs pretty much within one to six hours of when the flood events occurred. These are usually very random events. The hydrograph which means how the water, the flow of the water is distributed, it's usually a steep hydrograph and it results in smaller total flood volumes. Because of that they can be quite difficult to manage and result in a lot of damage. Now these are a few examples of floods in our semi-arid region. The first example is in China which shows excessive flooding which would have occurred after several years, months of drought, sorry, in this region. And the second example here is in Alexandria, Egypt, which is the second largest city in Egypt. They experienced a huge flood in 2015 and also in Jeddah in Saudi Arabia. They also had a lot of flood after some time of dry season. It's also important to note that these floods also occur in urban areas. It's not necessarily along river banks or other areas of the catchment but they can also impact urban areas. Also to be noted is flash flooding in these Wadi systems which are very prevalent in Middle Eastern countries. As can be seen in this example in Israel, because of the lack of water, a lot of these rivers, they don't have water throughout the year. But when the flood season comes, it's actually when water comes and with that comes a lot of debris. This example here on the right shows the before and after of a flood, a flash flood which occurred in Sinai. And you can really see the disaster and the power of that flood as it rose, completely destroyed here. So we're on to the second question. Based on the characteristics I've discussed, I'm wondering just how many people are in these regions, in the arid or semi-arid regions. So another reason why I asked this question is because I also want people to know that because these floods are so stated with the characteristics of flash floods, it's not necessarily only applicable to arid and semi-arid regions. So a lot of what we discussed there is going to be applicable to anybody who experiences flash floods. So don't think because I'm not in the semi-arid or arid region, this is not applicable to me. A lot of what we covered here will also be applicable to you. So no worries there. But I see we have 25% I believe that are in arid and semi-arid regions. So great. Moving along. Okay, so here we go. Looking at just to recap what some of those challenges are. We have large amounts of flood water within a short period of time. That makes them very difficult to manage and to forecast. In those regions, they're also significant, insufficient rainfall and hydrological monitoring stations, excessive damage and loss. And due to the lack of rain, as I mentioned, there's a lot of debris. And that infrequency also affects a lot of the type of infrastructure that you have in that area. As compared to other regions, like other tropical regions, which experience a lot more rainfall, there's less agency in a sense. And then it's not really as a priority to really justify the cost and large infrastructure projects. And in urban areas, as I mentioned, this is also prevalent in urban areas. We have other external stresses such as population increase and urbanization that increase this risk of flooding in those areas. So how do we manage these floods? What can really be done? In order to answer that question, the first thing we have to do is look at what is risk. We really just try to define what is flood risk and really what are the components of flood risk. So there's three main components of flood risk. The first one is hazard, which is looking at the magnitude of the flood and the probability of occurrence. Is this flood going to occur every year? We'll look at it every 10 years, every 2 years, every month, every week. Next, look at exposure. That is looking at who or what is exposed to this flood, who is at risk? So it looks at the human activity as well as the environment. And thirdly, we look at vulnerability. So what is the susceptibility of a region to losses from floods? And vulnerability can be expressed as geophysical, social, institutional, or economic. Another definition of flood risk is also the probability plus the consequences of flooding. And this is also tied back into the previous definition, which looks at the consequences of flooding is expressed as dependence on your exposure. And your vulnerability, so they're all linked. So when looking at flood risk management, the aim is really how do we reduce this risk? And that's really done by lowering your probability, so removing your hazards, reducing your hazards, preventing exposure to people, and then reducing the vulnerability of society. So if we can do that, hopefully the idea is that we can remove or reduce the risk. And this is achieved through the implementation of several structural or non-structural measures, and the last of those can be either policy or financial instruments. So the flood risk management cycle is one way in which we approach risk management. So at the first aspect, which is looking at the pre-flood measures that has to do with anything with defense or mitigation, or just pre-preparation. And when looking at the operational measures, these are real-time. So anything that happens while the flood is going on, or right before the flood. So it's real-time risk management. And then this post-flood measure, sorry, which has to do with relief or recovery. So it really encompasses all of this, but in terms of your approach or the measures that you choose, it's really based on your specific situation and the requirements and the capabilities of the users. Before there was more of