Historic Flood Models – Ghosts of Our Past

Episode Transcript

Mollie (Host)
Welcome back to {Brief}ly Speaking, the SEH podcast where we tackle the topics shaping architecture, engineering, and construction with quick expert insights. I'm Mollie Sabo, your host. Since it's Halloween season, we're talking about something that continues to haunt communities, developers, and infrastructure projects across the country, the ghosts of flood models past. Many floodplain maps and flood studies still in use today were developed decades ago using the best tools, data, and assumptions available at the time. But communities have changed, watersheds have changed, technology has changed. Yet those historic models can still influence where development happens, how infrastructure is designed, what permits are required, and even whether property owners are required to carry flood insurance. For communities and project owners, understanding these legacy studies isn't just an academic exercise. It can mean the difference between making informed decisions and inheriting risks, costs, regulatory challenges that nobody anticipated. So today, we're digging into the evolution of floodplain modeling from E431 to modern HEC-RAS and exploring why understanding the assumptions behind older studies can help communities make smarter decisions about development, infrastructure investments, and flood risk management. Joining me today is Jordan Thole, a project engineer specializing in floodplain modeling, stormwater management, and water resources. Before joining SEH, he served as a floodplain and dam safety engineer with the Wisconsin Department of Natural Resources, where he worked with FEMA flood maps and helped communities navigate complex floodplain regulations. So you could say he spent plenty of time uncovering the ghosts of flood models past. So Jordan, welcome to {Brief}ly Speaking.

Jordan (Guest)
Thanks, Mollie. I'm excited to be here.

Mollie (Host)
Let's start with a little bit of history. For listeners who may not be familiar, I threw this in the intro there. What was E431 and how did floodplain modeling look when many of these early studies were being developed?

Jordan (Guest)
Yeah, so you can think of E431 as it's kind of the first widely recognized flood model that any department of the government or anybody was really using to try to map hydraulic channels, streams, rivers, ditches, anything like that. You know, now we have maps everywhere and they're really common. People don't probably think too much about them, but starting in the mid 70s, nobody had really taken any attempt to really map and understand what these floodplains look like on a broad scale. Some communities had mapped some historic floods in the past. Some of them had maybe tried to do some really simple pen and paper estimates of what the floodplain was. But E431 was the first big advanced step towards a computerized flood model. And how it works, and I'm going to give just a quick kind of nerdy background on it, but how this all works is what we call it's a standard step-back analysis. So we always think flooding, if you think about it, when we talk about like a watershed or anything like that, you're thinking water, rain falls on the ground, flows downhill, and the flooding progresses downhill. Flood modeling looks at it backwards, the step-back side of it. So we're not looking at from the top down, it's starting actually in the downstream side of the river and working its way upstream. So what E431 did, and this is still the way a lot of flood models work, is you're starting with an assumed flood elevation or assumed channel slope on the downstream side. You're assuming an elevation at that point, and then it's gonna take that flow rate and that information that you have, go one cross section upstream, and estimate the new flood elevation based off of the channel losses and all the flow and everything going in between those two cross-sections. And then it checks what the flood elevation downstream comes back as. And if it doesn't match what I think it's supposed to do, it iterates, goes back, changes the number a little bit, tweaks it, and then goes back and solves it again. And it does this over and over and over again until the equations balance out. Once everything balances, it goes the next upstream cross-section. So it keeps stepping back on its way upstream. And this was revolutionary when E431 came out. You know, people had tried to do it by hand, but doing these computations over and over and over and over, well, that's what computers were designed to do. It takes out that human part of having to do this over and over and over. And so that's where this first big step took place. But computers at the time that this was developed weren't the little laptops that we carry around with us now. This was a computer the size of your house that costs millions of dollars to run. There were only a handful of them across the nation that could run these sort of things at this time. And the models themselves, we didn't have CDs, you didn't have thumb drives, you didn't have the internet. You didn't even have floppy disks for those of us who are still old enough to remember the old like big floppy disks that we carried around. They ran on punch cards. It was literally a stack of numbered cards that were punched out at different iterations. And you fed that whole stack of cards into the machine and it would read them out and then run the computations off that. So the amount of data that you could put into these models was very, very limited because somebody had to go through and hand fill out and hand punch out each one of these cards. So that's where, you know, a lot of the modeling started and there's a lot of spots in the nation still that they've never advanced beyond that because they're so cost prohibitive. That's still what's being mapped and what we're still looking at here today, you know, 50, 60 plus years later.

