How do you convince people and officials in charge to expect the unexpected? That question has come into the spotlight in recent weeks after two deadly yet highly contrasting U.S. flash flood events.

  • In Middle Tennessee, 21 people were killed as localized severe thunderstorms dumped more than 17 inches of rain early on the morning of Aug. 21, sending high water ripping through the town of Waverly.
  • Intense downpours on Sept. 1 related to the remnants of Hurricane Ida inundated thousands of homes and businesses across the Mid-Atlantic and Northeast, including many areas unaccustomed to high water. At least 57 deaths were reported, nearly all of them flood-related, including 11 who died in basement apartments in New York City.

Both events were well-predicted by historical standards. Yet for many affected, the suddenness and intensity of the flooding seemed to vault past expectations set by even the most alarming warning messages.

“By definition, record-breaking events have never been observed or experienced before,” said Maureen O’Leary, a spokesperson for the National Weather Service, in an email. “So when the [Weather Service] uses language such as historic or record-breaking, some people cannot fathom the amount of or force of rapidly moving water.”

Weather and climate researchers, with expertise in risk communication, are exploring ways to raise public awareness for the most dire situations, drawing from lessons of the past.

Modernizing a mid-century alert structure

Weather Service communication on flash floods relies on a system of watches, warnings and advisories first developed in the 1950s and still used for other hazards such as hurricanes, tornadoes and severe thunderstorms.

Experts have long pondered whether the watch-warning-advisory system has outlived its usefulness. Yet changing a paradigm so deeply rooted takes time.

According to O’Leary, social science sponsored by the Weather Service has confirmed that the “watch” and “warning” terms are easily confused with each other. “Watch” can also be confused with “advisory,” the other of the three commonly used Weather Service headlines.

As a result, the Weather Service announced in March that it will be abandoning “advisory” and moving in the next three years toward plain-language headlines that focus more specifically on the phenomena and impacts expected.

However, the Weather Service has no current plans to overhaul the overall watch and warning structure.

Within the legacy system, one high-profile attempt that has emerged to flag the most serious threats is the use of “emergency.” The phrase “tornado emergency” was first invoked by the Weather Service office in Norman, Okla., as an F5 tornado bore down on central Oklahoma on May 3, 1999. It has since been adopted formally, specified for use when a tornado is clearly underway and a “severe threat to human life” or “catastrophic damage” is “imminent or ongoing.”

Similar criteria are now in place to trigger a flash flood emergency, which was issued for both the Tennessee and Northeast flood events.

It’s often the piercing howl of a cellphone that warns people of the most dire weather threats. All tornado and hurricane warnings activate the Wireless Emergency Alert (WEA) system, which is embedded in all cellphones built over the past few years.

However, as of 2020, the only flash flood warnings that prompt the WEA into action are the subset for events deemed “considerable” or “catastrophic,” flash flood emergencies being the latter. Before this change, every flash flood warning — around 4,000 per year nationwide — triggered the WEA alarms. Many would go off while people were asleep.

“Impacts could range from a flooded low-water crossing to ongoing swift water rescues,” said O’Leary. “The perception was that the [Weather Service] over-alerted flash flood warnings.”

What was communicated — and when — in this summer’s twin flood disasters

Researchers working to improve weather hazard messaging have a potent ace up their sleeve: an increasingly rich and skillful set of meteorological guidance.

The day before the deadly flash flood in Tennessee, computer models highlighted the potential for a corridor of “training” thunderstorms, tracking over the same area repeatedly, to develop just west of Nashville. Forecasters at the Weather Service office in Nashville issued a flash flood watch at 4:55 p.m. on Aug. 20, to go into effect from 1 a.m. through the afternoon on Aug. 21. The watch included those areas that would be hardest hit.

Areas in and near Waverly were placed under a flash flood warning at 3:29 a.m. Aug. 21, and a flash flood emergency at 7:47 a.m. By that point, waters were rising rapidly along and near key roads that run through the center of town near Trace Creek.

Even residents who got the “emergency” message may have wondered which was the safer option: driving through flooded roads or staying at home. The first floors of many structures in Waverly ended up swamped with floodwater.

“Just seeing water around us is normal in most regions of the country,” said Kim Klockow-McClain, a research scientist at the University of Oklahoma. Klockow-McClain authored the flash flood section of a 2016 research report on risk communication and best practices.

“When does water become dangerous?” said Klockow-McClain in an email. “It's a matter of degree, unlike with a tornado or hurricane, where their mere presence inspires fear.”

