The real reason billion-dollar disasters like Hurricane Helene are growing more common
It rises like a mountain, up and to the right, and it has become one of the most potent illustrations of the perils of man-made global warming. It’s a chart showing the number of billion-dollar weather disasters that have struck the United States since 1980. When the toll is tallied from hurricanes Helene and Milton, they will become the 397th and 398th entries in the database.
The disaster data, maintained by the National Oceanic and Atmospheric Administration, has featured in multiple government reports on global warming. The Biden administration has referenced it at least seven times to help make the case for climate policies. Members of Congress cited it in a bill to curtail the use of fossil fuels. Last year’s National Climate Assessment, a congressionally mandated report on climate change, showed the disasters on a map under the heading “Climate Change Is Not Just a Problem for Future Generations, It’s a Problem Today.”
But according to disaster experts, former NOAA officials and peer-reviewed scientific studies, the chart says little about climate change. The truth lies elsewhere: Over time, migration to hazard-prone areas has increased, putting more people and property in harm’s way. Disasters are more expensive because there is more to destroy.
The billion-dollar disaster dataset is “quoted a lot and people use it as a way of saying that climate change is already influencing what we see. And yet, unless you get the economics right, you can’t really justify that,” said D. James Baker, the physicist and oceanographer who led NOAA from 1993 to 2001, the longest tenure as administrator in the agency’s history.
Baker and others say disputing whether global warming’s influence can be found in the disaster data is not the same as questioning whether climate change is real or whether society should switch from fossil fuels.
“We know that climate change is real. We don’t see it in the [economic] losses yet,” said Laurens Bouwer, an expert on the assessment of climate risks and a lead author on five reports by the U.N. Intergovernmental Panel on Climate Change, the advisory body through which scientists reach consensus on climate change.
Although traditional statistical methods cannot quantify the influence of greenhouse gases on rising disaster costs, many scientists say that global warming has intensified hurricanes, wildfires, droughts and other extreme weather, which must be leading to greater economic losses.
“A lot of these extremes are really ramped up,” said Adam Smith, the NOAA climatologist who has led the billion-dollar disaster project for more than a decade. “If you want to act like nothing’s happening or it’s minimal, that’s just not the case in what we’ve seen in these extreme events in the United States.”
In September, Smith experienced a billion-dollar disaster firsthand when Hurricane Helene’s record rainfall overflowed the rivers that run through Asheville, N.C., where NOAA’s National Centers for Environmental Information is located. Smith and the other NCEI employees survived the floods, but the agency’s trove of meteorological data, including the billion-dollar disaster website, was knocked offline and remained inaccessible for weeks.
“No water or internet but we just received power,” Smith emailed six days after the storm. “It’s a disaster here.”
As the billion-dollar disaster tally climbs, the question grows more urgent: Is global warming to blame?
To answer that, it helps to first ask another question: What do scientists really mean when they say that global warming is causing a trend? For that matter, how does scientific knowledge get created in the first place? In the case of the billion-dollar disaster dataset, the answer begins with two self-described weather nerds at NOAA’s office in Asheville.
Neither of them can remember exactly when or whose idea it was, but some time in the early 1990s, Neal Lott and Tom Ross started tracking weather disasters in the United States that caused at least a billion dollars in economic losses. They scoured government reports and phoned insurance companies to compile their list. The job took up a small portion of their time working at what was then called the National Climatic Data Center.
“At that point, it was not an attempt to do any sort of a scientific study,” Lott said. “As far as climate change, that wasn’t in mind at all.”
Each year brought a longer list of disasters, but the two meteorologists did not think the weather had changed. Instead, they noticed there were more people and property in areas prone to natural hazards.
When Hurricane Opal battered the Florida Panhandle in 1995, rapid growth had already rendered the area’s nine-year-old evacuation plan “critically obsolete,” according to an analysis prepared the following year. Fleeing residents were trapped on packed highways that were not built to accommodate so many people. The storm destroyed 3,300 structures and damaged 18,000 more. Costs ran to $9.6 billion in inflation-adjusted dollars.
