Clean Water Works
CLEVELAND, OHIO: From the Northeast Ohio Regional Sewer District, an in-depth and fun conversation led by Donna Friedman and Mike Uva on any and all topics related to clean water, wastewater treatment, stormwater management, and the people, projects, and programs serving Lake Erie and our local waterways and communities.
Clean Water Works
Advances in Urban Flood Management
Cutting-edge modeling techniques are being developed in an exciting collaboration between the NEORSD and Massachusetts Institute of Technology that focuses on pluvial flooding, promising a more nuanced view of water flow across urban landscapes.
This "rain on mesh" approach divides an urban topography into small, data-rich segments, unveiling a new level of detail in analyzing surface runoff, infiltration, and evaporation.
We explore the strategic use of various models, from the importance of selecting appropriate tools for specific tasks to the potential for these solutions to be adopted by other cities.
George Remus. Welcome to Clean Water Works a podcast on water Clean water, water-related things.
Speaker 2:I'm Mike Uva.
Speaker 1:And I'm Donna Friedman. And here we are.
Speaker 2:Great Clean Water Works.
Speaker 1:Is a podcast brought to you by Northeast Ohio Regional Sewer District.
Speaker 2:Yeah, and we are returning from thanksgiving break. Did you have turkey?
Speaker 3:I did. I had turkey a couple ways smoked a turkey, what like in a smoker?
Speaker 2:yeah, you have a smoker, I do, oh yeah, I've started making smoked.
Speaker 3:at one time I I was hosting I I would make turkey three ways. I would smoke turkey, I would do the conventional roasting of the turkey and then I'd also make turkey soup.
Speaker 1:I don't remember getting any invites. Did it get lost?
Speaker 3:I don't know what happened. Must have been all those mailers out there somehow.
Speaker 1:I just want you to remember this moment when you ask my team for data requests.
Speaker 2:So three options for turkey in your household. That's very accommodating.
Speaker 3:I love Thanksgiving. I love making the food, love hosting the roasting. And making of the soup was always the roasting of the turkey, always part of a traditional turkey Thanksgiving. For me it was the smoker. I just kind of got into the whole barbecue thing. You're hooked. I just decided let's give it a shot and I made it, and now you have to do that every year.
Speaker 1:I do, but it's okay. Did you make stock out of your I?
Speaker 3:did? We got a family recipe to make stuffing oh and um, I used the turkey soup to help flavor the stuffing and that's like the secret ingredient that I now will share with everybody. So you don't have to buy stock in the box or whatever. Make your own soup and that just makes it so much more flavorful.
Speaker 2:You're a pro, we're learning.
Speaker 1:We're learning so much, so we have George back on to talk about a collaboration with MIT.
Speaker 2:Well, we've had George on before to talk about stormwater and stormwater modeling.
Speaker 1:We're talking about data.
Speaker 2:We're talking about real numbers, facts, figures, talking about better ways to predict and track flooding in urban environments.
Speaker 3:That's right.
Speaker 2:How long has this collaboration been going on?
Speaker 3:So they were interested in collaborating with us near the end of 2022. So Katia at the time she was a PhD candidate was looking for a program to help her with her thesis. Who's Katia? So Katia Bukin. She was the PhD candidate from MIT, massachusetts Institute of Technology.
Speaker 1:And she's a doctor now.
Speaker 3:She is. Last Tuesday I got to watch her Defender dissertation on her thesis, and it was you know she did a great job.
Speaker 2:Cool, so she's studying the same kind of things that you are studying at MIT, and her thesis was on it was all about this pluvial flooding. What's pluvial, which is pluvial. There's pluvial and fluvial right.
Speaker 3:Yes, it's tough right, One is an F, the other one is a P. What?
Speaker 2:do those things mean.
Speaker 3:So fluvial flooding is a lot of what we traditionally look at here with our regional stormwater system. So you think of streams, you think of the floodplains, you think of the water that gets out of the stream channel into the floodplains and causes all that flooding. That FEMA the Federal Emergency Management Agency is always focused on is watching those big storms get out of the stream channel into the floodplain and causing all sorts of flooding and all sorts of economic damage because of that.