As for roadway safety, Klockow-McClain praised the Weather Service’s “turn around, don’t drown” message, now used widely to exhort motorists not to drive into floodwaters. However, she added, “Most people don’t know flooding is possible until they happen upon it. By then, they’re very motivated to get where they were trying to go. That motivation doesn’t just leave a person when they see water in the road. They need alternate options.”

In the Northeast, there was ample advance notice of impending large-scale trouble with the track of former Hurricane Ida well predicted. Flash flood watches went into place in and around New York City at 3 p.m. on Aug. 30, more than two days before the city was swamped — a phenomenal amount of lead time.

The watch warned that 3 to 5 inches of rain was possible, with localized totals even higher.

The potential for severe flooding was evident by the afternoon of Aug. 31, when the Weather Service’s Weather Prediction Center placed much of the Northeast under a “high risk” of excessive rainfall that could lead to flooding on Sept. 1. The center warned some locations had a greater than 95 percent chance of seeing rains so extreme they would typically occur about once a century — or even less.

As the rains intensified on the afternoon of Sept. 1, multiple flash flood warnings were issued throughout the Interstate 95 corridor. By evening, areas from Philadelphia to New York to Connecticut were in their first-ever flash flood emergency, and cellphones were screeching with multiple Weather Service and local government alerts.

Yet here too, the best options for safety were not always obvious.

Some residents were under simultaneous tornado warnings and flash flood warnings, which led to competing messages: Take cover and move to higher ground. These seemingly incompatible warning pairs are surprisingly common, with an average of 400 such overlaps per year detected in one study.

The heart-wrenching deaths in New York basements pointed to another awareness gap, according to Klockow-McClain: With so much stress on knowing how to escape flooding while out in the open, there’s less emphasis on the risk of high water at home. “If you don’t have a federally backed mortgage, do you even know if you live in a flood plain? What about living in a basement?”

Klockow-McClain points out that renters are especially unlikely to know their flood risk and that flood-susceptible housing is dominated by lower-income structures, typically on the cheapest land. “We manufacture this exposure and ignorance by design,” she says.

Laying the groundwork for flash flood response

Decades of research has found that the most intense rain events are getting heavier as human-caused climate change unfolds, a trend only expected to worsen.

Even the most torrential rains do not always translate into floods — that outcome hinges on soil moisture, the built landscape and other factors apart from immediate weather — but new research suggests that the most extreme floods may likewise intensify, while lesser-end flood events wane.

Ahead of Hurricane Henri on Aug. 21, Central Park measured 1.84 inches in an hour, the highest in online data going back to 1943. Yet New York City’s sewer system was only designed to handle 1.75 inches of rain in an hour.

Ida set the bar even higher. On Sept. 1, Central Park received what appears to be its heaviest one-hour rainfall on record: 3.15 inches, just ahead of 3.05 inches between 11 a.m. and noon on Sept. 4, 1913.

Some state and local leaders, including New York Mayor Bill de Blasio (D), asserted that no one projected the rainfall’s intensity, despite the Weather Service noting in advance that century-scale rains were possible.

It’s essential for key stakeholders in flash flood safety to build relationships during calmer times, says Eve Gruntfest, a professor emerita of geography from the University of Colorado at Colorado Springs. Gruntfest conceptualized the WAS*IS program, a pioneering effort at the National Center for Atmospheric Research that integrated meteorology and social science.

Gruntfest’s study of the behavior of survivors and non-survivors of the Colorado’s catastrophic Big Thompson Flood inspired the development of the “climb to safety” signs now widespread in Western canyons. The flood killed 144 people in northern Colorado on the night of July 31, 1976.

Most of the victims died in their cars when the road was washed away by raging floodwaters, yet could have survived had they gotten to higher ground, Gruntfest found.

Gruntfest noted that the record-smashing Colorado floods of September 2013, which included both river and flash-flood elements, took eight lives, far fewer than the 1976 disaster. “One major reason was the coordination and collaboration between forecasters, local governments, the media, and others that was carefully curated over the years,” she said.

Understanding that another catastrophic flash flood was possible, the partners built on mutual trust to take action, including closing roads, businesses and schools to reduce the number of people on roads — many of which were washed away. “The number of people lost was significantly reduced by the proactive behavior of the partners when it mattered most,” Gruntfest added.

Bob Henson is a meteorologist and journalist based in Boulder, Colo. His books include “The Thinking Person’s Guide to Climate Change” and “Weather on the Air: A History of Broadcast Meteorology.”