An Engineer's Perspective on the Texas Floods

0:01 This is an animation of the weather radar

0:03 in central Texas starting at noon on July 3, 2025.

0:06 You can see there was torrential rain across the state

0:10 throughout the afternoon from remnants of Tropical Storm Barry.

0:14 But focus on this area northwest of San Antonio.

0:17 Around midnight on July 4, a severe storm gets stuck in this area

0:22 and just stays in place for several hours.

0:24 When you put it in context with the rest of the system,

0:28 it looks kind of insignificant,

0:29 but that little storm dropped enough rain to raise

0:32 the Guadalupe River higher than ever in recorded history,

0:35 at least in the upper part of the basin.

0:38 The water quickly rushed through summer camps, RV parks,

0:41 and rural communities in the middle of the night.

0:43 And the result was one of the deadliest

0:46 inland flooding events in the past 50 years.

0:49 I live not too far from some of the worst-hit areas,

0:52 and although my family wasn’t directly affected by the weather,

0:55 it’s been a tough situation for me to wrestle with, personally.

0:59 I spent the better part of my career as an engineer

1:02 thinking about flooding and designing projects to cope with it.

1:05 I’ve worked on and played in the Guadalupe River.

1:08 And I have kids who are getting close to summer camp age.

1:12 As a dad, it’s almost impossible to comprehend a tragedy like this.

1:16 As an engineer, I’ve dedicated a large part

1:19 of my professional career to understanding events exactly like it.

1:23 So, as I’ve ruminated about this flood over the past few months,

1:27 I’ve collected some thoughts that might be worth putting into the world.

1:30 Let’s take a look at this event through an engineering lens,

1:34 talk a little bit about how technical and regulatory

1:37 decisions play out in the aftermath of tragedy,

1:40 and see if any lessons become apparent.

1:42 I’m Grady, and this is Practical Engineering.

1:53 One of the fundamental problems we face in engineering,

1:56 and really life in general, is that we can’t predict the future.

2:00 That sounds like a ridiculous thing to say,

2:02 but out of that uncertainty comes the framework

2:05 for how we think about so many things.

2:07 Because, we have to make all kinds of decisions- many

2:11 of them with extremely high stakes- in the face of the unknown.

2:15 In civil engineering, a lot of the loads we account for come from the most

2:20 classically volatile and unpredictable aspect of the earth: the weather.

2:24 Wind, ice, snow, waves, and rain- you cannot look ahead 50 or 100 years

2:30 and know what forces a structure will be subjected to.

2:34 You just have to guess.

2:35 And that’s a pretty hard thing to do, especially because you tend to have

2:39 two opposing forces pushing your guess around.

2:42 On the one hand, caution dictates overestimating

2:46 forces to leave a wide margin of safety, but on the other hand,

2:50 costs and budget constraints tend to push the estimate the other way.

2:53 I can make this dam taller or this bridge higher,

2:57 but it’s going to cost me a lot more money, and maybe it’s not necessary.

3:01 So how do you draw the line?

3:03 The same way we try to predict the future in so many other parts of life:

3:07 we look to the past.

3:09 Surely past performance is an indicator of future results, right?

3:13 I know that’s a stock line, but what else do we have?

3:17 Over the years, we have gone to considerable lengths

3:20 to apply historical data to predictions of future floods.

3:23 Of course, this gets pretty complicated.

3:26 One of the resources widely used in the United

3:30 States for decades is Technical Paper 40, published in 1961.

3:34 It represents a monumental effort to compile

3:37 rainfall data across the contiguous United States,

3:40 find probability distributions that fit the data, and map the results.

3:45 It’s divided up by duration and recurrence interval,

3:48 so you get this big group of separate maps.

3:51 But what is a recurrence interval?

3:54 I’ve talked about the so-called 100-year flood in a few of my videos,

3:57 but it’s a concept so widely misunderstood that it’s worth explaining again,

4:02 especially because it’s so relevant to the Guadalupe River flood in July.

4:06 We can’t really use historical data to determine

4:09 when a flood might happen in the future,

4:12 but we can make an estimation about how probable one might be.

4:16 The bigger the flood, the lower the probability that it might occur.

4:21 So there’s a relationship between probability and magnitude.

4:24 In hydrology, we often express the probability

4:27 as a quote-unquote “return period,” which means, on average,

4:31 how many years you would expect to pass

4:34 before you see that magnitude equalled or exceeded again.