Mollie (Host)
I think that background is so important just to show the growth in this field alone, but how revolutionary it was at the time. And it still is great data, but how limited it is now too. When you compare those early models to today's HEC-RAS capabilities, what are the biggest changes in how engineers approach floodplain analysis?

Jordan (Guest)
That's a really good question. And there's a lot of detail in there. I like to try to explain this to somebody. It's like we're trying to paint a sunset. This will make sense in just a second here. We're trying to paint a sunset. We're engineers. We're maybe not great at painting, but that's what we're tasked with trying to do. It's really complicated. There's a lot of colors. There's a lot of things going on. And in the 1970s, somebody was sitting there trying to paint it with finger paints. They were doing what they had, they had the primary colors and a few details and somebody's trying to paint And when it's all said and done, you can tell it's a sunset and it looks pretty good, but it's still done with finger paint. It's only going to be as good as you can. The only thing that's changed over the years is our tools. So now we've progressed. You have better paintbrushes, you have everything else. Now we could paint a much prettier sunset. It looks a lot better, but it's still never going to match the real world. There's just, there's so much complexity and so much stuff that's going on in the real world that all we're trying to do is take a perfect picture and use an imperfect approach to get as close as we can to what's actually going to happen during a flood. And that's really kind of how things have changed when we look at this transition from E431 and then later came HEC-2, and from HEC-2, there were also a lot of spinoff models that different agencies came out with, like, WSPRO or MIKE that were really focusing in on specific details that people were interested in within a floodplain. Dam-Break was another one looking at what happens when a dam fails or WSPRO  was primarily focused on bridges and structures because HEC-2 was really limited in that. And then in the 90s, it all kind of came back in to HEC-RAS trying to paint a better picture of what the floodplains look like. But at the end of the day, we're still trying to take a really complex system and run it through our computers in a way that makes sense that when it's all said and done, accurately reflects what's out there, but it's never going to be perfect. There's so much variability in nature that we can never get a perfect picture of it. But that's what we're trying to do. So every one of these steps that's come along the way, they've added detail and what we can model, what we can look at. Whether we can look at more complex flow regimes, whether we can look at bridges, it gets more into the physics of how water is actually flowing, reducing the number of assumptions that's in there - really started to paint a better picture of everything, but it's still lacking. There's still a lot of stuff that we just, we can't do. The computation power is not there. And models are continuing to improve at an even faster rate. And next year we're going to have a better picture of things than we do this year and it's going to continue growing that way.

Mollie (Host)
I love that example. I think that paints a really good picture of how it has changed, but it's still the same. It's just more detailed there. And like you said, many communities are still kind of relying on studies that were developed decades ago. So why do these older models still tend to stick around for so long?

Jordan (Guest)
Yeah, they definitely stick around a long time, sometimes longer than they should. But the first reason that always comes to mind that any community will tell you this, they're expensive to replace. Once FEMA has published a map, once they've put it out there, it's been published, sent out to the communities, it's now the community's responsibility. FEMA's not gonna pay for every change that has to come into there. It's really cost prohibitive, especially for a lot of these smaller communities, to say, hey, we recognize our maps are out of date and that they could probably benefit from some updating. But to hire an engineering firm to come out and resurvey everything and update the hydrology, update the hydraulics, remap everything, put together the full package and everything to send to FEMA to go through the package, I mean, just sending that to FEMA costs $8,000. For a lot of these communities, you're talking a hundred plus thousand dollars to send this stuff in. And it's a capital issue. And there are things that you can do to try to offset the cost of that, but it's still, it's an expensive and frankly, very slow process to get that stuff updated. Honestly, the second reason that some of these stick around for so long is that old doesn't necessarily mean wrong. You know, earlier I was talking about that standard step-back equation there. That's still how a lot of flood mapping is done to this day. Some of it's because that's how the regulatory framework was set up within the National Flood Insurance Program was to use that. But in a lot of cases, if you were living on a river that's relatively straight flowing and a well-defined channel, that methodology is really time-tested and proven and it's reasonably accurate. So there are a lot of situations that they weren't updated because they didn't need to be updated. And that's the tricky part is being able to pick it apart and be able to tell what needs to be done and what could maybe just benefit from some updated mapping.

