Weather Whiplash in Texas: Drought to flood

‘Weather whiplash’ is the abrupt transition from one extreme environmental condition to another that amplifies the effects of the individual extreme events. Such transitions include hot-to-cold and drought-to-flood.

Weather whiplash of extraordinary proportions affected central Texas in early July 2025. An exceptional and prolonged drought that led to reduced agricultural yields, livestock losses, and low water availability was abruptly followed by a 1 in 1000-year precipitation event. The result was catastrophic flooding, significant loss of life, destroyed infrastructure, and an altered natural landscape.

Here we document the extraordinary transition from drought to flood in central Texas in early July 2025 and analyze the meteorological and hydrological events to help provide improved understanding of such weather whiplash extremes and advance early warning and planning guidance.

Exceptional Prolonged Drought

As of July 1, 2025, 63% of the 20 central Texas counties included in a flood disaster declaration announced by Texas Gov. Greg Abbott in early July 2025 (Figure 1) were experiencing moderate, severe, or extreme drought conditions (Figure 2, top panel). Specifically, the southernmost counties under the declaration—Bandera, Kerr, Kendall, Gillespie, and Bexar—were in exceptional drought. Severe drought affected Travis, Caldwell, Guadalupe, and Comal counties, while areas west of Austin under the flood disaster declaration experienced moderate drought and abnormal dryness.

The 20 Texas counties included in the early July 2025 flood disaster declaration had been experiencing drought since late 2021, the second longest such uninterrupted period of drought in the U.S. Drought Monitor (Figure 2, bottom panel). This included 180 weeks of at least severe drought and 105 weeks with exceptional drought somewhere in these 20 counties. Despite the persistence of drought in this region from 2021 and 2025 the intensity of drought waxed and waned locally due to periods of above- or below-average precipitation, which is common during drought.

Figure 1. Map of Texas with county boundaries outlined and the 20 central Texas counties included in the initial Texas flood disaster declaration highlighted and labeled. Credit: ESRI/State of Texas/NOAA/PSL

Figure 2: (top) U.S. Drought Monitor on July 1, 2025 and (bottom) area covered by drought categories between January 4, 2000 and July 1, 2025 from the U.S. Drought Monitor over the 20 counties included in Texas’ disaster declaration: Bandera, Bexar, Burnet, Caldwell, Coke, Comal, Concho, Gillespie, Guadalupe, Kerr, Kendall, Kimble, Llano, Mason, McCulloch, Menard, San Saba, Tom Green, Travis, and Williamson. The image was produced by drought.gov.

The effects of prolonged drought in central Texas were acute and numerous. Ranching operations since 2022 have been affected by a lack of water and forage, which led to poor cattle health and sales. Water resources have been severely compromised as reservoir and aquifer storage fell to critical levels. The Edwards Aquifer dropped below the Stage 5 drought threshold, the most extreme threshold, for the first time since the Edwards Aquifer Authority’s inception in 1993 (Figure 3). Also, Canyon Lake, a reservoir on the Guadalupe River, was at 46% of total capacity and Medina Lake was just above 2% full prior to the extreme precipitation beginning on July 3.

Figure 3: Edwards Aquifer levels (feet above sea level) at the J-17 monitoring well in San Antonio from December 1, 2021 through July 14, 2025. Drought stages (colored lines) coincide with Edwards Aquifer Authority drought thresholds. Data source: Edwards Aquifer Authority.

Extreme low precipitation and record high temperatures for the 48 months ending on June 30, 2025 were principal drivers of this drought. Precipitation was in the lowest 10-percentile and the 9th-lowest on record compared to 1899-2025 (Figure 4) while temperatures were the highest on record, almost 1 degree Fahrenheit greater than the prior record set the year earlier (Figure 5). While 48-month precipitation ending on June 30, 2025 was exceptionally low, it was not unprecedented, as several other such lengthy periods of low precipitation have affected central Texas, including the 1910s, late 1950s, and the mid-2010s.

Figure 4: 48-month cumulative precipitation (inches) ending on June 30 of each year during 1899-2025 over the 20 counties that were included in Texas’ disaster declaration shown in Figure 1. Data Source: NOAA National Centers for Environmental Information nClimGrid.

Figure 5: 48-month daily maximum temperature (Fahrenheit) ending on June 30 of each year during 1899-2025 over the 20 counties that were included in Texas’ disaster declaration shown in Figure 1. Data Source: NOAA National Centers for Environmental Information nClimGrid.