4:37 But that “on average” is doing some heavy lifting in the definition.

4:42 This terminology is debated endlessly

4:44 in the hydrologic community because saying something

4:46 like the 100-year flood has an underlying implication that storms are cyclical;

4:51 that somehow if a particular magnitude of storm was to occur,

4:55 we might have a period of security before it happened again, or the flipside:

4:59 that if a flood hadn’t occurred in some time, we might be more “due” for it.

5:05 And that’s just not how it works.

5:08 Floods are statistically independent events.

5:11 Every year, the atmosphere rolls the metaphorical dice

5:14 to see what the biggest one is going to be.

5:17 The odds of rolling a two or snake eyes in craps are 1 in 36,

5:22 but if you go 35 rolls without a snake eyes,

5:25 the odds of rolling it on the next one haven’t changed.

5:29 The dice don’t remember what happened before.

5:31 No one calls snake eyes the 36-roll throw because we

5:35 understand it’s possible to do it twice in a row,

5:38 and it’s possible to go a lot more than 36 rolls without getting one.

5:42 So why do we call it the 100-year flood?

5:45 Probably because the only good alternative is

5:47 the storm with a 1% annual exceedance probability.

5:50 Just doesn’t roll off the tongue.

5:53 But it is the technically correct definition: the 100-year rainfall is the depth

5:59 of precipitation (over a given duration) that has

6:02 a one percent probability of being equalled or exceeded in a given year.

6:07 It’s a tough concept to wrap your head around,

6:10 but it’s fundamental to engineering hydrology.

6:13 If you take a look at these maps,

6:15 you can see that the 100-year rainfall over a 24-hour duration in Kerr County,

6:19 Texas is around 9.5 inches (or about 240 millimeters).

6:23 But again, this is from 1961.

6:26 And it’s based entirely on historical data.

6:29 So there are decades of rainfall not included in this analysis,

6:33 not to mention limitations in the statistical

6:36 methodology and data processing methods of the time.

6:39 TP 40 wasn’t the only resource for precipitation frequency data in the US,

6:44 but it was probably the most widely used until Atlas 14 came along,

6:49 or is coming along (it’s still a work in progress).

6:52 NOAA has been working to update this information

6:55 with the entire historical record and more rigorous statistical methods.

6:59 For most of the US, this is easy to navigate online.

7:03 Just mark a spot on the map and you get this table

7:07 of values and confidence intervals for a range of durations and return periods.

7:11 And you can see that the 100-year,

7:14 24-hour precipitation in Kerr County is 11.5 inches (or nearly 300 millimeters).

7:19 That’s a pretty big jump from the 1961

7:23 estimate- an increase of about 20 percent.

7:26 What was the 100-year rainfall in 1961 is now

7:30 just the 50-year storm AND look at those confidence intervals!

7:35 8 to 16 inches.

7:37 I know this is kind of long-winded,

7:39 but the whole point I’m trying to make here is

7:41 the tremendous uncertainty we have when it comes to hydrology.

7:45 In some ways, this rainfall data is extremely rigorous,

7:48 and I couldn’t even begin to explain some

7:51 of the statistical methods used to develop it.

7:54 It serves a really important purpose in the world of engineering,

7:58 planning, and emergency management.

7:59 But in another sense, it’s almost meaningless.

8:02 And I can show you a few of the reasons

8:05 through the lens of the Guadalupe River Flood.

8:07 Here’s an hourly map of the rainfall that hit central Texas on July 4, 2025.

8:13 That yellow area is the watershed for the upper Guadalupe River.

8:17 When I loop through it again,

8:19 you can see that cell right there caused the majority

8:21 of the flooding you probably read about on the news.

8:24 It was there and gone in four hours.

8:27 More rain came in later that morning and the next few days,

8:30 but this was a classic flash flood:

8:33 A relatively short burst of heavy rainfall on a small, steep, rocky basin,

8:38 where most of it runs off into a river within minutes or hours.

8:42 Here’s the thing: hourly rainfall records weren’t very common until the 1940s.

8:47 I counted about 100 rain gauges used by Atlas

8:51 14 within a 50 mile radius of Hunt,

8:54 Texas, where most of the fatalities occurred.

8:57 None had hourly records before 1940,

8:59 and of the group that did collect hourly data,

9:02 only four had a record longer than 70 years.

9:06 That might seem like enough data to understand flooding in the area,

9:09 but let me show you why it’s not.

