133
FORTUNE.COM // NOVEMBER 2019
Re really cares about: a curve showing the likeli-
hood that, in each of a number of geographic
swathes, a hurricane strike will produce a given
amount of insured loss. The company uses these
“loss frequency curves” to calculate a premium
price that—according to the model—will both
cover the expected payout and deliver an accept-
able profit.
Swiss Re’s storm modeling today has the
same basic goal today that it did a generation
ago: “to build the spaghettis,” Gloor tells me, “in
a way that they reflect the current climate.”
What modelers now are reckoning with, how-
ever, is the likelihood that climate change will
produce a future very different from the past.
WISS RE has struggled for years to
adapt its modeling to a warming
world. Initially it built into its
models an assumption that climate
change would produce a 1% yearly
increase in storms. 1% yearly increase in the
frequency of European wind storms. Later it
stopped that practice, concluding the science
wasn’t clear enough to justify it. Instead, it
added a dial that factors in the way sea-surface
temperature impacts hurricane frequency, a
factor it still uses today. But the recent surge in
the cost of disasters, particularly hurricanes, has
underscored the need to find a better way.
Now Swiss Re is helping to fund, on the other side of the storm-
prone Atlantic, work on a new model it hopes really will be climate-
smart. In New York City’s Morningside Heights neighborhood, at
the applied-mathematics and applied-physics department of the
engineering school of Columbia University, a team led by a former
trumpeter and current atmospheric scientist named Adam Sobel
is working to refine it. Rather than extrapolating from storms that
occurred when times were cooler, they’re deploying vast computing
power to create “synthetic” storms that—so the modelers hope—will
better reflect the realities of an era that’s getting hot.
The synthetic-storm methodology that Sobel’s team uses was
pioneered by Kerry Emanuel, a hurricane expert at MIT. The
Columbia crew’s secret sauce is the model it has written, which
incorporates particular assumptions about the physics of how cli-
mate change will affect hurricanes. Rather than employing Swiss
Re’s current standard, a top-down approach that starts with the
paths of past hurricanes and, as Sobel puts it, uses historical data
to “just jiggle it a little bit,” the Columbia modelers use a bottom-
up method, starting with data on weather-related factors they
think are both relevant to hurricanes and likely to be influenced by
global warming. Among those factors: “wind shear,” which is the
variation in wind speed and direction at different altitudes; sea-
surface temperature; and the amount of moisture in the air. Using
that data, their algorithm calculates when a hurricane will form,
how it will move, and how intense it will be. Theoretically, this
more-forward-looking approach could be more accurate.
Theoretically.
So far, uncertainty about which assumptions are the right ones
is causing the model to disgorge contradictory conclusions. One
key question concerns the interplay between rising temperatures
and moisture. Because higher temperatures stuff more ocean
moisture into the air, scientists generally agree that climate
change will make hurricanes more intense. Many have come to
believe the wetter air will make hurricanes more frequent too.
But warmer temperatures don’t just increase the amount of
water in the air; they also increase the amount of water the air can
hold. And the Columbia scientists have found that their models
spit out differing results depending on which of two moisture met-
rics they feed the algorithms. Using one, “relative humidity,” which
is the ratio of the amount of water the air is holding to the amount
it is capable of holding, the models say climate change makes
hurricanes more frequent. But using another metric, the “satura-
tion deficit,” which is the absolute difference between the amount
of water the air is capable of holding and the amount it’s holding,
the models say climate change makes hurricanes less frequent.
Warmer oceans affect both those metrics—and leave scientists to
debate which matters more.
“We sort of convinced ourselves that saturation deficit was
better” as a storm indicator, Sobel tells me late one afternoon in
his windowless office, one of its white-painted cinder-block walls
hung with a bright abstract representation of a hurricane’s move-
ment that his mother painted for him. “But we don’t know that,”
he confesses. In a scientific paper now under review, he and his
colleagues admit that, according to their model, whether climate
change makes hurricanes more or less frequent depends on which
moisture metric a modeler chooses to emphasize. “We would almost
be embarrassed about putting this result out there,” Sobel says. “The
COAL, FIRED An AEP power plant in Winfield,
W.Va. Swiss Re stopped insuring AEP property
this year because of the utility’s reliance on coal.
LUKE SHARRETT
—BLOOMBERG VIA GET T Y IMAGES