Opal and other hurricanes failed to slow Florida’s population boom. The combined population of the 15 counties hit by Milton’s hurricane-force winds swelled from 3.7 million people in 1980 to 9.1 million in 2023. As the population more than doubled, the area’s economic activity, as captured by inflation-adjusted wages earned by the people who lived there, quadrupled over the same time period.
“It doesn’t take as much to get to a billion dollars now than it did 20 years ago,” Ross said. “If a storm hiccups, it gets a billion dollars’ worth of damage now, whereas 20 years ago, it took a lot more than a hiccup to do it.”
As their disaster list grew, so did the public’s anxiety about global warming. Was climate change making the weather more expensive? It was plausible, but Lott and Ross didn’t know. They weren’t even sure how to disentangle how much was due to climate change and how much to risky development.
“I think you have to be careful rushing to judgments one way or the other on it,” Ross said.
Yet disentangling the various causes of an observed trend is just the sort of problem climate scientists excel at solving.
Scientists have used a global network of thermometers to calculate the Earth’s average surface temperature since 1850, which has risen outside the range of the random ups and downs in the historical record.
It’s not enough to show that the temperature is increasing. Scientists can also prove what’s causing it to rise.
This is what the temperature trend would look like if it were influenced only by natural causes like volcanoes, changes to the Earth’s orbit and solar activity. The observed temperature increase cannot be due to natural causes alone.
This practice of identifying trends and evaluating their causes is called “detection and attribution.” First comes detection: Show that a change is so big that it departs from historical randomness. Then attribution: Quantify the extent to which the trend is caused by various factors.
In addition to the increase in average temperature, scientists can also prove that greenhouse gas emissions have caused more extreme heat, higher ocean temperatures, and less ice in lakes, rivers and seas.
The billion-dollar disaster dataset includes seven categories of events, each of which responds differently to global warming. Proving global warming’s contribution to the rise of billion-dollar disasters would require conducting a detection and attribution study for each category separately. No one has done so.
“We’ve never sought to do attribution studies in relation to the extremes listed in the billion-dollar disaster dataset,” Smith said.
Other scientists, however, have sought the signal of climate change in economic losses from hurricanes, floods and other categories of extreme weather. They have not analyzed NOAA’s disaster database specifically. Still, if they were to find a link between global warming and disaster losses in other datasets, that would strengthen the case that global warming was also contributing to the rise in billion-dollar disasters.
The first two scientists to conduct such an analysis both studied hurricanes, but from different angles.
Chris Landsea was so obsessed with hurricanes that his three children are named after retired hurricanes, the ones so destructive that no future storm can bear their name. Landsea studied hurricanes so he could predict when they were likely to form and where they would make landfall.
Roger Pielke Jr. examined what happened afterward, how hurricanes toppled buildings and destroyed infrastructure. If he could understand why they caused so much damage, maybe he could figure out how to make them less costly.
In 1996, Landsea had just embarked upon what would become a distinguished career as a NOAA hurricane forecaster. That June, he published a paper showing that over the previous five decades, the frequency and peak intensity of North Atlantic hurricanes had decreased. Moreover, 1991 through 1994 was the quietest four-year period on record.
That’s strange, thought Pielke, then doing postdoctoral work in environmental policy at the National Center for Atmospheric Research. Pielke was studying the ravages of 1992’s Hurricane Andrew, and he realized the record quiet spell Landsea identified was also the costliest period for hurricanes on record.
That summer in Colorado, Landsea often drove from Fort Collins down to Boulder to play basketball behind the lab where Pielke worked. As they ran up and down the court, the two researchers talked over this conundrum. Were hurricanes’ frequency and intensity unrelated to the economic losses they caused?
To find out, they started with a list of hurricane damage estimates Landsea had compiled from the Bulletin of the American Meteorological Society, which also showed a huge increase over time, even after adjusting for inflation. But Pielke and Landsea went further, taking into account wealth and population trends.