9:12 Here’s that loop of rainfall again.

9:14 What do you see on this map?

9:16 Because I’ll tell you what I see: enormous spatial variability.

9:19 If you were to pick four random pixels on this map,

9:23 how good a picture do you think it would give you of what really happened?

9:28 That’s essentially what we’re doing with rainfall frequency analysis.

9:31 Compared to modern data collection methods, like the radar rainfall I showed,

9:37 our historical records are extremely sparse,

9:39 especially for data that varies so significantly across space.

9:43 Imagine trying to recreate the Mona Lisa

9:46 from scratch with just a dozen random pixels.

9:50 Most of the rain gauges we use to estimate flood probabilities have never

9:53 even seen an event of the magnitude we’re trying to use them to predict.

9:59 There’s a whole lot of extrapolation going on.

10:02 To hammer this point home: This is the 24-hour rainfall totals for the flood,

10:08 and you can see that even within this single watershed,

10:11 some areas saw extreme precipitation,

10:13 while others just got an inch or 25 millimeters of rain.

10:17 And actually, I mapped the percentage of the 100-year

10:20 rainfall that this storm amounted to, and you can see,

10:24 at least in the Upper Guadalupe Basin,

10:26 only a small area got close to the 100-year rainfall.

10:30 For most of the watershed, this was more like a 2- or a 5-year storm.

10:34 And here’s what makes this even tougher: When we’re talking about flooding,

10:38 we don’t actually care too much about rainfall.

10:41 We care about the outcome of rainfall,

10:44 specifically the rise in a river or stream.

10:47 Here’s the graph of a stream gage upstream of Hunt during the flood.

10:51 You can see that, starting around 2:00 on the morning of the 4th,

10:54 the river rose by 20 feet or 6 meters in three-and-a-half hours.

10:59 A little further downstream, similar story.

11:01 Starting at 2 AM, the river went up 35 feet

11:05 or nearly 11 meters in 3 hours before the gage broke.

11:10 That is a staggeringly fast increase.

11:12 In a hydrologic sense, it’s practically a wall of water.

11:16 And the results were devastating.

11:18 In Kerr County, there just wasn’t enough time to coordinate an evacuation.

11:23 More than 100 people were killed, many of them children.

11:26 So a rain gauge here, or here,

11:29 or here would have completely missed the fact that the watershed

11:33 it was within was experiencing the flood of record.

11:37 That’s the value of measuring the thing you actually care about.

11:41 Just like precipitation, you can take historical stream gage data,

11:44 fit it to a probability distribution,

11:46 and get a sense of the likelihood of major floods in the future.

11:50 But these gages are even more sparse in coverage than rain gauges,

11:54 their records often don’t go back as far,

11:57 they’re a lot more expensive to install and maintain,

11:59 and, as we saw in one graph, they can go offline,

12:02 ironically as a result of flooding, completely missing the peak.

12:09 Engineers or hydrologists actually often visit the affected area

12:12 and map the high water line after a flood

12:15 to validate and confirm the data from stream gages

12:18 (or to fill in the gaps if one breaks).

12:21 So, although they serve an extremely important role,

12:24 most of the time when engineers are trying to predict

12:27 flooding or its effect on infrastructure and the built world,

12:31 instead of using stream gages,

12:33 they’re using hydrologic models to convert rainfall into runoff and flooding,

12:38 a process that introduces a whole new set of uncertainties into the mix.

12:43 And there’s one more thing.

12:45 Everything we’ve been talking about so far

12:47 is predicated on a crucial underlying assumption:

12:50 temporal stationarity, basically,

12:51 the idea that the distribution of extreme events doesn’t change over time-

12:57 or put another way- that future

13:00 precipitation can be represented by past observations.

13:03 But, even though those past observations are

13:07 relatively sparse, in a lot of cases,

13:09 we can already see that it’s probably not a great assumption.

13:12 I understand this is a point

13:14 of pretty strong contention in the public discourse.

13:17 But within the professional community of hydrologists,

13:19 engineers, and climate scientists, it’s not really a question of “is the climate

13:24 changing” but more a question of how much,

13:26 how quickly, and where the effects of that are most pronounced.

13:30 For example, in the Texas Volume of Atlas 14,

13:33 the team tested for long-term trends in the data.

13:37 They found some scattered weather stations that did

13:39 show an increase in extreme rainfall over time; most of them didn’t.