a top-down approach when looking at flood risk management, but now we're seeing the benefits of actually having more of a participatory sort of approach, where you have input from all stakeholders. And also, because of the nature of these flash floods and these floods in these regions, because they're such quick and very sort of unexpected, that they really need to be focused on mitigation and warning to save lives, because you may not have enough time to save property or assets, which you might have in other river and floods where you have more time and more warning. And lastly, an important concept to note is managing residual risk, because the idea is that as much as you do, you may not be able to remove all risk. Therefore, it's really important to have these other measures such as having evacuation rubes or flood excavation exercises that can really add towards dealing with residual risk. And also, as I mentioned, there's several stakeholders involved, which is very important to this exercise. Which takes us to our next question. Oh, I'm sorry. Who are the stakeholders present today? So we have meteorological, hydrological agencies, maybe disaster reduction agencies, water managers, consultancies, academia, and we also have the public. So I see we have mostly from academia and a few from consultancies, so great. And the public, great. So moving on. Now we can't discuss flood risk management without really looking at insurgency, which is very prevalent, especially looking at the future. So there's insurgency in the climate, and these events are expected to increase. So there's also insurgency in the population as well as other economical situations that could contribute to changes in the future. So in the wake of this, there has been a lot of emphasis on the need for resilience strategies. And by resilience, it's a word that we hear all the time, and it's synonymous with really being able to absorb disturbances and remain functioning and recover as quickly as possible. And this is best described or explained in this graph here, which shows that if you adopt a systems approach to this, that for the first degree of disturbance, you would resist that. But after some time, as that increases, you really want to be able to be resilient and be able to recover. But as the magnitude of the disturbance increases, you will reach a point where you can no longer recover. And then there would be a regime shift. And at that point, you'd have to rethink how best to design for this or how to cater for it. But in terms of resilience, it's not only limited to design or structures and infrastructure, but you also have to look at your recovery capacity when looking at resilience. So you have to look at your social and financial capital, which is really important to resilience. Now, the flood risk assessment is a key part of flood risk management. The EU directive in 2007, they mandated that all EU countries do preliminary risk assessments, as well as hazard maps, and then come up with the flood risk management plans. So this first aspect is really looking at once you can define what your risk is, then you can choose best how to address these risks. And one way of looking at it is using a source pathway receptor consequence model. So the first thing is looking at the source. Now, the source is something you can't really change. You can't really reduce the intensity or change the river flow or things like that. So it looks at identifying corresponding frequency of your hazards, identifying flood extent and the adaptive velocity, so that's looking at your hazard. Pathways that has to do with overland flow of the plains, over topping, that is something that can be controlled to an extent. And then receptors are the inhabitants, like who is at risk, the industries. That can also be managed, but most importantly, what can be managed is the consequence by these different measures that we put in. So just to go through the quick framework of flood risk management, you define your hazards, you define your vulnerabilities. That will give you an idea of your risk assessment. You have to collect data in order to do your hazard analysis and also to do your vulnerability analysis. Once you define that risk, you determine what is unacceptable and what is acceptable risk. The unacceptable risk, then you have to plan mitigation measures, which could be land use, it could be structural, it could be non-structural. And any acceptable risk, that is something that you have to look at and constantly look at and review. So as I mentioned, flood risk maps are very important. It's a way to visualize what the risk is. So it also presents information in a way that's understandable for non-technical users. So the non-technical stakeholders. So you can define a hazard map, which gives flood extent, flood depth, as well as velocities. This will be combined with a vulnerability index map, sorry, which looks at probably building conditions. It will look at socioeconomic data. So these maps here will produce using GIS data. So it also gives spatial data. So you can actually see where that risk is. And then once you overlay these maps, it gives the risk map. So once again, it gives a visual representation of what area is at risk. And these can be very useful in planning developments, so people know where to develop, where not to develop, as well as when planning evacuation routes. So one of the biggest challenges that we have in flood risk management and modeling even is really unavailable or inadequate data. And going back to arid and semi-arid areas, this is also very prevalent because of the