They published their peer-reviewed results in Weather and Forecasting, another AMS journal, in 1998 — a study that was last updated in 2018 in Nature Sustainability. Hurricane destruction has indeed grown, but after adjusting damage to account for the growth in people and property, the trend disappears.
The research hints at the destruction past hurricanes would have wrought in today’s world. Four of the five most damaging hurricanes made landfall in the early 20th century, including the Great Miami Hurricane. That storm cost $105 million in 1926 but would have caused $236 billion if it had hit Florida in 2018.
“The degree of societal change dwarfs expected climate changes by orders of magnitude,” Pielke said. “That’s not to say climate change isn’t important. It’s just to say that we humans are really transforming the surface of the Earth in really profound ways.”
Similar studies have failed to find global warming’s fingerprint in economic damage from hurricanes, floods, tornadoes and crop losses. Of 53 peer-reviewed studies that assess economic damage from weather events, 52 could not attribute damage trends to global warming, according to Pielke’s 2020 review of the literature, the most recent and comprehensive.
To be sure, detection and attribution studies of disaster losses cannot capture all of climate change’s impacts to the economy. Insurance companies have hiked premiums and even withdrawn from hazard-prone areas, citing rising climate risk among other factors. The ongoing Southwest drought, which scientists have linked to global warming, has contributed to water shortages that threaten farms and hydroelectric dams. The U.S. government is spending trillions to adapt to and mitigate the risk of climate change.
Yet the consensus among disaster researchers is that the rise in billion-dollar disasters, while alarming, is not so much an indicator of climate change as a reflection of societal growth and risky development.
“A lot of the things we see in the disaster losses are obscured because there’s an overwhelming signal of more people in risky places, growing populations worldwide, more infrastructure, more assets, higher values,” said Bouwer, the five-time IPCC lead author. “This is the main signal. And that’s where the science is at the moment.”
Hurricanes are by far the costliest category in the billion-dollar disaster dataset, but they are not the most numerous. That distinction belongs to severe storms, a category that excludes hurricanes but includes hail storms, tornadoes and derechos. The category accounts for 202 of the database’s 396 events — slightly more than half.
Melanie Gall, an assistant professor at Arizona State University who maintains a separate database of disaster losses called SHELDUS, said she often struggles to match her data with the billion-dollar database’s severe storm entries, which often span multiple days and include different types of hazards.
For instance, the most recent billion-dollar storm in the database lasted from June 24 to June 26 and included “high wind, hail and tornadoes” across Nebraska, Iowa, Pennsylvania, Rhode Island and Massachusetts.
“They have a tendency to lump tornado outbreaks or severe storm outbreaks, like severe thunderstorms, together to make the threshold,” Gall said. “You cannot use these numbers because it’s sort of arbitrary what qualifies as a $1 billion event.”
NOAA has never changed its process for merging severe storm events and bases its estimates on property insurance claims that combine costs from different hazards, Smith said.
As NOAA’s disaster database has gained recognition, Pielke has emerged as its chief critic. In January, he sent the agency a request for correction alleging it had fallen short of its scientific integrity standards by failing to disclose how it arrived at its estimates for each billion-dollar disaster or to submit the dataset to regular peer review.
In its response, NOAA conceded the need to better document the sources of its estimates but said it could not always share private data provided by insurance companies, adding that it found no errors in any of the database’s entries. NOAA acknowledged the dataset is four years overdue for peer review.
Throughout the years, Pielke has clashed with other scientists, journalists and government officials over the state of the research. He will soon end a 23-year career as a professor at the University of Colorado, and this year he became a nonresident senior fellow at the American Enterprise Institute, a center-right think tank. Pielke plans to vote for Democratic presidential candidate Kamala Harris in November, but his views have found more favor among Republicans, whose nominee Donald Trump has dismissed concerns about global warming.
“I always want to emphasize climate change is real. It’s serious,” Pielke said. “But that doesn’t mean anything goes when it comes to science.”