Related Content: 4 Steps to an Effective Flood Management Plan

Mollie (Host)
Yeah, it's weighing that balance of is this minor sometimes amount of extra detail going to be worth the extra cost that can be fairly high, especially for smaller communities. And like you just said, many people will hear that a flood model is 20 to 30 years old and immediately assume it's outdated, but that's not necessarily always true. It's trying to find that balance. So how do you help communities figure out if it is necessary or if it isn't?

Jordan (Guest)
I mentioned earlier, I just briefly mentioned that, you know, in a lot of cases when you have a straight stream, well-defined banks, everything's very stable. The models are probably still reasonably accurate. And that's where I think you need the engineering approach or the scientist who's familiar with really how the hydraulics and the models work to help highlight any of these key changes. Because there are some really base assumptions that went into those older models that can really change the accuracy of the mapping. The one is that the older models assumed that the floodplain and floodway, everything flowed downhill, directly straight downstream. And I know that seems a little redundant. Water flows downhill. Nobody's going to argue that. But water doesn't always flow in a straight line. And that's what HEC-RAS is always assuming. It's always going to assume water is flowing straight down the channel. But we've all seen rivers that flood and all of a sudden it reaches up over a bank somewhere and floods across some farmer's land for two miles and dumps into another river. And we call those lateral flows where it's flow that's no longer going downstream. It's going laterally into a different area. Sometimes it breaches a natural levee or into a low point, and then it floods into a really large wetland complex or a really large storage area or something. Those historic models, again, it always assumed water just keeps flowing downstream. It didn't account for storage. It didn't account for any of that. And so you run into some of these situations where if you know that those are on the landscape and you feel like that might have a big impact, well, that's likely going to be a major driver to whether or not those maps are accurate. The other thing that we're seeing more and more now is the impact of urban development and also what that does to some watersheds. I don't think anybody's going to argue that if you put pavement on something that used to be a grass field, you're going to have more water that runs off pavement than you did off of a grass field. It's just, it happens. And the approach that development took across the Midwest and across the US for a really long time was we tried to mitigate the rate of discharge. So we were trying to slow down the peak amount of water that was being discharged off to at least match the existing conditions with the thought of like, well, we're not going to make things worse because we're at least pulling that back. And what's really started to show up over time is that flooding isn't just a function of peak discharge. It's not just looking at that initial big surge of water. It's also sometimes an impact of the volume of water. And so the other part that people maybe don't always think about is, yes, if you have a large paved area, you have more water that runs off really fast right away. But even if you put a detention basin or something in there, you also have an increased amount of water that's running off because it's no longer being able to infiltrate. It's not soaked up in the soil. There's none of those losses contributing to that. And so even if they slow down the water, it's still more water going downstream. And the cumulative impact of that across the board is also part of why you're seeing larger flooding downstream is because the amount of water that's being discharged has gone up. And that's the other part that we can try to look at is when we're doing any of these updates is to see if that discharge of water is still accurate. So that's I think kind of what we have to look at when we're talking about, are these models out of date? There's a lot of situations where they're not and they're fine, but there are some places where the assumptions that drove those historic models no longer are true. And we have to be able to recognize that, say, okay, well in this case, then maybe it bears looking back at and seeing if there's something that should be changed or tweaked to change those flood elevations.

Mollie (Host)
I think it's good to know. or comforting for communities to know that there are conversations that can be had that can help direct them and clues that could lead to you should probably redo your flood models or it's probably pretty accurate we could get a good idea based off of the old models. Now you brought up a couple of terms in that answer there that I want to hit on here. So one concept that often gets confused is hydrology versus hydraulics. They sound similar, they're different. Can you explain the difference in simple terms and why it matters when evaluating an older flood study?