infrequency of these events. There are not a lot of hydrological or data monitoring stations. Because they also occur in some instances on a smaller scale, they're not necessarily global archives, or they're just too small to be monitored or for that information to be collected. And this, in fact, hinders our ability to do or build hydrological in-endition models. And, of course, if we don't have available data, then we can't calibrate our validation models. But luckily, with advances in technology, there are now a lot more open-source data sites. And also, there's a lot of emphasis towards using citizens as sources, so crowdsourcing data or citizen science, where we can actually collect a lot more information. So I'm going to try to cover a few areas and a few sources of data. Now, I know it might be a lot of information, it might be a bit technical, and in the absence of time, I'm not going to go into detail of every one of them. But I think it's important for you to have the information here, so if you want to review it later, you can always come back. Now, remotely sent data, what that is, it's really just the acquisition of any data, not directly, but from a distance, from another point. And it's really changing the way that we access data right now, especially in engaged regions where we don't have a lot of data. And this is done mostly through satellites, the use of satellites. In particular, with precipitation data, which is absolutely needed in order to do flood modeling, there are a lot of satellites available listed here, like Trim, CMORF, or ERA Interim. These are all freely available, and they are available at 0.25 spatial resolution, which is cool, but it's still quite useful in some areas. It really just depends on the area that you are studying. And it has a pretty wide global coverage, which means that it covers a lot of those arid and semi-arid regions, and the two lacks also cover global regions. And in terms of the temporal resolution, that is also important in flash floods, because when you're trying to monitor intensity, which plays a huge role, you also want to have a very good resolution. So this can actually help. Moving on to DEMs, the digital elevation models, which are also very fundamental to building models. There are a few sources of freely available data. For example, the SRTM, which is available from NASA. There are two versions. There is the 90 by 90 resolution version, and then there's also the data resolution. These are actually quite useful in larger catchments, and it's really important to have good detail when using DEMs, because if your resolution is too large, then you may not accurately be picking up the variations in each apography. And finally, there's the AXA. This is provided by the Japanese State Agency, and this is actually one of the most precise DEMs that we have available. Once again, it's all freely available, which is the best part about it. Although it's not in terms of the resolution, it's not the best resolution, especially in urban areas, it may not be completely applicable, because the resolution is too coarse, but in other regions, it can be quite useful. So that's just a little caveat that you need to keep in mind. But in terms of that, there are other sources in terms of using LiDAR data, which is light detection and ranging data, but that is not really freely available, but it is probably the most accurate, one of the more accurate, I should say, sources of remotely sensing DEMs. Next, we have historical flood data collection. This is also very important when you want to maybe validate or just to confirm that a flood did indeed occur. They are these websites that are available. Once again, they're on different scales, and it is quite tricky for smaller catchments because of floods occurring on such small scales, it cannot be detected. But luckily, I was actually able to find some flash floods occurrences in Egypt, which are quite relevant here. So there are these three sites that you can actually check out, and they do give good information. Next, we have flood extent mapping. This also uses satellite data, but in this respect, one example is using these different water indexes, such as the Normalized Difference Water Index or the Modified Normalized Difference Moisture Index. These are actually quite useful in terms of flood extent, which as I said before, you can use to do validation of your models. So if a satellite takes an image on the day in which the flood occurs, then using the band, you can actually use these indexes to calculate or to give a flood extent. And there are many sources of these bands of this data, such as Landsat or Modest and Sentinel, and they all have different resolutions. You have to choose which is convenient to you or which is most appropriate. There's also the use of Sentinel-1, which is a synthetic aperture radar. This imagery has the benefit of, or I would say the advantage of the others in this respect, that it can penetrate clouds. So if you are worried about cloud cover, then these can actually help. But this was only launched in 2014, so it's not going to give you too much historical data, but in the future, it would. These actually, in order to analyze these, you would need these different tools. Luckily, these tools are all freely available. There are many tutorials online that can show you how to process these images, so it's quite useful, and you can actually use them. This is an example of a flood extent mapping that was done doing the NDN, MNDWI, for the flooding that occurred in Alexander in 2015. So that has a face-by-face resolution, and from this, we