Extreme Precipitation

Starting on July 3, remnant moisture from ex-Tropical Storm Barry, which hit Mexico a few days prior, combined with a mesoscale convective vortex (a small area of low pressure that develops from preexisting thunderstorm complexes) to produce extremely heavy rainfall over central Texas. NOAA’s Weather Prediction Center’s Mesoscale Precipitation Discussions issued 3 - 4 July 2025 describe additional small scale details about the processes that ultimately led to the intense and sustained local rainfall rates.

Figure 6: Infrared satellite image showing thunderstorm clusters and mesoscale convective vortex (red symbol) which was partially responsible for the heavy rainfall over central Texas as documented in NOAA’s Weather Prediction Center’s Mesoscale Precipitation Discussion issued at 12am CDT 4 July 2025.

Measuring extreme precipitation is a complex science and PSL scientists are evaluating a comprehensive suite of gauge-, radar-, and satellite-based measurements to help determine just how much precipitation fell for this event.

Localized totals likely exceeded 20 inches of rain over the multi-day period, with 10-12 inches of rain falling in just six hours over some locations. An example subset of datasets analyzed are shown in Figure 7, which highlights the differences in how products using various data sources, spatio-temporal resolutions, and data-blending algorithms represent the precipitation. The underlying hilly terrain of the region can make precipitation estimates less accurate. Small-scale interactions between the hill landscape and the atmosphere can produce areas of enhanced or decreased precipitation at scales that are smaller than the rainfall datasets can resolve (Bytheway et al, 2020).

Figure 7. Several maps show the maximum 24 hour rainfall measurements over South-central Texas between July 3 and July 5, 2025. Kerr County is outlined in bold.

On-the-ground (rain gauge-based) measurements of precipitation provide immediate information about how much precipitation falls at a point. Comparing that information against authoritative historical records can provide localized insight into how rare an event was for a given location. In the stations plotted below, the ~10 inch rainfall total was by far the largest 24-h event in the 70-year-long record captured from seven stations in the region.

Figure 8. Daily rain gauge data from San Angelo, Texas for 2025 (dots) overlain on historical daily rainfall from 1953-2024. Data comes from NOAA’s Applied Climate Information System. Two stations reported data for 2025 and seven stations reported data for some portion of time between 1953 and 2025. Image: Alex Thompson, PSL/CIRES

Our partners at NOAA’s Office of Water Prediction combine gauge analysis with radar observations and compare it to NOAA Atlas 14, a tool that describes how rare precipitation events are by “return periods.” This event was found to exceed the “1/1000 year” annual exceedance probability, or an event with a 0.1% probability over a given location each year.

Figure 9. National Weather Service Annual Exceedance Probabilities (AEPs) for the Highest 24-Hour Rainfall Period for Hill Country Texas from July 3-6, 2025. Image from NOAA NWS.

Deadly Floods

Heavy rainfall caused rapid increases in water levels, leading to flash flooding in some areas. For example, provisional data from the U. S. Geological Survey shows that on July 4th the Guadalupe River at Kerrville rose 32.5 feet in 1.5 hours (from 1.82 ft at 5:15am to 34.29 feet at 6:45am, CDT). Although the peak at Kerrville was below the 39 foot record, it was well above the National Weather Service (NWS) highest threshold category of “Major Flooding”.

Figure 10. The height of water (stage) in the Guadalupe River at Kerrville, TX and the volume of water (cubic feet) per second passing through this point from July 1 - 9, 2025 relative to NWS’s flood thresholds and record. The location of this gage is shown in Figure 11 in a red box.

The NWS has four flood categories — Action, Minor, Moderate, and Major — that are associated with increasing flooding thresholds. During the July 2025 flood event, 59 gages exceeded the Major flooding threshold and 3 additional gages exceeded Moderate flooding thresholds*.

* Gage sites without NWS flood thresholds are not included.

Figure 11: The height of water in the river at USGS gage locations relative to NWS flood thresholds during the July event. Counties where Texas made a flood disaster declaration are in bold (Figure 1, top panel). The red box highlights the Guadalupe River at Kerrville gage location shown in Figure 10.

Using the daily USGS stream gage records of discharge and gage heights, peak streamflow records from July 2025 were compared with previous annual maximas. While some record streamflows were observed - for instance, 3 USGS gages north of Austin, TX experienced their highest streamflow records ever - most were within historical ranges. On the Guadalupe River, two stream gages experienced their third highest flow on record.

Extreme Impacts

• Prolonged drought caused several years of decreased crop yields, livestock losses, and low water availability, which collectively resulted in substantial economic losses (NOAA Billion Dollar Disasters 2024).
• At least 135 people perished due to the floods, with several still missing, making it one of the deadliest inland flooding events in United States history. (Texas Public Radio)
• Economic damages estimated to exceed 18 billion dollars as floodwaters washed away homes, cars, and belongings. (Accuweather Press Release)
• Infrastructure damage as floodwaters washed away roads and bridges (KUT News Austin).
• Mudslides and debris flow altered the natural landscape (USA Today).