Jordan (Guest)
Hydraulics is maybe the obvious thing that people think about when you talk about flooding. It's the river, it's the stream, it's the channel, it's the water flowing across the landscape. That's the hydraulics. And that's the easiest part for people to kind of wrap their heads around. Hydrology is probably the more important in a lot of cases and more difficult part to wrap their head around. And that's the study of when rain hits the ground looking at how much is captured and stored in the ground, how much infiltrates, how much runs off, and what that resulting discharge looks like. That's the hydrology. And I think that's where a lot of people lose sight sometimes of the flood mapping when we're talking about floodplains and stuff too, is because we all focus on the hydraulics because that's the easy part to look at off the cuff. But hydrology is definitely this own complex monster in and of itself sometimes, because it's, again, we're trying our best to model a very detailed world with these imperfect models. And so we have to make some really broad assumptions. So we talk about things like the 100-year flood, the 100-year storm, you know, what is that? You know, how does that change? And we're basing it off of all this historical rainfall data, but the rainfall data has changed over time. Climate change has affected how that rain is arriving. In the Midwest, a lot of areas, we're still getting the same amount of rain in any given year, but the duration and intensity of those storms has also changed. We're seeing higher peak rainfall events, stronger discharge, some areas are seeing lower flows, and so you're kind of aiming at this moving target as well. which is why, you know, we will get really accurate when we're talking flood elevations and we'll give you elevations down to a hundredth of a foot for FEMA. But most engineers will agree that as far as hydrology goes, plus or minus 30% for that peak discharge is about as accurate as we can get. And it's not that, you know, we're not doing our jobs or, you know, this is wrong or anything like that. It's there's just there's so much variability in nature of - Has it rained recently? Is the grass cover good? Has the farm down the field, have they cut their crop down? Is it bare soil? Is it a big pasture of hay that they turned over and planted corn in this year? Things change so much year over year that a six-inch rain event today isn't going to be the same amount of water that you'd get running off in a six-inch rainfall in October, and it's not gonna be the same as what you'd do if you got six inches in March. So we're trying to again, paint this imperfect picture that's reasonably accurate, that's as good as we can do, but that's definitely, I think, the harder part of it that sometimes people overlook is that hydrologic modeling.

Related Content: Here’s Why There’s Water in the Street After a Rain Storm

Mollie (Host)
I think it's so interesting hearing you list out all those different potential factors because there is so much and it can be so minute, but make such a big difference. And I just have never thought about it that way. So when communities say our flood model is out of date, What is actually more likely to have changed than the hydraulics or the hydrology or both?

