Data Mining Reveals Sudden Climate Shifts
One of the frustrations of being a scientist who writes is that most of the good stories about science cannot be told — except to other scientists. Either they involve concepts that most people have not studied — say, isotope ratios or electricity — or they don’t lend themselves to a good narrative that will help readers remember the facts a week later. In my own writing experience over the last forty years, I estimate that fewer than 10 percent of what a scientist would consider a good story can be translated into a good narrative for general readers and policymakers.
However, here is one of the exceptions, my armchair discovery story about extreme weather shifts between 1991 and 2011. It’s the tale I’d tell to a general audience if someone asked me why I went looking for abrupt climate shifts.
First of all, you need to understand where I come from. Though nominally a neurophysiologist working on the brain’s circuitry and evolution, I wrote the first major magazine article on abrupt climate change back in 1998, a cover story for The Atlantic that the editors called “The Great Climate Flip- flop.”
Let’s start the narrative with augmenting two aspects that everyone knows by now, that the world has been warming since 1977, melting land ice so that sea level is rising. Note that since 1984, land has been warming much faster than the sea surface temperature.
Note, too, that sea level is now rising about twice as fast as it did from 1930–1992, and five times faster than a hundred years ago.
This data mining venture began when I stumbled upon NOAA’s project, now 20 years old, that is keeping track of episodes of extreme weather, ones that are big enough to cause a billion dollars in damage.
Going back to 1980, they collected data on the number of hurricanes, inland floods, drought & heat waves, severe windstorms (tornado, hail, derecho), wildfires, crop freeze events and big winter storms. That billion-dollar threshold involved a lot of data mining from insurance reports on claims, then correcting for inflation before seeing if they exceeded a billion dollars. Forty years of extreme weather data is a lot.
When they display the data as a stacked bar chart, it is difficult to see the time course of the various types of extreme weather. But you can plot each of them separately because the stacked bar chart is actually an interactive figure, one that allows you to plot each type of extreme weather separately.
Unchecking all of the other colored boxes allows us to look at the number of billion-dollar hurricanes (elsewhere known as typhoons and tropical cyclones, or TCs) coming ashore. There is data showing the biggest hurricanes are changing, but it doesn’t show up in this plot. For reasons I will explain later, I was browsing to find some really big changes, such as a three-fold increase that was sustained.
Next I plotted billion-$ droughts separately. There was never more than one each year; the only obvious trend is fewer years with zero episodes. Again, this type of extreme weather fails my arbitrary triple-or-worse-and-sustained criterion for being interesting.
How about those out-of-season freezes that fruit growers dread? They not only freeze half-grown fruit, killing additional growth, but they sometimes kill the fruit trees too — and thus the next dozen crops, while replacement trees reach fruit-growing maturity. But for this type of extreme weather, one cannot even make a case for more of them over the years.
So far, it’s a strikeout: no examples of big changes that would explain that uphill trend in the 21st century seen in the stacked-bars illustration.
Big wildfire episodes began in 1991 but have not exceeded one per year, with fewer skipped years in the 21st century. But one could argue that something happened after 2000.
One can also see a 2.3X jump in 2000 for total acres burned each year.
The costs of fighting forest fires took an even cleaner jump to 3X in 2000, increasing thereafter to 4X. Perhaps NOAA’s billion-dollar threshold for inclusion was so high that it obscured a jump.
So, wildfires had a sustained shift. Neither temperature nor drought took a leap that year, so perhaps it was the wind aspect of fire weather that changed.
The average annual number of billion-$ inland floods stayed at about 0.5 from 1980–2009, what with all of the skipped years. Then in 2010, it jumped up to averaging two a year, a 4X increase. Definitely a keeper, given my criterion for a maintained jump of at least 3X.
The other keeper is the annual number of billion-$-plus severe storms, a NOAA category that covers clusters of inland tornados, derechos, and hailstorms. The baseline, from 1980 through 2002, is one per year. Beginning with 2011, it steps up to an average 7X — after 2015, there are four years of 8X.
We now have three examples of a 3X-or-more step up in types of extreme weather in continental US data. And there are two more, from other sources than the NOAA analysis, where event duration increases in a big way.
In the last half of July, 2003, there was a heat wave in Europe that killed 70,000 people. How does that compare?
In Chicago in 1995, there were over 700 heat wave fatalities within a week.
The 1988 heatwave in the Midwest and Canada killed perhaps 10,000 — but that was over a 16-week hot spell that began in May.
So, the 2003 heatwave was several orders of magnitude more deadly than big 20th-century heatwaves. Someone called it a mega heat wave — and it stuck.
It broke new ground, certainly, but a single odd event is not a sustained shift. Indeed, there was talk that it was one of those rare, chance-in-a-million outlier events that we were unlikely to ever see again in our lifetimes. Comforting words.
