It’s time to revisit how we see Americans falling prey to hurricanes, according to new data from the National Hurricane Center (NHC). There’s always the risk of a storm-surge catastrophe like 2005’s Katrina, or of Category 5 winds slamming into poorly built homes, as with Andrew in 1992. Yet most U.S. hurricane deaths in recent years can be chalked up to factors that lie beyond these well-recognized threats.

A first-cut analysis was presented by NHC’s two top leaders at recent meetings sponsored by the American Meteorological Society:

Analyzing all fatalities caused by hurricanes between 2017 and 2021 in the contiguous United States, Brennan, Rhome, and colleagues found that indirect deaths outnumbered direct deaths by 299 to 271. (Had this analysis included Puerto Rico, the indirect-death numbers would have been overwhelmed by 2017’s Hurricane Maria. The initial death toll was 64, but the final official toll incorporated an estimate of more than 2,900 hurricane-related deaths during the six months after Maria.)

Almost 30% of the indirect deaths reported by Rhome and Brennan stemmed from disruptions to the power grid, an ever-more-consequential factor in an increasingly wired world. Carbon monoxide poisoning – a notorious risk when home generators are used improperly – claimed 48 lives. Another 39 people died from electrocution and other hazards related to power supply.

Other indirect fatalities stemmed from a variety of causes, including vehicle incidents (49 deaths), recovery/prep accidents (34 deaths), and heat-related issues (38). Most of the indirect deaths appeared to be among people 60 or older, Brennan said.

Destruction from Hurricane Laura
Figure 1. Widespread destruction from Hurricane Laura in Lake Charles, Louisiana, in late August 2020 included extensive damage to the city’s power grid. (Image credit: Marty Bahamonde/FEMA via DVIDS)

Are surge deaths receding?

Even when looking at the 271 direct deaths, the picture is far different from the classic hurricane image of storm surge and winds claiming lives along and near the coast. In fact, only 3% of the direct deaths (eight people) were from storm surge – five of those caused by Category 5 Hurricane Michael in 2018 – whereas freshwater flooding accounted for 65% of the direct deaths (175 people).

It’s not that the 2017-21 period lacked surge-producing landfalls. Category 4 Hurricane Laura pushed dangerous surge into highly vulnerable southwest Louisiana, including Cameron Parish, where storm surge claimed more than 300 lives during Hurricane Audrey in 1957.

NHC has upped its surge-awareness efforts dramatically in recent years. In the mid-2010s, NHC in public statements began to emphasize potential inundations above ground level, a more intuitive concept than the surge values that were long provided relative to mean sea level. Experimental mapping launched in 2014 was followed in 2017 by the first official storm surge watch and warning products. In 2020, NHC introduced an experimental graphic depicting peak storm surge.

For Laura, NHC addressed the threat of storm surge inundations as high as 15-20 feet with unusually strong wording. Several of the public advisories for Laura stated, “Unsurvivable storm surge with large and destructive waves will cause catastrophic damage from Sea Rim State Park, Texas, to Intracoastal City, Louisiana, including Calcasieu and Sabine Lakes.”

Peak inundation from Laura just east of landfall was estimated in NHC’s tropical cyclone report to be 18 feet above ground level, with several reports in the 12- to 18-foot range (and waves above that level) in the vicinity of Creole and Grand Chenier.

“We took a beating at the NHC for that,” Rhome said in his presentation, referring to the use of “unsurvivable” to describe Laura’s expected surge. Yet the wording may still have helped save lives. Rhome cited a 2015 post-Sandy study led by Jennifer Marlon (Yale School of the Environment) that found 22% of coastal Connecticut residents could be categorized as “diehards” resistant to evacuation under almost any circumstance.

“If you apply that 22% to the census information in Cameron Parish, somewhere around 1,500 [would] have stayed, ignored the message, said ‘phooey, I don’t care’.” Instead, Rhome noted, “there wasn’t a single storm surge fatality or a single rescue.”

Against this recent backdrop of success, NHC and emergency managers now are looking more closely at some less-emphasized threats. From 2017 to 2021, there were 38 deaths along the Gulf and Atlantic coasts from high surf and rip current related to tropical cyclones. This total includes eight fatalities from 2019’s Hurricane Lorenzo, the Atlantic’s eastern-most Category 5 on record, which remained thousands of miles away from the U.S. East Coast during its entire lifespan.