Prominent references to the billion-dollar disaster dataset make no mention of the role societal growth has played in the growing economic losses caused by extreme weather. One example came in September last year, when the Biden administration announced a summit focused on improving the country’s resilience to climate change.
“Across the country, Americans are experiencing the devastating impacts of climate change. In just the first eight months of the year, there have been 23 separate billion-dollar weather and climate disasters — more than any other year on record,” the administration said in a news release.
The White House did not answer detailed questions about how it reconciles its interpretation of the dataset with the consensus that global warming has played a relatively small role in rising disaster costs.
“Americans are already feeling the impacts of hurricanes, wildfires, floods, and drought. The science is clear: climate change is the existential threat of our time,” White House spokesman Angelo Fernández Hernández said in a statement. “President Biden and Vice President Harris have listened to and acted on the latest climate science.”
Other references to the dataset are more nuanced. In NOAA’s annual summaries and in the National Climate Assessment, the authors attribute the increase to a combination of societal growth and global warming.
The scientists behind those documents say that global warming has caused physical changes to the atmosphere that intensify extreme weather. For instance, as average temperatures rise, the potential for more water vapor in the atmosphere increases, which can lead to heavier rainfall and stronger storms because more moisture is available for precipitation.
They have bolstered this claim with an emerging statistical method called “event attribution.” Traditional detection and attribution research seeks to identify and explain trends in long-term data — decades or even centuries of temperature readings, ice measurements or hurricane observations.
An event attribution study, by contrast, examines how the likelihood or severity of one observed extreme weather event changed due to global warming. To do this, scientists use climate models to compare the event’s probabilities in a simulated world without greenhouse gas emissions against its probabilities in our real-life warmer world.
For instance, after Hurricane Harvey inundated Texas in 2017, Kerry Emanuel, an MIT climate scientist, reported in a peer-reviewed paper that the probability of so much rain in the Houston area had increased sixfold since 1970. Two independent studies reported similar findings.
“It’s a rare event by any standard. But that makes a big difference if you’re setting out to write the flood insurance policy for Houston,” said Emanuel, who serves on the board of two insurance companies, Plymouth Rock Home and Trusted Resource Underwriters. “If you rely on historical data that goes back 50 or 60 years, you’re going to vastly underestimate the risk.”
Some scientists doubt that the computer models used in event attribution studies can accurately simulate the probabilities of rare weather events. They also contend that event attribution skips the crucial first step of detecting a trend. Analyses of a century’s worth of Atlantic hurricanes have failed to find statistically significant departures from the historical range of natural variability, according to NOAA’s summary of the research.
Other scientists are enthusiastic about event attribution. They say that detection and attribution studies are ill-suited to capture relatively small but still meaningful changes in the behavior of rare events like hurricanes. By the time detection and attribution studies confidently find a trend, they say, society will have long since crossed into dangerous territory.
In event attribution studies of Hurricanes Helene and Milton, researchers at Imperial College London calculated that global warming nearly doubled the storms’ cost in Florida because their wind would have been about 10 percent weaker in a climate unaltered by greenhouse gas emissions. A separate study — which, like Imperial College’s analysis, has not been peer reviewed — found that Helene’s rainfall was about 10 percent stronger than it would have been without global warming.
Of 504 extreme weather events analyzed in this type of attribution study, 71 percent were made more likely or severe due to global warming, while 9 percent were made less likely or severe, according to a 2022 analysis by the the U.K.-based news and analysis website Carbon Brief. Global warming had no effect on the likelihood or severity of the other 20 percent.
“We know that climate change did have something to do with the increase in the risk of particular extreme events, not all extreme events, but certain extreme events,” said Kate Marvel, a climate scientist at NASA’s Goddard Institute of Space Studies and lead author of the National Climate Assessment chapter that references the billion-dollar disasters.
“I think then the burden might be to demonstrate that detected climate influence therefore has no effect whatsoever on the associated costs,” Marvel added.