13:44 Other studies have found more pronounced increases

13:46 by looking at only the past few decades.

13:49 So there are no broad statements that capture

13:53 the complexity of the situation as we understand it,

13:56 and importantly, this is a tough thing to figure out.

13:59 Say you have 100 years of historical data.

14:02 How many 100-year floods happened within that time?

14:05 Could be a few.

14:06 Could be none.

14:07 So, especially for very extreme events on the 1-in-a-century scale,

14:11 there’s a lot of uncertainty when it comes to teasing out any trends.

14:15 That said, there is a strong consensus

14:18 among the various climate models and recorded data

14:21 that a warming atmosphere has already resulted in an overall

14:25 increase in the intensity and frequency of rainfall,

14:28 a trend that will likely continue.

14:30 And you can see why that poses a problem.

14:33 Particularly for infrastructure with a design life of 50 to 100 years,

14:36 we need to design not just for the storms

14:39 of today but those decades in the future,

14:41 and our current methods of doing that is, on average,

14:45 systematically underestimating them if we assume a stationary climate.

14:50 Just to be clear, I’m not trying to blame a flood on climate change.

14:54 Although attribution studies can estimate the contribution

14:57 of extra energy in the climate system, there’s no way to ascribe any

15:02 particular weather event to global warming deterministically.

15:04 For many places, it might not even be

15:07 a major source of uncertainty compared to all

15:09 the other factors I’ve mentioned when it

15:12 comes to predicting the magnitude of future floods.

15:14 My point is that it’s just one

15:17 more confounding aspect of estimating flood risks.

15:20 And it gets to the heart of the entire issue.

15:23 Because why does any of this even matter?

15:25 There‘s been a lot of discourse about what should have happened

15:29 before the storm and what should be done in its wake.

15:32 But before you can take any action to mitigate flood impacts,

15:36 you have to know what the actual risks are.

15:38 On the upper Guadalupe, we’ve seen it with our eyes,

15:41 but how many similar watersheds just got lucky that night, or really, any night?

15:46 I think you’ll agree with me that this is complicated stuff.

15:51 And humans are notoriously bad at using

15:53 probabilities and risks to make decisions.

15:56 Almost nothing in our biology is optimized for long-term,

16:00 rational decision-making about rare and extreme events.

16:04 Almost every day of everyone’s lives, there’s not a flood.

16:09 That makes it really tough to consider it

16:12 as a priority and devote resources toward preparations.

16:14 And I think part of the problem is that we rarely talk about the uncertainties.

16:20 Even within the field of engineering, where we should know better,

16:23 we have a strong tendency to treat everything deterministically.

16:26 It sure makes things a lot simpler.

16:29 Take the bold number in the table,

16:31 plug it into your equations and computer models,

16:34 and just forget that those uncertainty bands even exist.

16:37 In some ways, it makes sense.

16:39 Ultimately, you do have to choose a number:

16:41 how high to build a bridge or how large a culvert to install,

16:45 or how wide to make a spillway.

16:48 But, in a lot of cases, those decisions get translated into a sort

16:52 of confidence that doesn’t actually exist.

16:54 The concept of the floodplain is a perfect example.

17:00 In the US, a lot of the framework for how we think

17:03 about and prepare for floods comes out of the National Flood Insurance Program.

17:08 And to participate in this program,

17:10 communities are required to regulate what happens in the floodplain,

17:14 or more specifically, what and how things get built there.

17:17 And so, a fundamental part of regulating the floodplain

17:20 is deciding where it actually is and isn’t.

17:23 We’re not going to dive into that process,

17:26 but billions of dollars have been invested in making

17:28 these maps and keeping them up to date in the US.

17:31 If you take a look at one, it’s a lot to parse depending on the location.

17:35 There are quite a few different hazard areas with different meanings.

17:38 The simplest for riverine locations is the base flood,

17:42 essentially the 100-year flood.

17:43 Some maps show the 500-year flood as well.

17:47 Many maps show the floodway,

17:48 which is kind of the main part of the channel needed to pass floods,

17:52 so it’s usually regulated more strictly.

17:54 But there’s something I notice when I look at floodplain maps.

17:57 All of these zones are bordered with nice crisp lines.

18:02 You’re inside the floodplain here, and you’re outside of it here.

18:05 And property owners often go to great lengths to refine these maps;

18:09 to shift the line just slightly and reduce their regulatory responsibilities.