can see that even with that cause resolution, it was able to show that flooding did indeed occur. So land cover, this is another important data feature that we need, and you can also use satellite information to do this using the supervised land cover classification and using the Semi-Automatic Plugin, which is also available on QDIS, which I should have mentioned before, is a freely available due information system software. And then secondly, we have OpenStreetMaps, which is a great tool to map locations. So we actually, at IHE, we've actually used this program quite often in terms of mapping areas. A lot of areas that have not been mapped because they're in very rural areas and people just haven't mapped them as yet. So we've done different map-a-tons where we come together and we build on these maps, and there are a lot of mapping communities. So the idea is that afterwards, once this data has been collected, then people can use them, or you can use them in GIS as well to build a foundation for roads and buildings. So they're actually quite useful. Now, citizen observation for data collection is something that's up and coming in a sense, and what it implies is that you're using citizens as either interpreters or as sensors. And why I thought it was important to really mention this here is that in arid and semi-arid regions where we discussed you may not have a lot of data, this can be used really to cover it or to substantiate a lot of that remotely-centered data that you're collecting. So if you can see in about the resolution, this really allows more data to be collected or more specific or defined data to be collected as they can actually help with ground-true things and also social-mini-mining, which means that you have access to YouTube videos or Twitter. If anyone's ever reported a flood, you know, there are opportunities to look at where indeed that flood did occur, what was the extent of that flood perhaps, what type of disruption there was, what were the impacts. So there's a lot of pushing the community right now to try to use this data as it is. It can be quite useful, especially when data is not available. And that can be used for water level, velocity, flood extent, land cover, and topography in some instances. Another example in the U.S., they have a program called the Flooded Locations and Simulated Hydrographs Program where they actually use this citizen science data and combine it with existing data to come up with locations of where flooding did occur. So if you're interested in that, you can check that out. And these are a mention of some other projects that also use a citizen science data. So if you're interested, you can check them out. Okay, so this takes us to question number four. Now, this is just to see if anyone was listening. What is the biggest limitation of using remotely sent data in arid and semi-arid regions? I think I said it a hundred times, so maybe everyone will get it. And don't worry if you don't get it because this is all a learning experience. It's fine. Okay, so just moving on, the answer is actually C, the third one, which is yes, cost and temporal resolution. Because as I said, some of those areas are much smaller than say large river and floods. So you have to be very cautious when using remotely sent data. So the next question I have is, have you ever used a citizen observed data? So this is, have you ever used a flood image? Have you ever looked on YouTube to ascertain if a flood occurred? Oh, I'm seeing some news. That's okay. So as I said, it's really some research that is up and coming now that they're using more. But there is a lot of potential because people don't realize that every time you take a picture of a flood, if you knew exactly what to look for, how to take that picture, it can actually be quite useful. Okay, moving on. So we're going to come back now to looking at flood mitigation strategies. Now, once again, in semi-arid and arid regions, you are quite limited in some of the measures that you can take. For example, as I said, the necessity or maybe the emphasis may not be there in some regions because in terms of balancing the economy of it, because if a flood only happens every one in three years, then you may not be prioritizing. But looking at the mitigation strategies that you would want to take on, they would really focus either on reducing the hazards or reducing the exposure or reducing the vulnerability. So for example, detention basins, which are used to really store the water and to reduce flooding downstream, is a good example of a strategy that could be used in those regions, as well as clearing obstacles. So as I said, in some areas where the rivers are dry for most of the year, if there's a constant process of clearing obstacles and once the flood comes in, that would most likely reduce carrying debris downstream. River child improvements are also an option, as well as terraces or check-downs in regions on slopes that can actually slow down the water. Just to look at some of the other examples here, which would focus on non-structural measures, as I may have mentioned before, flood forecasting is actually a very attractive measure in the sense that it's in what we call a no-regret measure. It's something that no matter what you do, if you do it, you will see the benefits of it. And because these floods happen so quickly, it is a little bit more difficult to do. But as I said, you have to react very quickly so having that warning is quite useful. And also land use planning, knowing where to build, where not to build. These are more long-term measures that can take