Disaster researchers counter that even when global warming made some extreme weather event more likely, that does not necessarily translate to more economic damage.
“Particularly for storms and landslides and floods and droughts, we simply cannot attribute the disaster to human-caused climate change, even when we can directly attribute what the hazard is doing,” said Ilan Kelman, professor of disasters and health at University College London and author of the book “Disaster by Choice: How Our Actions Turn Natural Hazards Into Catastrophes.”
Hurricane Katrina, for instance, was probably somewhat stronger than it would have been without global warming, but it made landfall as a Category 3, and “New Orleans should have been prepared for a Category 5,” Kelman said. Engineering flaws in the city’s protective levees resulted in flooding that submerged 80 percent of the city. More than 1,800 people were killed.
Although most Western wildfires are started by humans, global warming has amplified the hot and dry conditions that encourage their spread. At the same time, the number of housing units in the “wildland urban interface” — places at higher risk for wildfires — grew from 30.1 million to 44.1 million, a 47 percent increase, between 1990 and 2020, according to a Forest Service report.
“If we build in a burnable location, we’re going to get burned, and I cannot find any correlation between these fire characteristics and actual damage,” Kelman said.
Bouwer added that climate scientists should seek trends in meteorological data rather than economic damage, which is more dependent upon human behavior — including flood control measures. Economic damage from inland flooding is heavily influenced by the dams, dikes and canals humans build to manage rivers, he noted.
Rivers across Appalachia swelled during Hurricane Helene, but in some places, flood damage was mild compared with the catastrophe that unfolded in western North Carolina. On the French Broad River upstream of Knoxville, the Tennessee Valley Authority’s Douglas Lake dam held back 182 billion gallons of water, enough to fill about 275,000 Olympic-size swimming pools.
In the 1960s, the TVA proposed 14 flood control dams on the French Broad upstream of Asheville, but it scrapped the project in 1972 due to political opposition. It is impossible to know whether the unbuilt dams would have softened the impact of Helene’s remnants around Asheville, but the TVA estimates its dams have prevented about $9.7 billion in flood damage since 1936.
Yet some scientists believe that persuading people to take climate change seriously requires more than talking about fractions of degrees Celsius, inches of sea level or melting ice caps in parts of the world most people will never visit.
Margaret Garcia, one of Marvel’s co-authors on the National Climate Assessment, said in an email that they used the billion-dollar disaster dataset “to clearly communicate that climate change is a present-day problem here in the United States with significant impacts. We wanted to counter the common perception that climate change is a problem for the future and for someone else.”
Garcia, an assistant professor in the School of Sustainable Engineering and the Built Environment at Arizona State University, said the billion-dollar disaster dataset does not capture all the ways global warming impacts the economy because it includes only the most expensive disasters caused by a subset of extreme events.
For instance, the dataset does not include the houses in Rodanthe, N.C., that have crumbled into the ocean as a result of coastal erosion and sea level rise. Those losses are also influenced by societal factors like risky development and uneven real estate disclosures.
Kathie Dello, North Carolina’s state climatologist who co-authored a different chapter of the climate assessment that cited the billion-dollar disasters, said she prefers to focus on climatic data. But she added that she saw the logic in discussing disaster damage because most people care more about economic issues than climate change.
“Everyone always wants to know, what is this going to cost me?” Dello said. “There’s been a lot of pressure on climate scientists to tell the story better.”
Check my work
The data and code to produce the charts in this article can be found in computational notebooks: billion-dollar disasters, historical global average temperature anomaly and its causes, inflation-adjusted and normalized U.S. hurricane losses and factors influencing those losses. The data sources and analysis scripts to produce the data behind the hurricane charts can be found in this notebook. The data and code to calculate the change in population and inflation-adjusted wages of the counties hit by Milton’s hurricane-force winds can be found in this notebook.
To get in touch, email me or my editor, Monica Ulmanu. Photo editing by Sandra M. Stevenson. Copy editing by Frances Moody.