Trees brought down by high winds from tropical storms and hurricanes, sometimes well inland, are another risk that deserves attention. The study period included 28 wind-related deaths, many of them linked to tree falls.

Despite the massive damage inflicted on Louisiana by a series of hurricane landfalls between 2017 and 2021 – including two of Louisiana’s strongest hurricanes on record – the state reported just 10 direct fatalities in that period. Higher direct death tolls occurred in Texas (77), North Carolina (45), New Jersey (33), Florida (27), and New York (20). Most of the direct deaths over the Mid-Atlantic and Northeast resulted from torrential rains and subsequent flooding from the remnants of Hurricane Ida in 2021.

In contrast, indirect deaths were more prevalent closer to landfall locations, where lengthy recovery processes were common, including in Florida (127), Louisiana (55), Texas (46), and North Carolina (27).

Fatalities maps
Figure 2. Most of the deaths directly produced by Hurricane Ida and its remnants occurred in the Mid-Atlantic and Northeast, whereas the indirect deaths in Ida’s wake were mostly concentrated near the storm’s landfall location in southeast Louisiana. (Image credit: Michael Brennan and Jamie Rhome, NHC)

Where might climate change push hurricane fatalities?

With signs evident that human-produced greenhouse gases may be enhancing rainfall from some tropical cyclones, and further increases expected, the deaths from Ida (see Figure 2) could be a harbinger of more to come. Most of Ida’s 80-plus U.S. deaths came from catastrophic flooding in the Mid-Atlantic and Northeast, hundreds of miles from where the storm made landfall, whereas carbon monoxide poisoning caused a number of the indirect fatalities in Louisiana.

As Ida’s remnants pushed toward the U.S. East Coast as a post-tropical cyclone, forecasters gave ample notice of the dire flood threat taking shape. Some of the region’s first-ever flash flood emergencies were issued by local National Weather Service offices. Newark International Airport reported 8.41” of rain on September 1, the highest calendar-day total ever recorded at any major reporting site in the New York metro area, and Central Park recorded its rainiest single hour (3.15”) in more than 150 years of recordkeeping.

“We’re seeing rainfall rates that we haven’t seen before in some of these urban areas,” Brennan said at the AMS hurricane conference. “That’s certainly an important piece of this going forward, to make sure that the messaging continues downstream as the winds weaken and the storm may not even be a tropical cyclone anymore.”

Rhome, Brennan, and colleagues say they plan to expand their initial work to include tropical cyclones from 2013 through 2016. Despite the inherent limits in such a short study period, there are enough signals to merit closer looks at a broad range of hurricane and post-hurricane threats, especially issues related to the power grid and people’s desire to keep juice flowing. This hazard isn’t limited to hurricanes: In the Texas cold-wave disaster of February 2021, 20 people died from carbon monoxide poisoning, with many others dying as a result of power and heating loss.

As indirect deaths become a more common metric, challenges arise in making apples-to-apples comparisons with historical hurricanes. Recent techniques for assessing indirect fatalities (such as using death certificates to estimate “excess” deaths, as was done for Maria) seldom are applied to long-ago storms. What’s more, death tolls among Black hurricane victims were underestimated for many decades, including a longstanding undercount of more than 600 in the 1928 Lake Okeechobee disaster alone. All this makes it harder to confidently assess long-term trends in total fatalities.

What is clear is that the pieces holding together modern society – especially access to power – can be ripped apart at frightening speed in a hurricane, often taking days, weeks, or even months before they can be reassembled, leaving vulnerable populations in dangerous places. Such threats will be accentuated to the extent that landfalling hurricanes intensify more rapidly, become stronger on average, move more slowly, and/or dump heavier flood-producing rains.

Jeff Masters contributed to this post.

Website visitors can comment on “Eye on the Storm” posts (see below). Please read our Comments Policy prior to posting. (See all EOTS posts here. Sign up to receive notices of new postings here.)

Bob Henson is a meteorologist and journalist based in Boulder, Colorado. He has written on weather and climate for the National Center for Atmospheric Research, Weather Underground, and many freelance...