place but are also very useful. And flood insurance, which is something that is not necessarily available in all countries, is actually a very important measure that could be used to protect people in the future from disaster as it really allows you to recover a lot quicker. So in terms of the strategies for these areas, it really just depends on the climatology in the basin. As I said, some areas may be larger, some areas may be smaller. It depends on your socioeconomic conditions. But real and truly preferred solutions should be flexible. It should be a mixture of resilience and adaptable measures. And most importantly, it should case a friend's certainty. So just to... I believe I mentioned some of these things here before. Really, you want to have things that are no-regret measures where you're not going to lose anything by doing them and they will always benefit you. And then also having this balance of small-scale structures so it may not be necessarily a large dive or something. That's probably not suitable for some of these areas. But really having projects on a much smaller scale and also really an emphasis on nature-based solutions. These are ones that have multifunctional purposes and they have sort of environmental and societal and economic benefits. So for example, using these retention and detention funds, and this could be quite useful in arid and semi-arid areas as they can act as artificial recharges for aquifers. So that was just a few examples. Moving on to flash flood early warning systems. As I mentioned, it's a really important measure and in the SENDI framework for disaster risk reduction, early warning systems have been highlighted as a very important measure that should be implemented. And what it does is it allows you also to mitigate against these residual risks, as I mentioned previously, that risks you just cannot seem to design for. And you have to learn to live with these but it allows you to recover quickly and hopefully to make you more prepared. The approach, the system to early warning system is composed of detection, forecasting, warning and response. So it's not only just about forecasting when a flood would occur but it's also about how you issue that warning and where you issue that warning and where the warning is issued for and then finally response to really what sort of action is that you do take. And the level of sophistication of these early warning systems really just depend on the capabilities of the users and the technical capacity. It's also how to create the help, how severe these floods are and how frequent they are. So for example, along the Rhine here in Netherlands you would really have to have a sophisticated system because these floods happen every year. You know they're going to have these floods and obviously the Netherlands is at risk of flooding. But perhaps in semi-arid and arid systems you don't need very elaborate systems. So it really just depends on the risk that you have to quantify that. As I mentioned, I was talking about the Rhine. A lot of these systems are quite elaborate. So you collect your data, then you can use numerical weather prediction models as well as quantitative precipitation forecast in order to forecast the rainfall. And then you use hydrological and hydraulic forecasting systems to forecast water levels or inundation models to further extent. And once you have that information you disseminate your information. However, in arid and semi-arid areas we don't have that data. Sometimes it's quite difficult to build these models and they also require a long run time. And as I mentioned, because these floods occur in such a short notice, you may not have time to run all these models especially when you use these quantitative precipitation forecast. But these are actually quite useful, these quantitative precipitation forecast in the sense that they exist on different scales. So they can either be short term or medium term which has to do with the lead time. So they allow you to forecast maybe three days ahead or in respect of the medium term they go up to four to 15 days, for example. And this really helps to build longer lead times or to allow longer lead times but there's also a lot of uncertainty available with that. There's also a move towards probabilistic forecast which means that these forecasts give you more probability that an event might occur. So it looks like you likely heard that an event would occur. So in that respect it can give you the probability of occurrence of events and it also looks at uncertainty so it allows you to look at the insurgency of events as well. And right now there are many of these systems around the world but two that I will mention here is the ECMWF which is European Census and Medium Weather forecast and as well as the NOAA they have the Weather Research and Forecasting model. Now as I mentioned with these flood forecasting systems because you want to have a very short lead time well you may have a very short lead time you want to be able to make decisions as quickly as possible. And in flash flooding one method that is used is a threshold-based system and this method here, this one in particular it's proposed that instead of running these models you can actually directly compare critical rainfall thresholds with the forecast rainfall and it will take into consideration initial moisture conditions but in some instances maybe in urban areas because it's so highly urbanized and it's impervious that maybe that soil moisture condition is not as relevant but the idea is that you can do that and once that threshold is exceeded then you can