After The Deluge: The Science Behind The Texas Floods
In this article we unpack what went wrong with modeling, outdated flood maps, and more during the tragic floods in Kerr County, Texas.
In early July 2025, the Texas Hill Country was hit by one of the most devastating flash floods in modern U.S. history. It was fueled by extreme rainfall and a confluence of meteorological and hydrological factors. The towering floodwaters descended on towns like Kerrville, Hunt, and Comfort, leaving over 100 dead and many more still missing. The event exposed the fragility of existing flood forecasting systems, the consequences of outdated infrastructure, and the need for climate-adapted modeling and planning.
As water researchers and communicators, we at the ByWater Institute believe this event offers vital lessons for all communities facing intensifying flood risks in a warming world.
Unlike hurricanes, which build over days and allow for more advanced warnings, the July 2025 flood was the result of a slow-moving storm system that stalled over the Hill country. The remnants of Tropical Storm Barry interacted with a frontal boundary, drawing a monumental amount of tropical moisture from the Gulf of Mexico into central Texas. Atmospheric moisture levels were measured above the 99th percentile for the region, fueling rainfall rates of 2 to 4 inches per hour. In less than a day, some areas received 10 to 14 inches of rain—amounts normally associated with once-in-500-year or even 1,000-year events.
Hydrologically, the Texas Hill Country is susceptible to flash floods due to the region’s shallow, rocky soils and steep terrain. Rainfall is converted to runoff at breakneck speeds.It was like someone pulling the plug on a dam upstream—within 45 minutes, the Guadalupe River surged 26 feet, transforming from a gentle current into a raging torrent that bulldozed through communities.
These are not rare conditions for the region. Texas researchers have long called it "Flash Flood Alley." But even within that context, this event was exceptional, not only in its meteorology but in the human cost.
While attribution studies are not yet complete, the storm bore all the hallmarks of a climate-amplified event. The Gulf of Mexico was undergoing a marine heatwave, its surface temperatures several degrees above normal. This heat added energy and moisture to the atmosphere, which in turn supercharged the rainfall. The Clausius-Clapeyron equation tells us that for every degree Celsius the atmosphere warms, it can hold about 7% more water vapor, which raises the ceiling for extreme precipitation.
Researchers have confirmed that the atmospheric water vapor content over Texas in early July 2025 reached all-time records for the month. These conditions were not isolated; they reflect a trend of increasing extreme rainfall across the U.S. South, one that has grown steadily over the past three decades.
The Models Were Right—But Not Right Enough
Forecasting flash floods is inherently difficult, but the science is steadily improving. In the days leading up to the event, the National Weather Service and its Weather Prediction Center issued alerts that warned of heavy rain and flash flooding. On July 3, flood watches predicted 5–7 inches of rain. On July 4, Flash Flood Warnings and eventually Flash Flood Emergencies were issued as radar data confirmed what was unfolding.
However, the scale of rainfall that occurred (up to three times the forecasted amounts in some areas) exceeded what many models predicted. Global AI-driven weather models largely missed the localized intensity of the storm. Ironically, older physics-based high-resolution models, like NOAA’s High-Resolution Rapid Refresh (HRRR), performed better. They captured more of the hyperlocal storm dynamics, suggesting that, in this case, traditional modeling still had the edge.
Even so, forecasts were only part of the problem. Kerr County lacked a unified flood warning system. There were no sirens. Cell service in rural areas was spotty. Many people never received the alerts. Some received them far too late.
The warnings, though imperfect, were there. But the infrastructure to respond to them was not. In the years leading up to the flood, Kerr County had proposed and even designed a comprehensive flood warning system, including high-water sensors, sirens, and real-time river gauges. Yet the project was shelved, first due to a failed FEMA grant application and later because of political resistance to spending federal relief money from the American Rescue Plan.
Instead of investing in critical alert systems, local officials redirected funds to park trails, employee bonuses, and other priorities. That decision had fatal consequences. The flood hit overnight, during the Fourth of July weekend, with little to no redundancy in emergency alerts.
This is a cautionary tale of how disaster risk is not only a function of natural hazards, but of policy decisions. It reveals the gaps between what we can do with science and what we choose to do with resources and governance.
Outdated Maps, Underestimated Risk
Compounding the issue were outdated FEMA floodplain maps and rainfall frequency estimates. Many flooded areas were not marked as high-risk zones. Residents built homes, businesses, and summer camps under the assumption that they were safe. These assumptions were based on those data that no longer reflect our climate reality, and these assumptions proved to be deadly.
With rainfall events intensifying, what was once a 100-year flood might now be more like a 20-year or even 10-year event. Texas’s first-ever State Flood Plan, released in 2024, called for updated maps and expanded coverage of flood-prone areas beyond those covered by FEMA. The 2025 flood showed why that update is urgent.
What must happen next is not just a technological upgrade, but a cultural and policy transformation. Communities must invest in real-time flood monitoring and alert infrastructure, including robust rain and river gauges, county-wide sirens, and community-specific alert systems with redundancy. Forecasting models need enhancement through greater funding and innovation, particularly for high-resolution tools and AI-enhanced systems like NOAA’s Warn-on-Forecast, WoFSCast, and HRRRCast.
At the same time, floodplain maps and hazard assessments must be modernized to reflect the realities of a warming climate, and both FEMA and local governments need to keep those data up to date. Public education also plays a critical role. This is where Texas can take a page out of Louisiana’s book. Flood literacy should be widespread, from annual school drills to neighborhood preparedness plans. Long-term resilience will also depend on integrating climate adaptation into urban and regional planning. That means not only building infrastructure to resist floods, but also rethinking where and how we develop, favoring green infrastructure, preserving floodplains, and reducing emissions to mitigate future risk. Science can provide the tools, but the readiness to use them lies with policy, funding, and the collective will of communities.
Regional and National Relevance
The lessons from Central Texas are not limited to the Lone Star State. Similar risks exist across the Southeast, Midwest, and Intermountain West, anywhere topography, climate, and infrastructure converge. Eastern Kentucky’s 2022 flood, which killed dozens, followed a similar pattern of heavy rain, steep terrain, and under-resourced communities.
This is a national issue. And one that will grow worse unless the U.S. invests seriously in flood resilience.
The 2025 Texas flood was a tragedy, but we can learn from it. Understanding what went wrong can lead to a blueprint for how to adapt science, planning, and community resilience to a new climate era. As water rises, so too must our ambition to meet the moment. That means elevating not just buildings, but expectations and accountability.
The ByWater Institute remains committed to supporting this effort, through research, education, and collaboration. We encourage policymakers, scientists, and communities to treat this flood not as an anomaly, but as a call to action.
The next storm is already forming. The question is whether we’ll be ready.
Sources:
Matt Simon (Grist/Yale Climate Connections) – “The science behind Texas’ catastrophic floods” (July 14, 2025)
Chelsea Harvey (E&E News/Scientific American) – “AI Couldn’t Forecast the Texas Floods” (July 15, 2025)
Eric Holthaus (The Guardian) – “Texas floods reveal limitations of disaster forecasting under climate crisis” (July 6, 2025)
Dr. Ali Fares (PVAMU) – “Science in the Wake of Disaster: The 2025 Hill Country Flood…” (July 9, 2025)
Texas Tribune (Terri Langford & Dan Keemahill) – “Why Kerr County balked on a new flood warning system” (July 10, 2025)
The Environmental Blog – “Texas Floods 2025: Death Toll, Causes & Preparedness Guide” (July 2025)