Improving Tornado Warnings Will Help Save Lives

When we get the psychology of warnings wrong, disasters happen, as devastatingly shown by the southeast US tornado disaster last week. Warning information was provided in plenty of time, yet still dozens were killed. How can we do better?

Tornado damage in Texas in 2007.

Source: Ilan Kelman

The morning of the tornadoes, the US National Weather Service warned people across the region to be ready for severe storms and to ensure that multiple information channels were open to receive updates. Their language became increasingly urgent that afternoon followed by direct advice to be ready to shelter and then, when a tornado was sighted and the pathway forecast, to act immediately.

“ALERT*** Heads up Mayfield Kentucky. Tornadic storm moving your way, could arrive by 9:30 p.m. Be ready to shelter immediately! This is a dangerous storm!” was sent at least 20 minutes before the town was destroyed. Typical pre-tornado warnings give perhaps half that time.

This sequence eerily paralleled the July 2021 floods across western Europe with almost 200 deaths. Five days before the deluges began, the European Flood Awareness System forecast a high probability of major flooding. Hours before the disaster, clear warnings stated imminent inundation. It is hard to count how many lives were saved. It is easier to tally how many perished, all unnecessarily.

More Than Technical Information

In the US, western Europe, and many other disasters, the technical warnings were on target but the social warnings were lacking. We are working on understanding people’s information acquisition, decision-making processes, and behavioral responses.

This research dates back decades, such as a study of the 1976 Big Thompson, Colorado flash flood which killed over 140 people. Scientists reconstructed choices made by those killed to identify what information they needed and how it could have saved their lives. The results became an education program including action-oriented signs along the flooding canyons. This long-term warning process proved its worth in 2013 when similar areas flooded, yet deaths numbered under a dozen across the entire state.

Pooling all this knowledge helps us to better determine how to constructively influence behavior, especially in developing a mindset that thinks, plans, and acts long before a crisis manifests. One key is avoiding the myth of “the last mile” of warnings which assumes that saving lives is simply about having the technical capability to monitor the environment for threats and then to get the right information to the right people at the right time. Instead, we must flip it around to the “first mile,” which starts with people, to learn about their information, warning, and action needs. Then, technical systems can be designed to match people’s needs rather than people having to match the technical needs.

Evidence of Doing Better

Some such glimmers of hope emerged in the tornado tragedy. Thousands did receive the warnings, were able to act on them, and did so. Everyone affected was at a disadvantage anyway, because the tornadoes appeared in winter at night, with both factors known to increase recent death rates. Even though 100 dead is horrible, it is far better than 1,000.

The US, in fact, has had many higher-fatality tornado disasters. In 1925, at least 695 people died; in 1840, it was around 317; and 1896 as well as back-to-back catastrophes in 1936 each killed over 200 people. Kentucky has been hit previously with the 1890 tornado outbreak killing at least 76 people in the state and perhaps double that number overall.

Since these historic calamities, population numbers in the affected areas have increased, placing many more people in a tornado's potential path. Current disasters could be much more lethal than previous ones.

Yet in the meantime, awareness, knowledge, interest in acting, building codes, structures, warnings, evacuations, and shelters have significantly improved. Today, people have vastly increased opportunity to avoid harm irrespective of the tornado—as long as they have the resources and options to heed warnings.

They might not be able to afford a well-stocked underground shelter and mobile device alerting them. They might not speak the languages in which warnings are issued or might not be able to visualize the maps provided. They might be hearing impaired and so, especially when asleep, be unaware of local sirens. They might be told that if they do not arrive at work because they are sheltering, or if they head for the factory’s safe room with every warning, then they will lose their job. They might have untreated anxiety or other mental health difficulties precluding continued vigilance and robust actions when tornadoes are deemed to be likely.

Where these provisos are absent, we can succeed. On 13 July 2004, a tornado flattened a manufacturing plant in Illinois. Around 140 people were inside and survived because they reached the plant’s shelters less than five minutes beforehand. Then last week, multiple deaths occurred in commercial facilities in Kentucky and Illinois.

Environment Essential Reads

How Outdoor Activities Build Resilience in Children

Physical and Mental Health Risks in an Intense Heat Climate

We have decades of science and experience. Our mindset is often fixated on warnings being one-way technical information that a tornado is likely or witnessed, so now mosey on down to your safe space. Others realize that this part is a small section of the warning process. It must also incorporate people’s circumstances over time, including poverty, languages, resources, disabilities, and fear of assault while evacuating and sheltering.

Other aspects require deeper investigation. What happens in an underground shelter if flash flooding accompanies the tornado’s storm? If debris jams a shelter’s exits, how long until the people get out? What role do post-tornado fires and electrocutions play in casualties?

Warning processes can and should all address these factors, but long before a storm forms.

Beyond Climate Change

All this discussion covers people and society, not tornado traits. We know that we are changing the climate with observable consequences for weather, including tornadoes. We do not have enough science to state exactly how tornado formation and characteristics—including frequency, size, and duration—are being affected. Certainly, with climate change yielding warmer winters, tornado weather is more prevalent, although tornado weather does not always translate into tornadoes.

In any case, the southeast US typically displays a small peak of tornadoes in early December. Kentucky lists numerous December tornadoes dating back to the nineteenth century. Attributing last week’s tornado to human-caused climate change does not reflect the science.

We need to change our mentality to stop blaming climate change for all weather and especially to stop blaming climate change for storm-related disasters. Putting this psychology of warnings into practice draws on a wealth of knowledge with successes evident in the many lives saved during US tornadoes.

Given the recent death toll, there remains a long way to go.