Speaker 3:Or coastal flooding. So if you think of storm surges and how that could impact the capacity of pipes and streams and that can cause some flooding, that's kind of what you refer to as that fluvial flooding. Okay, the Cuyahoga River it's an 800 square mile watershed. When you think of that flooding that takes place in some of those select communities on the downstream end, you're really talking about the fluvial flooding. Now, pluvial with a P is really what we're dealing with in urban stormwater systems, where the water is on the surface, whether or not it gets collected into a catch basin, goes into a drainage system of some sort and could cause problems, either on the surface itself, because it can't get to a place that can be safely collected, or it just starts cascading onto the surface, spreads out because of the pipe surcharging and then spreading around and causing that, while it's typically shallow water, it can spread out much further into an urban system.
Speaker 3:When you start talking about these local locations at a parking lot level, at a street catch basin level, there's so much city texture about where these roofs are, where these buildings are, where these parking lots are and how all that interacts when it rains onto the surface. The topography itself that we generally from our models, have bypassed and loaded the water directly to a stream or loaded it directly to the storm sewer itself. So this rain on mesh approach, where the rain falls onto this topography, you almost call it rain on topography. Yeah, what is rain on mesh? Yeah, so the mesh, it's kind of think of the surface.
Speaker 3:Yeah, what is rain on mesh? Yeah, so the mesh, it's kind of think of the surface. And you think of how the rain, where it can fall onto a rooftop, it can fall onto a parking lot, someone's backyard, into a forest it takes that big topography and breaks it up into really teeny, tiny, small pieces, kind of triangular pieces. The rainfall falls onto that individual triangular piece and then it has its own properties of how much water evaporates, how much water infiltrates, and the rest of it is surface runoff that goes to the next piece of triangle and all those triangles kind of make up this big mesh that you can now route whatever rainfall to whatever area and see where it goes from there.
Speaker 1:I'll route whatever rainfall to whatever area and see where it goes from there. So sort of like, if you're looking at Google View right and you're looking at the overview where you can see your rooftop. If it falls onto a triangular piece that's maybe like a park by your house and it's trees, then that might have a higher characteristic for infiltration than for runoff compared to if you were to draw. If that triangle was on your rooftop, it probably had a lot higher runoff.
Speaker 3:Yeah, the pervious versus impervious right. How much water can infiltrate into some grass or some soil? It's got a higher infiltration rate than the parking lot, the asphalt how impervious is your surface that you're having this triangular mesh represent?
Speaker 3:is a big deal right. So the more pervious you are, the less surface runoff that is generated when the rainfall falls onto it, because more of it can be infiltrated into the ground or absorbed by that vegetation. But if it's on a parking lot, there's really almost no absorption whatsoever. It just turns into pure runoff and then it can go to the next place, and the next place it could tell you the surface, the elevations, it could tell you the land type, and so you can use that to help better understand. You know where's the water going to go and how much water is going to be absorbed into the ground. You're taking what we would normally have is a 50 acre catchment, a roughly a million square feet that we represent, into one hydrograph, one runoff hydrograph that we load to a stream segment or to a below ground pipe segment. This rain on mesh is like a thousand or three orders of magnitude more detailed, a thousand times smaller.
Speaker 2:So it's a much better understanding of where the rain is falling.
Speaker 3:And then where does it go on the surface? One of the things that we're learning in the urban landscape is we spent a lot of time trying to understand how the pipes and the streams, what kind of capacity can they handle, and focusing on making sure that they have enough capacity to safely convey water without causing flooding, and that's what a lot of our municipal ordinances, a lot of our federal regulations, state regulations, are all about making sure that you build a bridge or culverts or a detention basin. How much water can it handle, safely convey without causing some kind of issues. With this rain on mesh, though, you have all this runoff on the surface, unless there's like a catch basin or a yard basin. This water can go all sorts of places, and you think about how old our general area is and what kind of design standards were in place at the time, and how some places get inspected and maintained regularly, and some of them don't get inspected or maintained regularly.
