A Dangerous Case of Missing Water. (Part Ten)

-


It’s the evening of July 15th 2021 in the Valley of Ahr, Germany. Heavy rains are predicted and this is the fourth day of constant flood warnings issued by the German Meteorological Service. The Ahr River is equipped with a total of 8 supporting river gauges, four on the main tributary, and four on smaller feeders. The first gauge was installed in 1946, the newest in 1988. Based on modeling, a river rise of approx. 19 feet was forecast (ref 1). This would have kept water largely within the banks and town leadership did not evacuate. This decision would prove catastrophic.  The water arrived shortly after midnight. It was 34 feet, not 19 feet.  It was also too late to evacuate and 139 people lost their lives in this valley alone.

Why was the forecast off so much? Germany had just spent tens of millions on the European Flood Awareness System (EFAS). This river had 54 prior recorded flood incidents over centuries of settlement. 73 years of river gauge monitoring existed. The subject storm featured approximately 7 inches of rain. This was half of the 14-inch rain record set several years prior.  Why did the hydraulic model fail? 19 vs 34 feet is a 44% deadly underestimate. Why was the water missing from the forecast?

A clue may lie in what else is missing. During the July 4th floods in Kerr County, approximately 52% of the biological materials (trees, bushes, grass) was forcibly torn from the Guadalupe River areas. Normally, this presence is beneficial in slowing flood waters and reducing erosion, albeit at a slight increase in water level rise captured in Manning’s equation for surface roughness. Up to certain water levels, the obstructive value of this forest is fixed until water pressure increases to a critical level where destructive cascading begins. Instead of withstanding flood waters, trees are torn down and entrained by the flood waters causing large wood debris (LWD) fields and chaotic debris jamming. Uprooted trees are accelerated and slammed into downstream trees increasing the solid mass volume in the flood space. Finding a hydraulic model that specifically addresses the backwater effect of moving thousands of tons of obstructive mass through narrow canyons, over low head dams, and bridges becomes very complex, if measurable at all. Of note, evaluation of debris damming effects was specifically excluded from the July 24th, 2025 Guadalupe River Hydrological Assessment following the July 4th flood (ref 2). What else can we learn about it?

In 1997, Spain experienced deadly flooding in the Iberian Peninsula that struck at 230AM and resulted in the deaths of 20 people. This flood experienced a significant amount of wooded debris prompting the Geological Survey of Spain to conduct a study titled the “Two-dimensional modelling of large wood transport during flash floods” (ref 3).  The study can be reviewed below but the salient point is simple. LWD damming can significantly affect backwater levels and cause flood behavior to become unpredictable. In certain model scenarios, backwater levels increased from a “no wood” height of 4.7 meters to an obstructed 8.3 meter level. This is a 43% difference which is astoundingly similar to the 44% error difference between forecast levels and measured high water mark levels in the Valley of Ahr flooding.

The German flood experienced massive levels of LWD recruitment and transport. The chaotic forming and breaking of debris dams resulted in “Tsunami-like” flash waves creating sudden rises in downstream water levels. This same phenomenon was reported via witness reports at four separate locations beginning at Camp Mystic and ending 13 miles downriver at Howdy’s Bar and Grill on the Guadalupe River in Kerr County.

'What does this mean for Texas? First, this is highly problematic for predictive flood forecasting. Hydraulic modeling can likely predict “no wood debris floods” or perhaps with enough research; estimate debris damming effects but this doesn’t solve the problem. The exact location and water pressure where destructive tree cascading begins will likely remain unpredictable making accurate level forecasting difficult with errors likely to be deadly “underestimate” events. More worrisome, LWD damming results in unpredictable high-water events that completely undermine any validity derived from flood plains lines. 100-year, 500-year…it doesn’t matter. Water behavior becomes unpredictable and can rise quickly. Secondly, it may invalidate high water mark hydraulic projections further devaluing post event water reporting while overestimating volume. 

Interestingly, as LWD damming is reduced, the effect would likely reverse resulting in “missing water” from models utilizing high water mark measurements. Curiously, the 2025 Guadalupe River Hydrological Assessment identifies the model overestimating water surface downstream as measured by the river gauges at Spring Branch and Bergheim by a difference of 10-20 feet…more missing water.