Jordan (Guest)
It varies based off of a lot of different factors. I will say hydrology often varies based off of where you are in the US. There are areas in the US where your rainfall information is changing faster than it is in other places. Some states haven't seen a huge impact due to climate change. Some states are seeing very big impacts due to climate change. And a lot of it's also looking at what data do you have available? You know, really as far as hydrology goes, the best data we can have is stream gauge data and rainfall data to pair it with it. So the USGS has a number of stream gauges that are out in rivers all around the US that are continuously recording flood elevations, flood depths, and flow rates in all of these rivers. And what we'll do is we actually go back and pair it with that rainfall data to match, okay, well, they got X number of inches of rain over this period of time, and this was the result in the stream. And the USGS is also updating this information constantly. And so looking at their newest reports and stuff that are coming out and reading those, saying, hey, has there been a big change in the hydrology in this area in the last 10, 20, 30 years? Are there new gauges that have come online or have we had record floods that maybe skew that data set and maybe it should be higher than we thought it was? Or when it was done 20 years ago, we only had 10 years of data, so we didn't have a lot of information. Now we have 30+ years and we've never had a flood even remotely as high as they previously estimated it. It can go either way on that. Hydraulics is probably the easier part to just double check and look through that. I talked earlier some of the major assumptions where the old models fall apart. So looking at your community and going, hey, do we have any spots where breakout flows can happen? Are there storage areas? Could those impact the flood elevations? And we've definitely seen situations where that has a huge impact on it. The first one that comes to mind, there was the city of La Crosse, Wisconsin, kind of my hometown back there. They had a flooding issue in a smaller stream coming off the bluff there called Ebner Cooley that the new maps that were coming out had this massive floodplain because it was using the older 1D model and it had water that left the channel and flooded out this whole community, this whole neighborhood back there. What we were able to do is look at and say, well, hey, this isn't necessarily water that's going to flow anywhere. It's getting stuck in this pool and then basically runs out through the city storm sewer. Should we be looking at this as a different sort of flooding and kind of tying it in with this urban flooding and tying it in with this lateral flow that's leaving it? And by changing the modeling approach and adding more detail than what FEMA would typically put in there, we significantly reduced the map floodplain. And it's not our goal to necessarily say that we're coming in here to reduce the floodplain to try to get you guys to not have to purchase flood insurance. Our goal as the engineer is, hey, maybe we need to have a better understanding of what the actual flood risk looks like. And there are people who are being inadvertently put into the floodplain that shouldn't be, and they're paying for flood insurance they don't need. It's impacting the cost of their property values. But also on the flip side, there were houses that we mapped into the floodplain that previously weren't. And for them, it's a burden now that they have to bear. But on the same side, if they weren't aware of this flooding impact and they got flooded, then they wouldn't have had the impetus to buy the flood insurance and they wouldn't be covered in that effect. Your homeowners insurance doesn't cover flooding. So why nobody wants to be mapped into the floodplain. It is a good thing sometimes just to understand your risk and really understand what's happening there. So yeah, there are some really good examples of when you should go in and update those. And ultimately, when I tell people is that you live in a community. Especially if you're on the community administrator, the floodplain administrator, public works, you know your community, you know where it floods. You get phone calls every time there's two feet of water in an intersection or a stream is overtopping a culvert. You know that better than I ever will. You hire the engineers to come in and map this and everything else, but sharing that information and telling us what you've seen out there is the best data that we can get when it comes to building these models. So we make sure that what we're modeling matches what you've seen. And that calibration is so important. So the first thing I always tell people when they ask me, a community official asks me, hey, should we update this? It's like, well, what do you think? You live here. Does the flooding that you've seen living here for the last 30, 40, 50, 70, 80 years in some cases, does it match what you see on the floodplain maps? Or do you look at this and go, wow, that makes no sense? Because if it doesn't make sense, it might not be right.

Mollie (Host)
I think that's great advice. putting it back in their own hands. They know this. If they're seeing some of those red flags, then you can kind of go through those triggers like we were talking about earlier of the warning signs that you look for as somebody in the industry. And if all of those are kind of lining up, then it weighs higher in that cost versus new mapping. One other thing that might contribute to that conversation here is improvements in terrain data, computing power, modeling technology. How have those changed the ways that communities understand and manage the flood risk?