Speaker 3:You can have clogged catch basins. You may have places where water is going that people aren't aware of, and so there's a lot of water runoff surface runoff that this rain on mesh will show you in a lot more detail. Where it's going, it could be causing some flooding that we weren't aware of where the source was. We have a lot of customers in the service area. Some of the times they call you know we've seen clusters of calls where our models show that there's plenty of capacity in the pipe. It's not surcharging and getting onto the surface, it's actually starting on the surface and so this rain on mesh, this pluvial focus type of modeling, really helps us understand more, gives us more insight to the cause and source of problems.
Speaker 2:So MIT Katya reached out for our data.
Speaker 3:Yeah, they're looking for a program that had these catchment models to do initial start with and wanted to compare those findings from the catchment model to replacing that catchment topography hydrologic model with this Raynon mesh and see what happens. And so they needed someone who had detailed catchment models and a lot of data to share, because to build this much higher level detailed model you need a lot of data to compare to and make sure it looks good, and we have one of the best data collection programs in the country.
Speaker 2:What did Katya do with the data that we gave.
Speaker 3:So it started with making sure the catchment model was functioning just fine, which it was, and then it was just a matter of replacing our catchment hydrology with this rain on mesh.
Speaker 3:Then looking at a series of storm events that we had a lot of flooding, and so there were three big events in 2023 that we focused on July 2nd, on the east side, where we actually had some pretty significant underpass flooding.
Speaker 3:That took place July 20th, where we had a lot of surface flooding throughout the surface area, and then August 23rd of 2023.
Speaker 3:And so we have a lot of great data that was able to be used to compare to it.
Speaker 3:We have some field photos, we have some monitors, we have some trail cameras, we have some spherical imagery, and then we had lots and lots of GIS data to share and then several years of customer and media reported flooding that we've been collecting over the many years that the program exists, and all that was used to help run the model for our storm events, where we used rain gauges and gauge-adjusted radar rainfall data and compared the flooding that was either recorded from monitors, from field photos or customers to what the model predicted, from field photos or customers to what the model predicted. And the good thing was, a lot of these surface floods that were being reported or measured were lining up with the Raynaud mesh model. If someone was reporting flooding, the model was predicting flooding, oh okay. And so if the monitor was showing a high rise of water in the system, or overtopping and causing roadside flooding, the Raynon mesh model was doing the same. So we were validating the model with all this great data that we've been collected and shared with them, proving that it worked.
Speaker 2:Proving that it worked and in return, did the Raynon mesh give us a better understanding of those events?
Speaker 3:It did. One of the most interesting things that came out of this analysis was there's a lot of underpasses in our service area. Think of railroad underpasses, think of highway underpasses, essentially these underpasses which are a low point along somewhere on a road. It's a low point, so you think of all the things that kind of can drain to it, and one of the things that we were finding is this underpass flooding wasn't necessarily the result of the below ground pipes having lack of capacity, surcharging. It was starting on the surface and then cascading down to the slow point, and some of these locations the rain on mesh model was doing a you know very insightful to show where could the water come from, and it wasn't just nearby locations. If you think of that cityscape, that topography, there's some places that are really high up where, if with enough steep incline think of a catch basin is supposed to capture the water along a roadside gutter and if it's intense enough, all that water isn't necessarily being captured by that catch basin.
Speaker 2:It's just rolling on past.
Speaker 3:So if it doesn't get captured by that catch basin, it goes down the street and goes to the next catch basin.
Speaker 3:Well, if that's not capturing everything, it keeps cascading, cascading, cascading until it gets to the low point, and that's typically where these underpasses are. It's not only capturing water, but it's also transporting sediment and debris and trash that gets collected on the road, collected on the road, and then, when it makes it to the low point, which is typically where these underpasses are, those catch basins that were meant to capture some runoff could very well be clogged. So now you have this big pool of water that is very slowly draining and sometimes, when these storms are happening in the dark, you don't see where that water is and how deep it is or how unsafe it is. And that's one of the challenges that we ran into is, some of these storms had those deep underpass flooding where people had to get rescued. So we have pictures of these cars being with emergency rescues from those events, and this kind of helped show where the source of this water could have actually been coming from.
Speaker 2:So prior to this information from the pluvial analysis, we wouldn't have had as good an idea of where the water was coming from.