issue a flood warning. An example of how that is used is the Flash Flood Guidance System which was first developed by the US National Weather Service and they actually, what they do is that they try to estimate the amount of rainfall for given duration that's required to cause a flood and these warnings are issued when they think a flood might occur and I was actually in a state recently and I received many of these flood warnings and this is just an example of how they might occur so it really is something that can be done quite quickly and it's very applicable to flash flood areas and it has been done in a lot of other countries such as in Central America and quite recently in the Black Sea and in the Middle East they are also using this method. So question five. What is the biggest hindrance to efficient flood early warning systems? I've given quite a couple of options here so is it data gaps, is it poor availability of data, is it lack of integration of risk information or is it governance or organizational issues? Okay, so everyone's chiming in. Okay, great, that's nice to see. So this is actually quite interesting because I want to say that maybe this was a bit of a trick question because the answer really is all of the above but the answer that I was really looking at was really the fourth answer which is the lack of integration of risk information and communication and dissemination of mechanisms because the reality is as well as you build these models and as well as you forecast it doesn't make any sense if you can't get this information out to people and people can't respond. So while they are all very important and they should be all considered having proper communication and dissemination is very, very important. Based on that, I just wanted to mention this very quickly which looks at impact-based warnings which is a way that we're trying to, they're trying to move warnings in a direction where you can actually have more descriptive warnings so people, it's not just about there's a flood, there's no flood but it can actually give you an idea of what depth of water, what area and so forth and what the impacts are most like, most importantly and this is an example of how it might be done operationally and this is what is used in my office in New Yorkie where they combine impacts as well as probability in order to issue warnings so whether it's a yellow warning or a red warning so it's either severe, there's a high likelihood or if it's just to be prepared or whether to take action. Moving on, so now we're at the point where we're going to look at the case study so it's really everything that we've talked about so far like how we can really apply it to an actual study and this would have been part of my thesis when Alexandria, once again, they have a Mediterranean climate but it's still based on the annual average precipitation it's an arid climate but it's one of those situations where they have a lot of uncertainty so one year they'll have 300 millimeters of rainfall and one year they'll have 75 millimeters of rainfall so it's really a lot of uncertainty and variability in terms of the rainfall it's the second largest city in Egypt and they have the largest ports and industry in Egypt and they have a thriving tourism industry but in addition to that they also have a lot of vulnerability so there's informal settlements there's in some instances inaducative drainage infrastructures and there are some low-lying areas which make them very susceptible to flooding, sorry. In 2015 they had a really bad flood it was probably the worst flood that they've ever experienced seven people died, a lot of areas were impacted transportation industries were impacted and it really demonstrated a lack of resilience in the country and what we noticed is that that region that Middle East and North Africa region because of their arid and semi-arid regions climate the studies have been done that have predicted that they are going to be more exposed to flooding this is all part of the anticipatory flood management project which hopes to increase resilience and increase participation, stakeholder participation it looks at what sort of measures that you can take before flood that would actually reduce the impact however in order to do this study there's a lot of work that needs to go into so part of my thesis was really looking at what sort of first approach that we can take while these studies are being done because all now the country is still at risk of flooding so what sort of interim or immediate action you can take so the idea was really to build an operational approach a operational based early warning system so this as I mentioned before would have used the ensemble forecast using these rainfall thresholds that would have been derived as well as coming up with a decision based rule in order to determine when to take actions or when not to take actions and then you can characterize that in a hazard matrix that will take probability into consideration as well as the likelihood of a hazard and potential impacts so the important part, another factor that was also looking at how to identify what those critical rainfall thresholds were so because it's also a very data limited region as I would have mentioned I had to figure out how to collect rainfall data how to collect historical data so I would have used YouTube I would have done social media mining when a flood occurred how bad was this flood so I would have also have to speak to persons in Egypt about their information on this flood as well as local drainage capacity data so this is a method that I use it uses