Seven years later, there was a second mega, located several time zones farther east in Russia, mostly south of Moscow. It killed about 56,000 people over several weeks. I decided not to wait for a third, and so added megas to my sudden shift list, redefining my criterion to 3X in either severity or recurrence rate.
There is likely to be an American Mega in our future, and we are not preparing for it, just as we were unprepared for the 2020 pandemic.
What makes megas so bad is not just their peak daytime temperature; more important for heat stroke fatalities is how hot it stays at night — and for how many successive nights one stays awake fanning oneself and drinking water. After several sleepless nights in a row, bad things start to happen.
Longer duration is also the key to the fifth shift, stalled hurricanes.
Hurricane wind maximums are the usual measure of “strength.” But even a minor hurricane can inflict major damage if it sticks around too long. Someone coined the phrase, “The hurricane that didn’t leave.” An early example was the 1991 “Perfect Storm” which hung around for five days off the east coast.
What is the normal amount of time that it takes a hurricane to pass over a coastal city? The speed of advance along the average ground track is about 11 mph. If the diameter of the storm is 55 miles, then start to finish inside the city is about 5 hours. (A half-century ago, it used to be about 4 hours.)
So how long did the stalled hurricane named 2017 Harvey batter the Gulf Coast near Houston? Five days, 24X longer than usual. And it dumped a year’s worth of rain in the first four days. More than $126 billion in economic losses were attributed to Harvey.
Much of that damage was because Harvey reversed course, but there has also been a more general slowdown in ground speed since mid-century. Tropical cyclones over land have slowed down 20 percent in the Atlantic, 30 percent in the western North Pacific, and 19 percent in the Australian region. Ground speed over the continental USA is down 17% over the longer data set from 1900 to 2017. But people will argue with you endlessly about significance when claimed changes are that small, one reason why, in my survey, I was looking for steps of 200% or more.
Most of this exploration on the web, you may have noticed, could have been done by anyone interested enough to do some data mining; many high school students would qualify. Being a scientist only helped me to frame the search and to avoid some beginners’ mistakes.
And, as I mentioned earlier, I also had a lot of motivation. I’d been following the abrupt climate shift story since 1983, simply because of my neuroscientist’s fascination with why the human brain enlarged 3X during the climate fluctuations of the last ice age, starting about 2.3 million years ago. In species evolution, nothing that big happens that fast without a big boost from repeated climate change.
So I started paying attention and got to know the climate science researchers here at the University of Washington. I’d had a year as a graduate teaching assistant in physics, plus a Ph.D. in physiology and biophysics, so I could usually understand what the oceanographers and atmospheric scientists were talking about, even when they went off into Coriolis effects and such.
Brain enlargement is another good narrative with a vocabulary suitable for general readers, but I’ll have to save it for another day.
What caused the sudden shifts? Is there a common cause for all five?
That is not established yet, but the middle of my Extreme Weather book addresses the promising jet stream possibilities that the atmospheric scientists are coming up with. But a cause common to them all doesn’t rule out an individual type from having an even more important cause. We’ll see, as the basic science gets fleshed out.
You forgot to tell us why you were looking around for sudden shifts.
I had been following the paleoclimate research, which showed quite abrupt shifts in climate, where things went from a cool-and-dry climate to a warm-and-wet climate in just five years. It stayed that way for centuries — a sustained shift — and then flipped back into cool-and-dry in about fifty years. There are now dozens of examples of this instability of climate. That was my mental model for a sustained shift.
There are many aspects of climate, but the only two that researchers can infer from tree rings and ice core layers are changes in temperature and precipitation rate (there are some hints of high winds and waves which are also preserved in the ice, such as dust particles carried all of the way from East Asia to Greenland). Nothing like that sudden step has happened to us in the last 11,000 years, just a little glitch at 8,200 years ago from a Labrador ice dam breaking.
The agricultural era has been a very quiet period for climate change. We are about to destabilize things and, so far, we lack the leadership to even reinstitute the Pharaoh’s seven-year grain storage plan.
It doesn’t have to involve temperature and moisture to be an abrupt climate shift. I had started wondering whether other aspects of climate, such as the afore-mentioned types of extreme weather, might flip without temp and precip flipping at the same time. And 2011 is sure looking like a good year to tag it with; it has taken a decade to make sure that the step was maintained, worthy of being a good candidate for an abrupt climate shift.
The climate scientists have not yet spoken on such a shift, so treat it as only a candidate, but by 2011 the data suggests five sustained steps: 3X, 4X, 7X, 24X, and about 100X. One would be bad enough. We have a problem.
William H. Calvin, Ph.D., is a professor emeritus at the University of Washington School of Medicine in Seattle, and the president of CO2Foundation.org. This piece is adapted from the preface to his 17th book, “Extreme Weather and What to Do About It.”