Jordan (Guest)
We're in the era now where people expect instant information. You can go online right now and in most communities across the US, you can download really accurate LiDAR data for the whole community that maps out all the terrain and really high resolution. You can now download HEC-RAS for free. It's from the US Army Corps of Engineers. Anybody can install it on their computer. And the laptops we have today and the computers we have there are infinitely more powerful than the ones that they were running in the 80s and even the 90s. Our options of how we look flood modeling have changed dramatically. Take a really small like side tangent on here, but I've described cross sections or I mentioned cross sections a few times in our conversation. And I think it maybe bears just a little bit of explaining. You know, we think of a river as this three-dimensional object. It's a, it's a channel, it's a low point that continues downstream and water flows across it. But in these early modeling days, even until recently, we couldn't looked at everything. We couldn't model that whole stream channel. So the methodology that the engineers 50, 60 years ago had come up with was, okay, so we're going to cut cross sections, which is to say you're going to take a line perpendicular to the direction of water that's flowing, and we're going to look at what that channel looks like from left to right across there. And it basically kind of looks like a large description, large "U" shape where you have the channel in the middle or dips down, water's flowing through there, floods up onto the overbanks and you have some water that spills out to the sides. And in the seventies, sixties, seventies, eighties, the only way that we could get that information to see what that cross section looked like was to send out a surveyor and have some poor surveyor go hiking up and down these floodplains Sometimes going through the river, having to drill through ice to get river bottom shots or try to do it off of a boat. And it was really time consuming, really expensive because it's all done by hand. There was no total station. There was no computer or anything. And so these old models that you look at, you might see cross sections that have, you know, 6, 10, 12 points for a cross section that's a half mile wide because that's all they could afford to go out and get. with LiDAR and these new terrain models that we have available to us for pennies on the dollar to what it used to take, we can go out and we can cut 10 times as many cross sections that have 300 points across them. So we have a much better representation of what that terrain looks like, filling in a lot of those gaps just because that information is so widely available to us now. And that's really something that's really become a thing in the last 15, 20 years. And at the same time that this better and better terrain data has become available, we've seen a big shift in also how modeling is done. Everything in the floodplain world was always done on this 1D model because that's all that our computers could handle. It was always an issue of what computation power do we have to work with. Well, as computers got stronger, got faster, we've been able to switch from this 1D model into a 2D model. And all 2D means is now instead of water always flowing straight downhill, it can go straight downhill, but it can also go left and right. It can flow out of the channel, it can flow into the overbanks, it can go places that the 1D model never could interpret it as going. We can include things like storage, we can include then the breakout flows, and it doesn't rely heavily on the engineer making sure that they're spending the time to model this and put all these little fine, fine details in there. The models will do it right away for you as long as you build the model correctly. And so how that's changed dramatically changes how we can look at floodplains, how we can map floodplains. But the issue that we're running into now, which is, I think, kind of the new ghost that we first starting to be haunted by, is that that's not how our regulations were written. Our regulatory practices were written for those 1D models, for these really simplified things. And that's just not how flooding really works. It's now we're flipping it and kind of looking at it the other way of like, now we have this much better picture that we need to kind of go back and try to shoehorn it into an older framework for regulatory purposes. And the modeling and the engineering pace of this now has outstripped the regulatory side's ability to keep up. And it's not like our regulators don't want to use better data, but they're held to federal law. It takes an actual act of Congress to change this. And so now we're dealing with that as kind of this new ghost of, you know, everything was written 60 years ago for what was pertinent and a really good assumption then, and nobody knew where this was going to go at that point in time. And so now trying to deal with that on the back end has definitely created a whole new issue that we're trying to work with. But the new modeling and the new tools that we have available to us do such a good job of really painting a better picture of what that flooding could look like that the engineers back in the 60s never could have imagined that we'd have available to us. And now it's not just a handful of people across the nation trying to do this. Every engineer in the nation has access to these tools and, you know, we're just getting better and better and faster and faster at doing it.

Related Content: 6 Keys to Long-Term Flood Resiliency

Mollie (Host)
As we've uncovered here, ghosts of the past, not so bad. They're okay. They're not so scary. Ghosts of the future with regulatory issues might be a problem, but that's, we'll see.

Jordan (Guest)
It's something scary, but I will say we've got some great people working on it. Our friends at the Association of State Floodplain Managers are really working day and night to try to work with Congress and to work with FEMA and the regulatory agencies to address these issues. And they've done remarkable work getting 2D models accepted by FEMA and progressing how we can look at this. And they're doing great work on it. I'm hopeful for the future that it's just going to continue to get better.

Mollie (Host)
Well, I think that's a perfect way for us to end our conversation here. Really, by understanding where those models came from and the assumptions that they're built on, communities can make smarter decisions about development, infrastructure, permitting, and long-term flood resilience. So sometimes the scariest thing isn't an old model. It's relying on one without understanding what's behind it. A big thank you to Jordan for helping us separate floodplain fact from floodplain folklore and for showing us how both historic and modern modeling have an important role to play in managing today's flood risk. And then thanks to everyone listening. We hope this episode helped make some of the ghosts of flood models past a little less mysterious. So until next time, I'm Mollie and this has been {Brief}ly Speaking.