Speaker 3:Yeah, the fact that it's so detailed. But all that water that started on the surface that never made it below ground, this really gives you more insight to what's happening on the surface that never made it below ground. This really gives you more insight to what's happening on the surface, particularly for these big extreme events.
Speaker 2:So, looking forward, what are the impacts and the opportunities that this collaboration presents?
Speaker 3:Well, I think what they're trying to show is a new approach to understanding the source and what to do about it from an urban perspective, and so Cleveland is kind of like one of their case studies to help demonstrate that you know what can you do with a catchment model, what can you do with a rain on mesh model, and then help you understand what are the sources of the flooding and then what are the proposed projects to mitigate it.
Speaker 3:In this underpass flooding piece, I think is one of those that is intrigued both of us, because underpass flooding happens everywhere in the world and there isn't a framework necessarily to deal with that specifically. So this may be the beginning of working through that. Better quantifying the flood risk on the surface that's something that we've been trying to do here at the district and this is something that will come out of this project. But to calculate the flood risk across the entire study area, where before we really focused on our regional system and some extensions of the local. This starts on the surface, so everything ultimately gets quantified accordingly. So it gives us more insight to share with the communities and can we notify officials when the flooding, like an early flood warning detection system, to share it with others. Can we close off the road to prevent people from driving into deep, unsafe waters?
Speaker 2:Was this really eye-opening? This collaboration and the results that came from plugging in our data to MIT's Rhein-On-Mesh.
Speaker 3:Yeah, I want to say in some areas it was definitely more validation of what we were trying to do here at the district with collecting this data and then be able to apply it to such a big area with so much detail and help demonstrate. Because when you build a model, you know one of those classic sayings is all models are wrong. Some are useful. So if you're going to create something that's three orders of magnitude more detailed than our current modeled approach, you want to have a lot of data to help demonstrate that it produces findings that you can have confidence are correct, so you can give insight and direction and recommendations for others to take action. So being able to collect all that data and help validate in so many locations was really important for taking something that we've not done before and then give us confidence that it's giving you good results that we can come to our communities, think about under our programs and provide insight to others.
Speaker 3:We had customers reporting flooding and our models weren't showing that flooding was coming from below ground. Why is that? You knew the customers were right it was. What analysis could we lend ourselves or to our communities to help understand why that problem is caused or is being caused and what to do about it. So now we have this new tool that we can use for that. So, when is the rain on? Mesh approach, the preferred approach, when might be the catchment approach, the preferred approach, and then creating this complex model, with it to provide much more insight to a much bigger area, to provide, hopefully, more value than what we had done before, is the ultimate goal.
Speaker 1:Does it take more computing power to run the Raynon mesh for the same amount of area?
Speaker 3:It does, it does, and that's, I think, one of the original purposes of the catchment model. Originally, it was developed in the 70s and the first thing you think of in the 70s, well, the computing power may not have been as strong as it is today and much more data to share, whether it's electronic GIS or field photos, even or custom reports. All that as we collect more data, have stronger computing power, a lens itself to create more detailed models, to help try and understand and explain more questions.
Speaker 1:For our listeners. One way of understanding how we use our models is if we were to build a project so if we're redesigning a stream channel to have floodplain capacity we might want to see what level of storm that project can handle before it starts to be overwhelmed and flood out into the surrounding area. So we can build the system in the model and then run various storm events. So we can run a 24-hour rain event or a five-year rain event and see how that system that we built works, based on the model for that area. And so it does take a lot of computing power when you're continually tweaking the system. So maybe you add a basin or maybe you expand the floodplain even more, and then you have to rerun that model.
Speaker 3:So yeah, we're always actively managing the model. So, as new construction goes on, one of the things we try and do is identify where that takes place and then make sure we update it based on new construction. And I do want to highlight that not all areas require a highly detailed model to understand the source and causes of problems and what to do about it, but there are cases where, in some cases, it's just a complex location and what we're finding is there is opportunity to learn a little bit more about it than what we were doing before.
Speaker 1:I'm thinking of, like some of our flatter communities like Brook Park and Middleburg Heights, where sometimes that flooding it is kind of tricky because the topography is so slight, the changes are so slight yeah absolutely, and you can represent that cityscape with higher detail.
