Flash Floods: From prediction to protection

Written by Ana Barros

Op-ed from the Center for Secure Water (C4SW)

My heart is broken. Again. Watching the coverage of the flash floods in the Guadalupe River in Texas on TV, we remember the flash floods in the Central and Southern Appalachians over the last five years that killed so many, including small children. While these tragic events feel rare in the moment, and within a local area, flash floods are occurring somewhere all the time in complex terrain: in the headwaters of the Rockies, along the Appalachians, in the Texas Hill Country, and in mountainous regions around the world. Flash flooding is the severe weather disaster that causes more deaths nationwide and globally.

Photo Credit: Kamran Khan

Like the classical question of whether a falling tree makes a sound if nobody is there to hear it (spoiler: it does!), flash floods often strike in remote areas, far from the public eye. Yet, when flash floods occur where people live, the result is almost always catastrophic. These floods can trigger landslides, debris flows, and massive erosion—relentless forces of natural terraforming. The height of the limestone bluffs along the upper reaches of the Guadalupe River tells the story of the floods that shape the river, time and time again.

Photo Credit: Thomas Park

Traditional risk analysis that relies on historical records severely underestimates the real, widespread risk of flash-flooding regionally. When severe weather interacts with topography, runoff generated by localized heavy rainfall in steep terrain quickly reaches streams to form tall fast-moving (flash) flood waves. Whereas flash-flood watches are commonly issued under heavy rainfall conditions, the rapid escalation to more specific flash-flood warnings can leave little time for people to act. Unfortunately, the potential for loss of life can be exacerbated overnight when people are asleep.

The right response to flash flood warnings is to get out of the floodway, and to do so quickly. Even if people receive a warning though, they may not know what to do. Flash flood risk must be acted upon with a heightened sense of danger that requires more than running headlines hours ahead at the bottom of TV screens. Beyond physical sirens and radio communications in community-operated warning systems, mobile technology is now available that enables location-based individualized warnings. A national flash flood system to provide actionable information is needed and is at our fingertips.

New modeling supports a new flash flood system

The basic operational requirements for such a system are the following: 1) actionable precipitation forecasts with long lead-times (1-2-3 days ahead of the event); 2) localized flood response forecasts with long-lead times; and 3) situational information that directs people to the nearest safe location to wait the flood. The first indicates potential occurrence (when it can occur) and severity over a region. The second is essential to position flash-flood risk (where it can occur) and potential impact. The third is the essential final leg of disaster preparedness (to protect the public).

Despite the National Weather Service (NWS) making great strides in precipitation forecasts over the last two decades, accurately forecasting flash floods in complex terrain—especially with long lead times—remains a scientific challenge. However, operational numerical weather prediction models, such as the NWS High-Resolution Rapid Refresh (HRRR) with its 3 km spatial resolution, offer valuable long-range information that can be used to drive flood models and issue actionable alerts. Furthermore, these operational weather forecasts can be tailored to place-based conditions using AI technology. Geospatial information at spatial resolution of 30 m and less is available to enable mapping evacuation routes and defence strategies that can reconcile situational information with guidance from emergency response services. This was demonstrated twenty-five years ago already. However, the practical implementation of such systems was not feasible then due to the lack of computational capacity at scale and the lack of data resources made possible by the Cloud. We have the tools today to do better.

Photo Credit: Mathew Risley

We can do better

Often the spectres of flash flood warning fatigue and fear of inconveniencing the public for “nothing” are raised if a predicted disaster fails to happen. It is not for nothing. Civil protection is a pillar of secure and prosperous communities. It requires that communities develop comprehensive disaster preparedness plans addressing their unique vulnerabilities to flash floods. We can do better to educate and empower citizens on situational flood risk.

On my TV screen, a commentator claims the Guadalupe flash flood came "out of nowhere." Statements like this—rooted more in fiction than science—reinforce a troubling sense of helplessness. All of it unfolds against the haunting backdrop of smiling children still unaccounted for at a Texas summer camp. We owe it to them, and to future generations, to do better. Better flash flood forecasts, better public education, better communication, and better preparedness can—and will—save lives.

Ana P. Barros is the Civil and Environmental Engineering Department Head and the Donald Biggar Willett Chair of Engineering at The Grainger College of Engineering, University of Illinois, Urbana-Champaign, and also serves as the director for the Center for Secure Water (C4SW).  Barros conducts research on mountain hydrology and hydrometeorology including aerosols, cloud and precipitation processes across scales; land-atmosphere interactions; vegetation dynamics; and soil-vegetation-water cycle interactions in land-margin environments using field and lab experiments and high-resolution process-resolving models. 

In the news: U of I professor says flash flood emergency response should be reevaluated, WCIA