Is it time to rethink flash flood warnings?

I wish this post had an answer, but flooding is difficult

After the devastating South Fork fire in Ruidoso, New Mexico last year, the road to recovery has been difficult. The community is facing financial pressures, a slow response to fund access, as well as frequent flooding concerns. How frequent? Between July 1st of last year and the end of July this year, Ruidoso has been placed under a flash flood warning at least 56 times. That’s (on average) once per week. Obviously in New Mexico, it doesn’t rain every week. There were 15 issued in July 2024, 12 in August, 1 in September, 2 in May of 2025, 6 in June, and 20 last month. Ruidoso is unique in that they’re in a burn scar location, which becomes especially vulnerable to flash flooding.

In that same time, Kerrville, Texas has been under 19 flash flood warnings, Houston 12, Austin 13, and Dallas 22 flash flood warnings. I went through the same time period for various Texas NWS offices to see the verification rate of storm-based flash flood warnings. About 40 to 60 percent on average verify. A huge, huge caveat to that is that it’s likely a fair bit of flooding is never reported given the enormous size of Texas and large rural areas. Although the official verification rate is 40 to 60 percent, it is likely higher than that in reality. The New York City NWS office has had 80 percent of their flash flood warnings verify in the last 13 months, much easier to do in a more urban location. So, let’s say for the sake of argument that 80 percent of flash flood warnings verify. Not bad, right? When you’re in NYC and issuing 35 flash flood warnings in the last year-plus, certainly. When you’re in a place like Texas, where the rate of issuance is much higher, that would inevitably mean many false alarms. For example, the Austin office has issued 114 flash flood warnings in the last 13 months. That still leaves about 20 to 25 warnings that would be unverified if their hit rate was 80 percent. Dallas has issued 141 of them. Houston has issued a more charitable 35, likely due to a combination of slightly less heavy rain than usual in the last 13 months and perhaps a higher threshold for pulling the trigger here.

So why am I running through a bunch of statistics and numbers? 141 flash flood warnings in a region is a lot of flash flood warnings. A lot. 56 in one specific location in a year? A lot. Houston has seen 117 flash flood warnings in the eight years after Hurricane Harvey hit in 2017. It’s just…a lot. Let’s be honest.

What are flash flood warnings anyway?

The NWS defines a flash flood warning as follows:

A Flash Flood Warning is issued when a flash flood is imminent or occurring. If you are in a flood prone area move immediately to high ground. A flash flood is a sudden violent flood that can take from minutes to hours to develop. It is even possible to experience a flash flood in areas not immediately receiving rain.

That seems reasonable. In the fall of 2019, a transition to IBWs or impact-based warnings occurred within the National Weather Service. You now had 3 tiers of flash flood warnings to choose from. There was the basic flash flood warning as described above. But the forecaster issuing warnings also now had the ability to “tag” the warning as “considerable.” How is that defined?

Use rarely, when there are indications flash flooding capable of unusual severity or impact is imminent or ongoing and urgent action is needed to protect lives and property.

Then, there is the highest echelon of flash flood warnings: the flash flood emergency, a “catastrophic” tag, which is to be used “exceedingly rarely, when a flash flood threat to life and catastrophic damage is occurring or is imminent, and floodwaters have risen or will rise to levels rarely, if ever seen."

Notably, if a flash flood warning is tagged as considerable or catastrophic, it will trigger a wireless emergency alert (WEA) to your phone. A basic flash flood warning will not. However, in most cases, if you’re using an app to get weather alerts, any flash flood warning that’s issued will push an alert to your phone if you have that app’s notifications enabled.

Lest we forget the flood advisory. That is issued “when a specific weather event that is forecast to occur may become a nuisance. It is not expected to be bad enough to issue a warning. However, it may cause significant inconvenience, and if caution is not exercised, it could lead to situations that may threaten life and/or property.”

So your hierarchy of warnings…the alerts you should be taking action on is essentially:

Flood Advisory: Be aware, use caution.
Flash flood warning: Take action.
Considerable flash flood warning: Take action because this is worse than usual.
Catastrophic flash flood emergency: Take immediate action to protect lives and property.

That’s a lot to digest. And it’s kind of interesting. Through the end of July in 2025, the nation has had:

5,918 Flood Advisories
3,811 Flash Flood Warnings
591 Considerable Flash Flood Warnings (excluding additional statements or upgrades from basic flash flood warnings)
50 Catastrophic Flash Flood Warnings (excluding additional statements)

Something stands out to me here. Of the 3,811 warnings this year, over 15 percent have been labeled “considerable.” (Editor’s note: This was originally reported as 40 percent. I had accidentally left in duplicates, and this is now corrected. I apologize for the error). Recall, the definition for considerable tagging began with the phrase “use rarely.” If 15 percent of all flash flood warnings are for considerable damage, is that rare? Do 15 percent of all flash flood warnings truly rise to the occasion of needing to sound the alarm? Meanwhile, 1.3 percent (originally cited as 2.5%) of the warnings rise to the level of catastrophic. That, to me, seems more than reasonable.