Outstanding Research Questions

Based on this drought-to-extreme precipitation analysis, there are several open questions that PSL, NIDIS, and the research community at-large can investigate to apply knowledge gained from this and similar events to improve understanding of water-related extremes and their impacts.

1. Is it more likely that extreme precipitation events can alleviate long-term drought in some areas more than others, and can that be predicted?

We already know that several prolonged droughts in central Texas were ended by extreme precipitation events. Examples include multi-year droughts in the 1910s, 1950s, and 2010s. However, further research is needed to determine if these mitigating precipitation events are predictable in Texas as well as other areas of the United States.

Better prediction of how a certain area’s proclivity to having a drought ended by extreme precipitation could help water managers and decision makers better anticipate the cessation of drought and capitalize on the opportunity to capture water from the extreme precipitation event for future use.

2. Does long-term drought affect the land surface in ways that influence the frequency and magnitude of devastating floods?

The ability of water to infiltrate the land surface is reduced when certain soil types are exceptionally dry. Research is needed to inform the complex relationship between drought, soil type, soil water infiltration, topography, runoff, and flooding related to extreme precipitation.

Knowledge and understanding of how long-term drought could affect the land surface and the characteristics of floods could provide forecasters and local decision-makers with enhanced situational awareness of where and when devastating floods in the midst of drought are more likely to occur.

3. How can we improve our understanding of a precipitation event of this magnitude, its rarity, and its historical frequency?

This specific event in central Texas has underscored persistent gaps and challenges in observational and model-based datasets to represent and forecast extreme precipitation. NOAA’s Modernizing Probable Maximum Precipitation (PMP) project is referencing a recent National Academies of Science, Engineering, and Medicine study that lays out an approach for NOAA Research to modernize authoritative extreme precipitation guidance. The PMP initiative is an important step towards filling these persistent gaps and solving these challenges related to extreme precipitation.

Regarding the extreme precipitation itself, it is an outstanding research challenge to understand the actual rarity of this event relative to historical frequency and to modernize authoritative guidance to reflect more updated datasets, methods, and environmental trends.

The PMP project is actively seeking novel methods for integrating historical events with recent observations. For example, the PMP team led by PSL is digitizing and incorporating into user tools analog storms to this event such as the 1932 flood detailed in the historical document below, courtesy of NOAA collaboration with the US Army Corps of Engineers:

Improved understanding of the historical frequency and overall guidance related to extreme precipitation events in a specific area will help engineers improve the design and build of high-risk infrastructure (e.g., dams) to reduce the chance of a catastrophic failure, and give key decision makers updated precipitation information with which to make critical decisions when facing an extreme precipitation event.

4. How well do NOAA forecast models and reanalysis datasets (datasets that look back and analyze represent past weather) extremes of this magnitude?

Similar to factors complicating measuring extreme precipitation and representing it appropriately in observational datasets, forecast models are also acutely challenged to provide accurate and reliable forecast guidance for rare events. The highly nonlinear nature of physical processes that interact on multiple space and time scales to actually form extreme precipitation makes these events both infrequent and inherently less predictable, even in high-resolution, state-of-the-science modeling systems.

Research recently published by PSL as part of leading NOAA’s Modernizing PMP study evaluates NOAA forecasts of extreme precipitation, specifically focusing on the models that were highly relevant to the Texas flooding event. Prior to the recent publication of these findings, despite NOAA’s High-Resolution Ensemble Forecast (HREF) system being operational for more than four years, little evaluation had been done to quantify how well the ensemble forecast system predicts extreme precipitation. PSL researchers, in partnership with NOAA’s Weather Prediction Center and Global Systems Laboratory, thus constructed a climatology of heavy and extreme precipitation forecast performance.

The in-depth analysis of forecasts over a comprehensive set of metrics, regions, seasons, and phenomena types provides critical contextual information for forecasters trying to understand how confident and reliable the ensemble’s predictions are for extreme precipitation. The results also provide valuable guidance for researchers seeking the most effective areas to target for improving high-resolution, ensemble-based, quantitative precipitation forecast guidance.

Particularly relevant findings for the Texas flood include Stovern et al. (2025)’s evaluation of techniques used in weather forecasting to generate a more realistic ensemble mean from a set of forecasts, particularly for precipitation. These diagnostics appear to have offered considerable skill in the Texas flooding event, for example, the “localized probability matched mean”, below. Stovern et al. (2025)’s findings suggest that forecasters can be confident that enhanced skill can be found in this diagnostic.

