Storms born over water—hurricanes, cyclones, and typhoons—can be downright destructive when they make landfall. What’s more, theory has suggested that intense storms are more likely in a warmer climate. Now researchers have used a nearly 40-year record of storm observations to show that to be true. This finding is cause for concern, the team suggests, because the strongest storms cause disproportionate damage and mortality.
Upping the Speed Limit
Models have consistently demonstrated a link between a warmer climate and stronger storms. That makes sense, said James Kossin, an atmospheric scientist at the National Oceanic and Atmospheric Administration in Madison, Wis. A tropical storm’s potential intensity—its “speed limit”—is directly dictated, in part, by the difference between the temperature of the ocean’s surface and the temperature of the upper atmosphere.
To test that hypothesis using real data, Kossin and his colleagues collected infrared satellite imagery of over 4,000 storms observed between 1979 and 2017. Satellite imagery affords the most unbiased observations of storms, said Kerry Emanuel, an atmospheric scientist at the Massachusetts Institute of Technology (MIT) not involved in the research. “Satellites can see every single storm on the planet.”
So Many Data
The data set included more than just a single look at a storm: Some of the tempests were imaged at up to 133 different points in time. In total, Kossin and his collaborators had a lot of data on their hands—about 225,000 observations.
The first step was to ensure that the data were all of roughly equal quality—the images varied significantly in resolution because they were obtained by different instruments. To obtain a more homogeneous data set, Kossin and his colleagues resampled all of the observations to a spatial resolution of 8 kilometers.
The researchers then applied an automatic algorithm known as the Advanced Dvorak Technique to estimate wind speed from the images. This technique basically inputs parameters such as a storm’s structure (including its eye) and infrared intensity into a flowchart to estimate wind speed. “It’s not much more sophisticated than a trained forecaster looking at a photograph,” said MIT’s Emanuel.
Despite its indirect nature, the Advanced Dvorak Technique has been widely used to estimate wind speed from storm images, said Kossin.“It sounds very coarse and rudimentary, but it’s stood the test of time.”
The team recovered wind speeds ranging from 25 to 170 knots (roughly 46–315 kilometers per hour). The scientists then grouped the observations according to the Saffir-Simpson scale, a commonly used metric for categorizing storms based on wind speed. The Saffir-Simpson scale ranges from category 1 (wind speeds between 119 and 153 kilometers per hour) to category 5 (wind speeds above 252 kilometers per hour).
More Common “Monsters”
Kossin and his colleagues limited their analysis to images of categories 1–5 storms. They calculated the proportion of images classified as category 3, 4, or 5 and found that these “major storms” increased in prevalence as time went on: Over the 39-year span of the data set, the proportion of major storms increased by 25% (roughly 6% per decade).
The researchers also divided their data in two based on when the images were obtained. When they compared these “early half” and “late half” subsamples, Kossin and his collaborators found that more recent storms were about 15% more likely to be major storms.
These consistent findings are troubling, said Kossin, because intense storms are the ones that pack the biggest punch. “The major hurricanes by far dominate damages and mortality,” he said. “They’re the monsters.”
It’s difficult to pin down precisely what’s driving this evolution. Natural variability in storm strength might well play a role, the authors noted, but the data’s consistency with simulations of greenhouse warming is telling. “There is a likely human fingerprint on this increase,” the researchers concluded in their study, which was published this week in the Proceedings of the National Academy of Sciences of the United States of America.
In the future, Kossin and his colleagues plan to explore other ways to estimate wind speed from satellite imagery. “We have enough data to start applying machine learning algorithms,” he said. “It’s ultimately a pattern-recognition problem.”
—Katherine Kornei (@KatherineKornei), Science Writer