It would seem the German, Swiss and Belgian authorities are catching on and again visited the subject conducting additional research regarding bridge clogging in 2024-2025 (ref 4). An interesting quote from the conclusion; “More research on the effect of debris shape and type on backwater is urgently needed!”  


So what are we studying here in Texas?  After receiving a grant for $4M in research funds, Rice University and UT Arlington have teamed up to study high resolution weather monitoring for better storm prediction and weather monitoring. Texas State is studying Mental Health and Volunteer Impact. The water authorities are buying more water gauges and interestingly, the San Antonio Botanical Garden TREES initiative plans to plant 50,000 native trees in the Guadalupe River Valley. 

Returning the Guadalupe River to its prior natural splendor is a worthwhile endeavor, but we must account for the lives lost and understand the increased risk generated by thousands of 100-foot cypress trees in the floodway. The July 4th storm was significant, but Texas has seen much worse storms. Top estimates measured 17 inches in 24 hours, less than half the rain record in Texas.

Risk on the Guadalupe remains high and will only increase. Our warning systems must match that risk.

Ref 1: https://nhess.copernicus.org/articles/25/581/2025/nhess-25-581-2025.pdf

Ref 2: https://d3ey0ivtc68uxj.cloudfront.net/19e93847-dd19-437e-8b0f-cb02dbb05f0b/b8df7a9ca7e38a2fc5551b4ecd89510e.pdf

 Ref 3: https://www.researchgate.net/publication/260076636_Two-Dimensional_Numerical_Modeling_of_Wood_Transport

Ref 4: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024WR039218

Comments

Post a Comment

Popular posts from this blog

River Sentry Project RIVER SHIELD GUADALUPE commences! (Part Nine)

-
Image
Over the holidays, River Sentry formally committed to installing 100 flood warning towers along the Guadalupe in Kerr County before campers return this summer. The project is privately funded by donors and an association of youth camps that is represented by Camp Mystic. River Shield will encompass both the North Fork, South Fork, Honey Creek, and Turtle Creek with the goal of providing localized flood warning protection for all youth camps in the area. After the installation is complete, River Sentry will also build a mobile app that will allow the public, including camp parents, to see the location and status of all sensors located along the rivers. This is expected to be launched later this year. “This is to build trust” River Sentry Founder and CEO Ian Cunningham explains. “We are building the most advanced and capable collection of flood warning systems that exists anywhere. We want people to see it, understand it, and if desired, interact with it.” The River Sentry design i...
Read more

River Sentry: Will the New Sirens Work? (Part Two)

-
Image
The new Texas SB3 legislation calls for installation of sirens. What is their purpose? Flood tragedies and nighttime have a deadly relationship. Approximately 75% of the flood deaths happen at night. Exclude low water crossings and this jumps to 80%. Like a volatile fire triangle, flood tragedies feature a three-ingredient chain: sleep, proximity, and rising water. To prevent the horrors, we must break the chain.   Mother nature owns rising water, and proximity cannot be affected if one is resting so you must attack sleep first. Will the sirens do it? Commercial Fire Code NFPA 72 Section 18.4.5.1 requires an alarm minimum of 75 Decibels (dB) in each sleeping space. A typical city siren produces around 130 dB. Only 500 feet away, heavy rain absorption, loss over distance, and structure insulation lower the level toward 58 dB, far lower than fire code and unlikely to wake anybody. The new legislation defined these as outdoor sirens so perhaps that’s expected but if so, where is t...
Read more

River Sentry: Fire Code in the Flood Domain (Part One)

-
Image
It’s 2 AM. You are sound asleep on the fifth floor of your hotel when a fire breaks out. The danger is immediate. The escape window is limited. How do we solve for alerting in this time critical scenario? Do we utilize distant city sirens unlikely to generate sufficient decibel levels to wake up those indoors? How about SMS text warnings to guest phones that may be silenced? Fire code dismisses any such remedies. Loud alarms are required in each room as are lighted egress paths. It makes sense, one must be awake and moving to escape. Before that escape can occur, timely detection of the danger must quickly and accurately occur. Fire alarms use smoke and heat detection to autonomously activate. How do we bring such autonomy to flash flood warning?   River Sentry philosophy utilizes perimeter warning sensors to detect rising water as it approaches. The sensor methodology differs but the desired effect is the same. Get up and get out! We can further examine the flood warning chall...
Read more