After 20 hours of testimony, flood committee hears the science

The catastrophic July 4 flood that killed 117 people along the Guadalupe River resulted from an inherently unpredictable weather system that overwhelmed computer forecasting models. Still, scientists told state lawmakers Thursday that targeted investments in technology could dramatically improve future warnings.

About 12 inches of rain fell across the upper Guadalupe River Basin in just a few hours during the early morning of July 4, causing the river to surge more than 30 feet and rise 26 feet in 45 minutes near the hardest-hit areas.

“While the conditions were ripe for severe flooding, computer models struggled to pinpoint the exact location of heavy rain,” said John Honore, chief meteorologist with the Texas Division of Emergency Management. “The system was weak and unorganized, making it difficult for models to predict specific rainfall amounts or locations.”

Local officials testified earlier in legislative hearings that they had little advance warning beyond routine flood watches issued across broad areas of Central Texas. The National Weather Service didn’t issue a flash flood warning until about 1:15 a.m. on July 4, giving communities minimal time to respond before the river began its catastrophic rise around 3 a.m.

Honore described the complex conditions that preceded the disaster: a persistent onshore flow had created unusually humid conditions for weeks, while remnants of Tropical Storm Barry moved through Central Texas. The combination created an atmosphere primed for extreme rainfall, but pinpointing where it would occur proved nearly impossible.

Texas State Climatologist John Nielsen-Gammon told the joint legislative committee that his 2005 research showed it remains impossible to predict extreme rainfall locations at the county level days in advance, regardless of data quality.

“National Weather Service models were poor at providing 24-48 hour warnings for the July 4 event, partly because air causing extreme rainfall enters Texas from areas with scarce or low-quality observations,” Nielsen-Gammon said, citing the Gulf of Mexico and northern Mexico as examples.

He noted that a key weather indicator came from a balloon launch at Del Rio — the only upper-air monitoring station between El Paso and Brownsville, a gap of more than 700 miles along the Texas border.

Despite the July 4 flood’s severity, Nielsen-Gammon noted that several past floods in the Guadalupe River Basin rivaled or exceeded it in terms of rainfall, citing examples from 1869, 1915, 1932, 1978 and 1987. He provided sobering examples of extreme rainfall totals in the Hill Country, including 36 inches in 18 hours during a 1921 storm east of Austin and over 48 inches over two days in 1978 between Hunt and Medina.

The July 4 flood resulted from 10 to 15 inches of rain, while floods near Mason County and Leander were produced by about 20 inches. Nielsen-Gammon said Texas should expect a 35-inch rainstorm somewhere in the area on average every two decades.

However, Nielsen-Gammon said the July 4 flood was “particularly damaging in terms of loss of life due to the time it occurred and the difficulty for people to understand the evolving situation.”

He contrasted the disaster with the 1932 flood, when cabins housing campers were destroyed but no lives were lost because the flood happened during daylight hours. “Camp managers reported that campers could see what was going on and they could evacuate safely,” Nielsen-Gammon said.

The nighttime timing meant the situation rapidly changed from a general flash flood warning — where the appropriate action might be to stay indoors — to a severe flash flood requiring people to seek higher ground. Nielsen-Gammon noted that “weather warnings did not keep pace with that threat and with the change in action necessary.”

The weather experts, testifying nearly nine hours into the hearing at the Hill Country Youth Event Center, provided the committee’s first scientific analysis of the disaster. The joint committee had conducted nearly 20 hours of hearings across two days without hearing from meteorological or scientific experts before Thursday’s testimony.

After the weather panel concluded, Senate Committee Chairman Charles Perry acknowledged the delay, saying, “We could have had y’all the first one probably last week and saved a lot of trouble.” Perry noted there had been “a lot of confusion last week and rightfully so, we’re not weather people” about data collection and forecasting processes.

The scientists identified specific technological improvements that could dramatically enhance forecasting accuracy.

John Schrader of Texas Tech University said the Hill Country has about 36 gaps where high-quality weather stations should be installed. He recommended adding advanced radar technology closer to headwater regions and establishing a Texas Prediction Center to complement national weather models.

“We need very quick reporting intervals of five minutes or less for measurements,” Schrader said. “Minutes matter for forecasters.”

Carla Guthrie of the Texas Water Development Board noted that while her agency operates 118 weather stations statewide, most partner stations report data too slowly. Only 23% of stations report at five-minute intervals or faster, which she called “problematic during intense rain events where rapid data is crucial for decision-making.”

Nielsen-Gammon recommended considering additional observations along the Texas border, such as profilers measuring moisture, temperature and wind aloft, to improve short-term forecasts.

David Maidment, a University of Texas professor emeritus, presented data showing how real-time detection technology could have provided crucial early warning.

A gauge on the Guadalupe River at Comfort recorded a “velocity spike” at 8:10 a.m. on July 4, when water velocity increased from 2 to 14 feet per second in five minutes — about 90 minutes before the water level peaked.

“If such a gauge were at Camp Mystic, it could have detected the spike and triggered an alarm for evacuation,” Maidment said, referring to the camp where 15 people died.

He recommended installing radar streamflow sensors on 800 of Texas’s 25,000 bridges to create a statewide early warning network.

Phil Bedient, director of Rice University’s SPEED Center, analyzed radar data showing the flood’s rapid development. His team runs a real-time flood forecasting system for Houston that provides data every 10-15 minutes using radar-based rainfall data adjusted by rain gauges.

“Real-time flood warning systems based on radar and advanced technology are crucial for both urban and rural areas,” Bedient said. “The necessary tools already exist.”

Despite advancing technology, the experts agreed that implementation requires sustained commitment and investment.

“We are safer today than 20 or 40 years ago due to increased information, gauging and models, but the critical missing link is communication to those on the ground,” Bedient said.

Nick Fang of the University of Texas at Arlington emphasized that successful warning systems require sustained investment and cannot depend solely on automation.

“Human decision-making is essential,” Fang said. “Systems must be tailored to different zones” across Texas’s varied terrain and climate.

Fang stressed that systems require “upfront and recurring funding for infrastructure, models, computer systems, and ongoing operation, maintenance, and future research and development.”

The experts provided cost estimates for improving Texas’s forecasting and warning capabilities:

• Schrader estimated a comprehensive forecasting solution at about $20 million, including additional gauges, radars and a prediction center
• Nielsen-Gammon said his recommendations for additional border observations would cost less than $1 million
• Maidment recommended installing radar streamflow sensors on 800 bridges at $7 million for capital costs
• Bedient suggested 20 to 25 automated river gauges in the Guadalupe Basin at $10,000 to $20,000 each

Schrader noted that his proposed comprehensive solution would include filling 36 weather station gaps in the Hill Country and adding four high-resolution radars, plus establishing a Texas Prediction Center focused on the 0-48 hour forecast period.

The testimony came as lawmakers work to craft legislation addressing the disaster that destroyed multiple youth camps and riverside communities between Kerrville and New Braunfels during the early morning hours of July 4.

Lt. Gov. Dan Patrick, who leads the Senate committee, has promised to pass emergency legislation immediately and continue work “in the next session and the session after that.”

The joint committee plans additional hearings as lawmakers prepare flood warning and emergency response reforms for the current special legislative session.