This is a difficult article to write. It’s heartbreaking to think about the families grieving loved ones and the people still unaccounted for. There’s a sense of helplessness that comes with covering a tragedy like this because sometimes, there truly are no words for such profound loss.
When devastating weather events like this happen, we’re often left asking: Why did this happen? How did it get so bad so quickly? I want to walk you through the meteorological setup behind this event, not to diminish the human toll, but to help explain how the atmosphere aligned to create such a dangerous situation.
The setup:
On the evening of Thursday, July 3, I was on-air discussing the growing flood threat across the Hill Country. Tropical moisture from the remnants of two systems, Barry and Flossie, had moved into Texas, fueling rounds of heavy rain.
The early forecast had called for 1″ to 3″ of rain, but when you’re dealing with tropical remnants, it’s not unusual for totals to double or even triple. That’s what makes these systems so dangerous; they can escalate quickly.
The graphic below shows the slight flood threat that was in place for the Hill Country from Thursday night into Friday morning. Unfortunately, what followed exceeded expectations.
“Flash Flood Alley”
This part of Texas is known as “Flash Flood Alley” for a reason. Flooding isn’t new here; it’s happened before with heartbreaking consequences. In 1987, ten teenagers tragically lost their lives when floodwaters rose rapidly near this very same area.
What makes this region so vulnerable is a combination of factors: it’s prone to extreme rainfall, the Guadalupe River runs right through it, and on a busy holiday weekend, the area was packed with people enjoying the outdoors.
Kerr County Judge Rob Kelly put it plainly: “This is the most dangerous river valley in the United States, and we deal with floods regularly.”
Why did the flooding happen?
You may have heard us talk about training thunderstorms. This is when storms repeatedly move over the same area, much like train cars passing over the same section of track. It’s one of the key ingredients for flash flooding.
That’s exactly what happened here.
Between 1:00 a.m. and 4:30 a.m., an estimated 10 inches of rain fell just west of Kerrville. All that water rushed into the Guadalupe River, funneling downstream with incredible force.
At midnight, the river level was 7 feet, moving at 8 cubic feet per second. Three hours later the river the massive runoff resulted in a river level of 29.45 feet, moving at 120,000 cubic feet per second!
This video shows what training storms look like. It is torrential between 1 and 4 am!
a
The answer to whether there was a warning ahead of this event is a resounding yes.
- At 3:35 p.m., a Flood Watch was issued, forecasting 1″ to 3″ of rain, with isolated areas possibly receiving 5″ to 7″. That was the first alert that conditions could become dangerous.
- Then, in the early morning hours 1:14 a.m. and again at 3:19 a.m. Flood Warnings were issued. These alerts specifically warned of life-threatening flash flooding.
- By 4:03 a.m., the situation had escalated. A Flash Flood Emergency was issued one of the most serious weather alerts we have. These are rare and only used when catastrophic, life-threatening flooding is underway or imminent.
But tragically, the timing couldn’t have been worse. These alerts came while most people were asleep. If phones were silenced or notifications were turned off, many may not have heard the warnings in time.
And that leads to another critical question: even if people got the alert did they know what to do?
These are the kinds of questions investigators and emergency officials will be looking into in the days ahead.
Did NOAA cuts have an effect on the warnings?
Staffing wasn’t the issue. In fact, the National Weather Service office in San Antonio has more meteorologists on staff than the one in Houston.
But when it comes to forecasting extreme weather, there are hard truths we have to confront; there is a limit to predictability.
As we got closer to the flooding, the high-resolution HRRR model began picking up on the potential for 10″ to 20″ of rain. But even then, the model didn’t place that extreme rainfall west of Kerrville, where the worst ultimately unfolded.
And even if it had, what then?
Would emergency managers have issued a preemptive evacuation in the middle of the night? That kind of action is almost unheard of and for understandable reasons. The statistical odds of a single location receiving 10+ inches of rain in just three hours are incredibly small. But when the risk is high, even low-probability outcomes can become devastating realities.
So we’re left with this question: How do meteorologists, emergency managers, and the National Weather Service communicate extreme but unlikely threats? How do we message worst-case scenarios when 98 times out of 100, they don’t happen?
Meteorologist Alan Gerard captured this dilemma perfectly in his Balanced Weather newsletter:
“The reality is that the state of the science does not allow us to know that a small location like the south fork of the Guadalupe River is going to get 8” of rain in 3 hours even a couple of hours ahead of time, let alone with the kind of lead time that would, for example, keep people from camping in that area on a given night.
The best we can do is to develop better models with better probabilistic information that can be communicated in a way that helps people make better, more proactive decisions.”
Next-level
We’ve made tremendous strides in forecasting over the last decade, and the potential to deliver earlier, more targeted warnings is real. That future is within reach.
But the recently passed NOAA budget cuts nearly all funding for the kind of research that would make these next-level forecasts possible.
That’s a conversation for another day, but it’s one we need to have.
If you’d like to watch my full video on this topic, click below.