As the past few weeks have shown, flash floods can develop very quickly in both rural and urban areas, with mild to catastrophic impacts. That’s part of why flash floods are so critical to study, according to Andrew Kruczkiewicz, a meteorologist and senior staff researcher at the Columbia Climate School’s International Research Institute for Climate and Society. “Some parts of the world will never see potential disasters like wildfires or tropical cyclones, but there are very few areas at zero risk of flash floods,” says Kruczkiewicz. And because of the wide variability in potential impact, “it demands extra sensitivity in terms of how you communicate risk, as a run-of-the mill flash flood is very different than a 30-foot wall of water.”

Kruczkiewicz’s current research focuses on extreme weather events such as flash floods, and the application of climate and weather data and forecasting to reduce disaster risk and facilitate humanitarian action. In the following discussion, Kruczkiewicz talks about why the Texas floods were so devastating, how warning systems need to consider very different populations—e.g. recreation-seekers vs. locals—and how we might integrate both technology and local knowledge to avoid such tragedy in the future.

What caused these floods, and why were they so catastrophic?

There are many factors at play when we’re talking about flash floods: meteorological, climate-related, geophysical, and those related to the built environment. On the meteorological front, the moisture associated with the remnants of Tropical Storm interacted with a mesoscale convective vortex (MCV) over central Texas. The MCV interaction allowed for new, intense and rainfall-rich thunderstorms to form. In addition, these storms were both very slow moving. Underlying drier-than-average conditions on the ground were also a contributing climatological factor when determining why the impact of the floods on lives and livelihoods was so great.

One of the fingerprints of climate change is that warmer atmospheres lead to wetter tropical systems. But flash floods have been happening in this area for a very long time. Did climate change make this worse? We should be asking that question. Other factors at play include socioeconomic shifts, land cover changes and land use changes. We must consider all of this when trying to assess increased levels of risk, broadly speaking, in order to develop revised policy and disaster preparedness, warning and anticipation approaches.

Some news reports are calling this “Flash Flood Alley.” The area impacted by the floods is indeed part of a flash flood alley, and there is this swath of central Texas where we see a higher base level of risk. But these are factors we see not only in the Texas Hill Country, but also up into northern Texas, perhaps even into Oklahoma and down into New Mexico. “Flash flood alley” is not a well-defined term, and more work should be done to define and map the various types of flash flood alleys, as they are areas that experience elevated risk, and therefore require more specific approaches depending on where they are. For example, can there be flash flood alleys in cities too, mostly due to poor urban planning? Yes, if there are neighborhoods or other areas where we see higher levels of risk compared to other areas of the city.

Would better communication around the risk of flash floods have helped save lives?

There are very few parts of the world that are at zero risk of flash floods, and because of this, we need different types of risk communication. What does “flash flood risk” mean to the people who already live in Flash Flood Alley? How many flash flood warnings have been issued over the past five years, and what did those resulting flash floods look like? There have been some false alarms, for sure. But for those flash floods that did occur, what were their impacts? Certainly nothing close to what we saw with this one.

In areas where flash floods are common, a “run-of-the-mill” flash flood is likely to be perceived as something locals can deal with—it might wash away your car or your cow—and while that leads to economic stress, which disproportionately impacts lower income populations, it’s very different from a 30-foot wall of water. And I think something we’re learning is that the binary of “no flash flood warning” versus “flash flood warning” is not sufficient. The “flash flood emergency” is a relatively new mechanism. It was meant to address the issue of different intensities of flash floods. So I think we need more sensitivity in the way we issue these different types of messages.

Also, many of the people who died weren’t living in this area. They were there for recreational purposes. Who has the responsibility to communicate risk and raise their awareness—to tell them they’re entering a zone where the baseline risk is low, but definitely non zero? What seasons or months will the baseline risk be elevated? And what does an enhanced risk of flash floods mean in an area where the baseline level of risk is already higher than surrounding areas? As flash floods increase in intensity, we are at a critical moment to review and revise our approaches to flash flood risk mitigation, early warning and anticipatory action—and these questions will support moving toward more comprehensive systems.

What tools or technology might’ve helped warn people to evacuate?

This is an opportunity for us to better integrate new technology with “traditional” technologies such as a siren system, or what we call NOAA weather radios, which use a VHF frequency. People don’t use these radios much anymore because they think their phone will do the trick. They assume that if risk suddenly increases, they’ll get an automatic push notification. But sometimes those notifications don’t come. The VHF frequency is much more reliable—weather radios don’t need to be charged. You put some batteries in, and they last for a very long time.

I haven’t heard many people talk about leveraging local knowledge—or we could even say indigenous knowledge—of how flash flood risk was perceived and managed for hundreds of years before this area was built up. Even before sirens and radios, there were other approaches. What were those? How can they inform the next generation of risk awareness and warning systems for flash floods?

Could new technology, such as AI modeling, have helped here?

I think there’s an interesting lens here related to our trust in AI, in climate tech, and in our weather apps. In many locations we are learning that AI-driven forecasting is better than numerical weather prediction when we’re talking about average or moderate amounts of rainfall. That’s because we have a lot of training data representing those types of events. However, when we’re talking about events that are so far outside the norm, we don’t have the historical data to train the AI modeling infrastructure, and so we don’t have evidence that AI-driven products are doing much better at forecasting disasters in a given geographic area.

How can we better prepare for such extreme events?

At the very least, everyone should have some sense of the potential for extreme weather in the place where they’re going to be sleeping on a given night. When we have periods without disasters, social science shows that we become comfortable. In the back of our minds, we’re aware there is a risk of a flash flood, but we don’t like to think about it, especially when we are trying to enjoy a vacation. This is a common phenomenon in disaster response and risk reduction, and we need to do a better job of addressing complacency and the underestimation of risk.

Who should be responsible for communicating risk?

The primary question to ask is, what type of communication motivates people to take meaningful action at the right time? Up until recently, one approach was based on frequent and consistent messaging by trusted and authoritative sources. However, the “trusted” element is key here. To what extent are sources trusted by everyone in a certain community or county? Do people at the highest level of risk trust the same sources as those at moderate risk? If you are traveling in a new area, for work or tourism, for example, is your perception of “trusted source” the same as those who live in that area?

If there’s a policy reaction to the Texas flash floods that’s focused more on improving risk communication and warning messaging for those away from home, such as people on vacation or camping, it could overshadow an equally important factor: there are people living in these areas, and data shows that, particularly for flash floods, people at the highest risk levels are more likely to be economically underserved. Even if they receive a warning, they may not be able to evacuate or can only evacuate into an area that is at even higher risk. For this event, how many people received the warning and couldn’t leave for various reasons? This is a good opportunity to ask these questions and see if they can inform policy for people living in areas at all levels of risk for flash floods, in addition to developing approaches to communicate risk and warning messages for people moving into such areas for work, vacation or other reasons.