what data is currently available but it can be updated once new information becomes available and once you, just to mention here very quickly that once you have these thresholds the idea is that to combine it back into the risk matrix so the hazard matrix that I showed previously and to see how well you could have used this ECMWS data which I would have used how well you could have used that data in order to classify hazards of past hazards whether they were there was a high likelihood of flooding or there was a significant likelihood of flooding and so on so this is why as I mentioned I didn't have any data so I had to go online and look for floods and I used YouTube blogs and newspapers and I was able to find that in 2010 there were floods so I used the historical floods for one period so I had testing or calibration period which was from 2010 to 2012 and then I used that to derive these different rainfall thresholds so I saw in 2010 there was a lot of damage and a lot of flooding for a 20mm rainfall and then I also looked at a flood in 2011 and this showed also significant flooding and this was for a 50mm flood a 50mm rainfall, sorry, excuse me so at this point I would have assumed particular thresholds and this also would have considered rainfall frequency as I said what is the design capacity of the system and just made some assumptions of what these rainfall thresholds would have been and assigned based on exceedance probability assigned any different classifications so no flooding, minimal flooding minor flooding and significant flooding so all these different things and as I said this would have been limited data once again but it's just really testing a concept here and how well it can really be used and as I said in my study I would not have I would have assumed minimal changes to the system as well as some more conditions did not have an impact because the system is so urbanized so here are a couple floods that I looked at the same 2015 flood as well as the 2013 flood and from here I saw that I looked at forecasting up to 96 hours and from this information it showed that you really can't really predict anything from 98 to 72 hours and 48 hours it's still not giving great results because if you compare it with the observed here you can see that in some instances it was able to accurately predict it in some areas it was not but up to 6, 12 and 24 it did show a good representation of how extreme those floods would have occurred in some instances not to agree but you can actually see in this instance when you actually had major floods it was able to predict that when you had smaller flood incidents the models did not perform as well and just to bring this up again looking at the flood extent I was able to validate some of these areas to show that flooding did indeed occur for this 2015 event and that would have been using information from the drainage department as well as newspapers and the University of Alexandria so it shows once again that the SAFLAID data here using this remotely sensed data it was able to help map a flood extent and if you have citizen science observed data you can actually go back in now and fine-tune some of that information that can be used to validation so, here we go, final thoughts I've just wanted to add just a few words I think a few important takeaways here which is really there is no sorry, there are many components of flood management but there is no one blue print it really just depends on what the situation was and knowing how to approach it really depends on that in iron and semi-iron areas challenges flash floods are challenged with flash floods due to the nature of a climate geographic and meteorological characteristic so you always have to keep those in mind when choosing what sort of measures you use and what sort of approach you take to flood risk management an effective integrated strategy requires the use of both measures and these can be implemented over the short term or the long term and you have to care for uncertainty and these must be resilient and flexible measures and in data sketch regions as you would have seen there are many opportunities to supplement the data gap for my research I would have shown that it is possible to predict extreme events and flooding and issue warnings so this is really just a first step but it's something that can be done initially and as you get more data you can update your system and then finally you have options so it's really what options are you going to take will you do nothing or will you wait for the data or would you try to capitalize on what information is currently available so that's it I think I'm a little bit over time but I'll take any questions now thank you very much Adele for your presentation your very clear explanations of quite a lot of content and complex issues so we have received a few questions to start with let me put myself up also there's one question from Simon and he says thank you for raising impact of floods and risks very relevant and he just received images from another colleague who is in Yemen in Mukalla showing severe flooding and he asks which organizations stakeholders in your experience should be responsible for the warning of floods like who should it be national or local governments etc maybe international organizations what is your good question and really and truly it is something that requires stakeholders to work hand in hand so for example in terms as I mentioned there would be detection forecasting, warning and response in terms of detection and forecasting those responsibilities are usually taken on by your