18:13 But consider everything we’ve talked about

18:16 with estimating flood risk and ask yourself,

18:19 what’s the difference in the risk profile between here and here?

18:24 Is it enough to have a sharp line between them?

18:27 And if not- if the true situation is more nebulous- is

18:31 the map doing a good job of communicating flood risk to the public?

18:35 Because, just to be clear,

18:37 that is one of the stated purposes of floodplain maps.

18:40 Of course you need to delineate zones clearly to be able to regulate

18:45 where permits are required and where buildings can be built and so on.

18:49 But, to me at least,

18:50 it sends a complicated message to have this binary definition of inside

18:54 the floodplain or outside of it as a way to explain to individuals,

19:00 homeowners, renters,

19:00 and the general public about the risks that they’re actually exposed to.

19:06 You look at these maps and there is absolutely no indication about uncertainty,

19:10 despite the fact that almost every step of the process

19:14 that goes into creating them has huge margins of error.

19:17 And then, when we get more historical data, or land uses change,

19:21 or our understanding of the floodplain evolves,

19:24 and we try to change the map, that immediately sows distrust.

19:28 You hear it all the time (at least if you run in similar circles as I do):

19:33 “We’ve had two hundred-year floods in the past 5 years.

19:36 These engineers don’t know what they’re talking about…” Part of that, of course,

19:40 is just a misunderstanding about what the hundred-year flood actually means,

19:43 but part of it is that we don’t

19:46 do a good job communicating risk and uncertainty well.

19:49 The meteorologists get the same thing.

19:51 People get salty when forecasts are wrong without any acknowledgement

19:55 at all that the job is essentially predicting the future.

19:58 You know, it’s wizard stuff.

20:00 Weather is really complicated, and I think we have a lot of room to grow

20:05 in how we discuss and disseminate the things we don’t know for sure.

20:10 Because flooding is capricious.

20:11 If you look back at the maps from July 4,

20:14 you can see a lot of places where rainfall was

20:17 more intense than in Kerr County and the Guadalupe River.

20:20 Many areas of central Texas received more than the 100-year,

20:24 24-hour precipitation from Atlas 14.

20:26 And there were severe storms and flooding across

20:29 the region in the days that followed as well.

20:31 But nearly all the fatalities happened in this one place.

20:34 I don’t have a good answer for why.

20:37 Maybe some combination of timing, warning systems, the rural location,

20:41 differences in floodplain regulations, and plain bad luck.

20:45 I think scientists, engineers, and emergency planners can probably learn a lot

20:49 by simply comparing the flooding between Kerr

20:51 County and some of the other areas in central Texas hit by this storm system,

20:56 and why the outcomes were so drastically different.

20:59 My heart goes out to the victims

21:01 and their families who were affected by this flood.

21:04 I’ve been thinking so much about it in the weeks since,

21:06 and why these kinds of risks can go

21:09 so underappreciated that we wouldn’t bat an eye

21:11 at having such a large population of people

21:14 sleeping in the floodplain of a flashy watershed.

21:18 I think there are a lot of lessons to learn here,

21:20 but the one that keeps coming back to me is about communication.

21:23 People can’t act to reduce their risk

21:26 unless they can internalize what it actually is.

21:29 Professionals think about these issues every day;

21:32 they have technical training, knowledge,

21:34 and experience to make informed decisions about infrastructure,

21:37 land use, and zoning.

21:39 But most people don’t have the same cognizance of the hazards.

21:43 You can’t blame them.

21:44 It’s a crazy world we live in, and even individuals who live, work,

21:48 and play in areas at risk of flooding might

21:51 not come face-to-face with the danger in their entire lives.

21:54 Like I said, weather is complicated,

21:56 and we don’t all have the headspace to try and understand spatial variability,

22:02 annual exceedance probabilities, climate stationarity, and so on.

22:07 So I think the professional community has a responsibility

22:10 to improve how we communicate flood risks to the public,

22:15 not only for accessibility but honesty.

22:17 We need to have language that anyone can grasp,

22:20 but we also need to be better about acknowledging uncertainty.

22:23 It sounds counterintuitive,

22:25 but I think facing the limitations of our understanding head-on actually

22:30 instills more trust than pretending like we have all the answers.

22:35 And when people understand those uncertainties,

22:37 they get a deeper appreciation for how flood hazards vary across the landscape,

22:42 giving them more insight, not less, to prepare for what’s ahead.

22:47 Thanks for watching, and let me know what you think.