Are we overwarning?

The numbers above break down somewhat how you’d expect. There are more advisories (nuisance flooding) than warnings than considerable warnings than catastrophic ones. But I want to get back to the 15 percent “considerable” rate. It’s almost as if we’ve tried simplifying the warning system only to complicate it more. That feels like a lot. Honestly, it feels today like the basic flash flood warning is almost akin to a flood advisory and you would probably wait til a considerable tag gets reached before getting concerned. Maybe that’s the point. But I guess there are a couple layers of this to peel back. This is what I’m thinking about. Edit: My original math had this as 40 percent, which is incorrect. Still, I do think 15% is a higher number than I’d expect for considerable events.

Do forecasters feel like they need to pull the considerable tag out more frequently than they actually need to because the phone alert will only get tripped if that happens? I’m going to speculate that if you asked a lot of forecasters they’d tell you that’s not really a consideration, but I don’t know for sure. No one wants to be the one that didn’t warn the bad thing that happens, right? And flooding is tricky because sometimes street flooding can be considerable. It isn’t just the awful stuff we see on the news like in Texas last month. It’s highways that flood in Detroit routinely when it rains. It’s streets in Houston. It’s terrain driven road flooding in West Virginia or Maryland. It’s near a wash or arroyo in the Southwest. Any of those instances can be life-threatening. But are they rare? I suspect it’s less rare than a lot of people realize. Again, the considerable tag was intended to be used “rarely.” So, what do we do?

Does the public even know or care about the difference? A flash flood warning is a flash flood warning. After a while, they may get sick of them, particularly if it’s just “the typical spots” flooding. Most people in Houston know that certain frontage roads flood routinely when it rains. Our local NWS office doesn’t push flash flood warnings in those instances usually because it’s common, it’s isolated, most people are aware or expect it, and usually a flood advisory will cover things. But if a flash flood warning is issued here, it tends to carry a little more weight. But if I asked 100 Houstonians if they knew the difference between a basic flash flood warning and a considerable one, I’m not sure many would. Would they know a flash flood emergency? Yes. But delineating between considerable and typical is a grayer area.

Further, does the fact that most other weather apps people get warnings from push alert for any flash flood warning dilute the message? I don’t know enough about every app to know if there is a clear demarcation indicated between a classic flash flood warning and a considerable one. But I do know that a lot of flash flood warnings get issued each year. Frankly, do we have too many alerts in general? We have the whole color spectrum of alerts now from public safety agencies. They’re all well-intentioned, but they’re often also abused. If I’m in Houston, I don’t need a blue/amber/silver alert for something occurring in Midland. Every local news media entity issues push alerts or declares Quadruple Alert Weather Days when there is the slightest hint of mischief. At a certain point, you get sick of it and say, “Wake me when it matters.” But if you’re constantly bombarded with this stuff, how do you know when it matters? And this is before any actual weather warnings get issued.

No easy answers

All of this to say: I don’t know if the current system is the right one when it comes to flooding. But I also don’t know that it’s the wrong one either.

There’s a clear difference between flooding that’s disruptive or annoying or injurious if you make a poor decision and one that is truly, definitely life-threatening. “Turn around don’t drown” is a phrase we use regularly to emphasize never driving through flooded roads because you certainly don’t know how deep that water is. And we lose numerous people annually to those decisions or unfortunate accidents that occur. But do we need to stop trying to verify on a specific metric and start strictly considering the situation? A street flooding mess at 3 AM is a lot different than at 3 PM. A paper published in 2021by Hatim Sharif of UTSA concluded that of the over 5,700 known flooding deaths in the United States from 1959 through 2019, roughly 86 percent occurred from people driving or walking into floodwaters. 58 percent of the total were people in cars or trucks. And over half of all flooding fatalities occur at night.

This gets complicated. Because flooding is indeed a dangerous threat that requires ample warning and notification is the saturation of warnings that exists today actually unnecessary? Are there ways to utilize flood advisories instead of flood warnings? Or since a lot of the deaths from driving into flooding occur due to people ignoring barricades or barriers, is the solution derived more from good partnerships between the NWS and local emergency management agencies? It sure seems like there were issues in Kerr County last month with how the emergency preparedness and response folks handled their duties. Not every municipality has the resources or manpower to manage things like that. That’s a topic for another time though.

In the meantime, I don’t want to sit here and just lob a bomb and walk away. I genuinely think that the topic of how we define and issue flash flood warnings needs to be revisited. Perhaps it turns out that the current method is the right one. Or the least worst one. But it sure seems like we’re at a bit of a crossroads with this problem, and usually that’s a good time to re-evaluate where we stand.