mythological and hydrological agencies because they will have the technical capabilities of doing that but in terms of actually issuing the warnings what we suggest is that they work very closely with your disaster agencies and they should work hand in hand to actually issue those warnings so in terms of when a flood should occur and what not usually I would say the hydrological agencies but in terms of issuing what sort of what can be done like what response would be taken and things like that to alert people I think a disaster agencies should be involved in that but once again nobody is working singularly all have to work together and slightly related so the pictures that were just received is from a citizen and you mentioned in your presentation that with just a little bit of maybe training or knowing what to post and how that added value could be made by the social media or the crowdsourcing of data and you also refer to several websites but is there something that you can refer to like an example that really worked well something that you came across a lot of literature there's a lot of literature one of the tables that I placed that article is actually a complete literature review of Juno's and particular instances where they have all used citizen and safe data so they have used it to validate information and also those projects some of those projects are actually up and running and are associated with specific projects with everything within the beginning everyone is trying to figure it out because I think one of the limitations of that right now is really how to standardize these methods of collecting data there are some examples and I think it's really a great opportunity for people thinking what can I do, I'm not really doing anything but people don't realize that the potential that they actually have we will also upload the presentation so the slide that you were just referring to is for everybody to feel also later on then I have a next question from Oklaia is there a difference in managing flood risks in urban and rural semi-arid regions can you say something about that even though you're a research focus person in urban? I would say that particularly because urban eras the drainage system exist so you have you don't only have for example in rural eras you only have flooding probably from the rivers so you have overflow from the rivers and that would flood but in urban eras you actually have flooding from your sewer system so that's really complicated things in a sense it's a physical the dynamics of that is added complexity as well as you have more exposure, more people are exposed whereas in rural areas perhaps you have a lot more flood plains so there's not as much exposed so there is actually absolutely a difference in how they're managed it also limits your measures because you have a lot more space in rural areas whereas in urban areas you're very limited in space and you have to be very creative next question from Johanna is there a level of confidence you should set to decide whether or not to warn the population can you say something about that well I think I think yes I think it also depends because you can have a lot of false warnings absolutely but I think this is something that builds over time when you look at your these forecasting systems and one of the important ways of monitoring like how well your forecasting systems are working is looking at how well they were able to monitor events and from that you can it's something that's constantly being tweaked in terms of your threshold so your thresholds don't stay constant if you've noticed that and you're constantly getting false alarms or they're constantly being underestimated and I think you can adjust that but absolutely there needs to be a level of confidence people have to have that level of confidence in making those forecasts then I have a final question by Okunbambi so what about the influence of the Mediterranean Sea or whether on precipitation or hydrological modeling okay the Mediterranean Sea well the case study that I did look at was only Mediterranean Sea discussed one thing I did notice or one thing I did observe about that whereas in some instances like flash flooding might be convective so it's based on like temperature right along the Mediterranean Sea most of their their rainstorms or their their systems are from frontal rain systems so whereas what that means though is that these forecasting systems they can actually forecast that rainfall a little bit better I would say then say convective systems where because they occur so quickly because of that change in temperature it's a lot harder to do that so in terms of the hydrological modeling I'm not sure if they were referring specifically to developing the model in terms of data availability but you can actually you have that data available and I do think it's compared to other regions it might actually be easier in that sense but once again there is still a lot of variability in those circumstances I'm not sure if I asked this question thank you very much for your presentation if people want to stay in touch with you I saw you shared your email address and what I understood is that you are now also working to publish your research and is there anything we need to know from you for the near future how to keep in touch no I think well my email address and it's a Gmail email address so I think I should be easily then I would like to thank you again and thank you IHC DEL for all the support and thank you for all the participants to stay here throughout and we hope to see you at the next webinar thank you very much