Figure 12. A forecast of 24-h precipitation amounts based on the “localized probability matched mean” diagnostic derived from NOAA’s High-Resolution Rapid Refresh modeling system. Image from NOAA/NWS Storm Prediction Center.

Additionally, Bytheway et al. 2025 provides critical information regarding individual ensemble member to member skill and variability, which forecasters noted as being a challenging aspect of the forecast in this event. Knowing which ensemble members are more reliable or biased when predicting extreme precipitation can help forecasters increase their confidence when making decisions based on relatively low probability forecasts of extreme events. NOAA Research, led by PSL and GSL, continues to focus on how to make research analyses of forecast datasets more useful for forecasters faced with making predictions for rare events.

The more that is understood about how NOAA forecast models and reanalysis datasets are representing extreme events, the better these models and datasets can become, improving the overall ability for NOAA and the weather community using these tools to predict these events.

5. How do uncertain precipitation forecasts affect streamflow forecasts?

Uncertainty in meteorological forecasts, particularly extremes, can create challenges for hydrological forecasting. Current ensemble streamflow forecast models use a statistically corrected meteorological model ensemble mean to generate the forcings (the set of atmospheric data used to drive hydrologic models) necessary to model streamflow. This technique allows the streamflow forecasts to be statistically robust, but dampens extremes, particularly when the meteorological forecasts include significant spatial uncertainties.

PSL is leading a research effort, in partnership with the Office of Water Prediction, to enhance the meteorological forcings that are used in NOAA’s streamflow forecasting models. A recent effort by Towler et al. (2025) shows how an alternative post-processing approach can improve the representation of heavy precipitation and subsequent streamflow. Ongoing work aims to understand how representation of extremes is impacted by use of more information from meteorological ensembles and through inclusion of machine learning techniques.

Streamflow forecast models that better represent extremes will empower more informed forecasts.

6. How can and should an event like this inform and guide improvements to extreme precipitation planning, early warning, and preparedness guidance going forward?

The recent Texas flood event emphasized that disasters occur at the intersection of environmental and societal realities. To help reduce the chance of future harm when natural hazards impact society, scientific research must be directly shaped by the real-world needs of communities, and improvements through research aimed at improving the ability of decision makers and community planners to build infrastructure where risk can be reduced. Recognizing the work to be done at this interface is a priority area for NOAA Research as exemplified by the partnership with the American Society of Civil Engineers.

PSL and its partners are focused on use-inspired research to improve extreme precipitation planning and preparedness as part of NOAA’s mission to protect lives and property. It is through this use-inspired research informed by the user communities that progress can be made in improving extreme precipitation science to serve societal planning and preparedness needs going forward.

Summary

The central Texas "weather whiplash" event in early July 2025, an abrupt shift from exceptional, prolonged drought to a 1-in-1000-year flood, resulted in catastrophic impacts including loss of life, destroyed infrastructure, and significant economic losses. This event underscores the critical need for continued research into water-related extremes to improve understanding, early warning, and planning guidance for future occurrences.

Key research questions emerging from this disaster focus on the understanding and predictability of extreme precipitation events ending droughts, how long-term drought influences flood magnitude, gaps in current guidance for extreme precipitation rarity and frequency, enhancing NOAA's forecast models for extreme events, and better translating precipitation uncertainties into streamflow forecasts.

Ultimately, use-inspired research, shaped by community needs and aimed at improving decision-making and infrastructure planning, is essential to mitigate future harm to society from such natural hazards. PSL continues to focus significant research efforts in this area to better understand, predict, and reduce the societal and economic impacts of such extreme events.

On this page

Exceptional prolonged drought

Extreme precipitation

Deadly floods

Extreme impacts

Outstanding research questions

Related links

NOAA Research: Decoding the warning signs of weather whiplash

Drought.gov: From Dust to Deluge: Weather Whiplash Devastates Texas

Download a summary causal map of PSL's analysis of the 2025 central Texas floods PDF (442 KB)

PSL authors

Andrew Hoell, Research Meteorologist

Kelly Mahoney, Research Meteorologist

Mimi Abel, Research Meteorologist

Janice Bytheway, Research Scientist

William "Ryan" Currier, Hydrologist

Alexander Thompson, Research Scientist

Erin Towler, Research Scientist

NIDIS authors

Jason Gerlich, Regional Drought Information Coordinator, Pacific Northwest DEWS and Missouri River Basin DEWS

Joel Lisonbee, Regional Drought Information Coordinator, Souther Plains DEWS