Speaker 3:That might give you a little more insight to where the water is coming from. And then where is it going after that? It may not be making itself to the pipe might not be making it to the catch basin. There may be some local drainage issue that is the single source of why that flooding is taking place. There could be things that are going on on the surface that you're trying to simplify with that catchment hydrograph, some problems that you're completely bypassing. That now you might have some more insight on.
Speaker 2:This is telling us where the work has to be done first.
Speaker 3:Where the problems are, where we think could be the sources of them, and then where the work needs to be done to help mitigate that.
Speaker 1:When you're getting into these models and you're getting into, like, the very deep details of the models, do you ever like? What is this saying? Have trouble seeing the forest.
Speaker 2:For the trees.
Speaker 1:Yes, something like that. You know what I mean.
Speaker 2:Like do you?
Speaker 1:ever like run into the tree and you're like this tree is really in the way, but really like you need to be focusing on the forest. Like does that? Is that a problem that you run into? I?
Speaker 3:think, anytime you're investing a lot into data or trying to simplify the real world into a model and make it more complex because you think it's going to give you more insight, you always have to take a step back and go. Why did you need this model to begin with? And so I think that's one of the things we'll always have to be cautious about the model. It's a simplification and, going back to that old saying once again, all models are wrong. Some are useful. What could this be useful for? You always have to think about what was the end game goal of having this model to begin with. Our catchment models, its original end game goal of having this model to begin with. You know our catchment models. Its original end game goal wasn't to solve all pluvial flooding in the local system, but we recognize, with some of these really big storms that have happened recently, that there are problems out there that may need more understanding of why it's being caused and what to do about it. And so this pluvial approach to modeling this rain on mesh modeling looks like it's giving some of that insight that we never had before because we weren't necessarily looking for it. But if the end result of that was to help you with pluvial modeling. It looks like it's a good tool, then that's something we want to share with others.
Speaker 3:I think our models that we've been focusing on the fluvial, the stream aspect of things, it does a good job. If our models already doing our much more simplified model is doing a good job dealing with fluvial flooding, do we need to make it do anything else other than that? And I think the answer is probably not to make it do anything else other than that, and I think the answer is probably not. But if we really want to focus on this pluvial side of things and realize the shortcomings of this catchment model, well then it sounds like this radon mesh approach is probably the better use of that for an urban landscaping area where those pluvial sources may be starting on the surface.
Speaker 3:What are you trying to solve here? Which model approach makes the most sense? And that pluvial you might just want to have it for a very isolated area like a neighborhood. I think the part where people can get caught up is if they try and make the model answer everything versus the one thing that it really was intended to do. What's our leadership? Say, is the juice worth the squeeze?
Speaker 1:Just do? They say that.
Speaker 2:That's the title of this episode, the right tool for the for the right job.
Speaker 1:We're just a whole, whole bag of metaphors today.
Speaker 3:Oh yeah, we're full of happening here today. That, yeah, we're full of them. That's what's happening here today, that's right.
Speaker 1:Are you living the dream? That's my last question.
Speaker 3:Living the dream. I recognize there's a lot of problems due to flooding that we have a long ways to go to dealing with, a long ways to go to dealing with, and I think, as I have a little better understanding of what's going on and why, being able to help people mitigate those and then demonstrating that and then using that to apply to another place, I think is something that I look forward to doing. Right there's, you know, there's always a problem that needs to be solved that I think we, the district, can help with when it comes to the flood risk, and I think our program, as MIT called it, cutting edge. Hopefully it can lead to other programs wanting to adopt this. If it helps us and helps communities solve these types of problems, then it's a tool that is worth exploring.
Speaker 2:Thanks for joining us again, giving us insight into this new technology and how it can help us understand our flooding problem.
Speaker 3:Thank you, as always, for having me.
Speaker 1:Always a pleasure to be here and hopefully sometime we could talk about some other topics down the road, like what you're going to make for Christmas Ham three ways what are you going to smoke?
Speaker 3:Others host Christmas.
Speaker 1:See, I usually host Thanksgiving, so that's always like my frame of mind.
Speaker 3:I'm always willing to bring something, but that one's that holiday. That's my favorite, so it's the one I like to think about and make food for.
Speaker 1:Perfect.
