Hunting Prehistoric Hurricanes

Storm-tossed sand offers a record of ancient cyclones

The transformation was stunning: One moment a nondescript tropical storm, the next, a hurricane of an intensity no American alive had ever experienced. The storm did not grow through some gradual accretion of power; it exploded forth like something escaping from a cage. The Weather Bureau of 1900 had a code word for winds of 150 miles an hour—extremebut no one in the bureau seriously expected to use it.

Galveston, Texas, after the 1900 hurricane that killed 20 percent of the city’s inhabitants. NOAA

A strip of sand separates Alabama’s Lake Shelby from the Gulf of Mexico. Fearn and Liu

Liu, Fearn, and a colleague draw a core from the lake. Fearn and Liu

Sediment cores taken from coastal lakes and marshes contain sand layers deposited by catastrophic ancient hurricanes. This chart documents the frequency of such storms at four sites. Each black line represents a sand layer. Liu

Erik Larson, Isaac’s Storm
(1999, Crown)

On September 9, 1900, the impossible destroyed Galveston, Texas. A fierce hurricane roared over the thriving coastal city that day, flooding its island with water and claiming more than 6,000 lives, about 1 in every 5 residents. The storm arrived with almost no warning. The U.S. Weather Bureau had ignored forecasts of Cuban meteorologists, noting that a severe hurricane had never before hit the town.

Judged simply on its strength, the hurricane that leveled Galveston a century ago was indeed a rare phenomenon. Meteorologists today classify it as a category 4 storm—one with sustained winds of 131 to 155 miles per hour. Few of those monsters ever arise in the Atlantic Ocean’s hurricane breeding grounds, let alone smash into the U.S. mainland. In the past 150 years, fewer than a dozen have struck the U.S. coast along the Gulf of Mexico or the Atlantic.

The most catastrophic hurricanes, known as category 5, are even more uncommon. Just two have run into the United States in the past century. In 1935, on Labor Day, one flattened the Florida Keys. In 1969, Hurricane Camille roared through Mississippi.

The infrequency of severe hurricanes is welcome news, of course. Yet it also poses a problem. Reliable data on hurricane landfalls in the United States is available only for the past 150 years. And given the small number of category 4 and 5 storms during that relatively short time span, scientists don’t have enough statistical power to estimate confidently how frequently catastrophic hurricanes strike the U.S. coastline.

So, to better look forward, investigators have decided to look further back in time. As part of a fledgling discipline called paleotempestology, they’ve begun to search for signs of hurricanes that predate recorded history.

At the forefront of this effort is Kam-biu Liu of Louisiana State University in Baton Rouge. By unearthing sand layers deposited by massive hurricanes in coastal lakes and marshes, his research group has identified storms that have struck the U.S. coast over the past 5,000 years. In February, Liu described his results at the annual meeting of the American Association for the Advancement of Science in Washington, D.C.

“It’s the first time we’ve been able to peer back before the historical record to see how hurricanes vary in time,” says Kerry A. Emanuel of the Massachusetts Institute of Technology, who would like to use such data to test whether the anticipated global warming will increase the number of severe hurricanes.

Scientists aren’t alone in taking an interest in paleotempestology. Most of the field’s funding comes from the Risk Prediction Initiative, an effort bankrolled by insurance companies in need of better data with which to predict the odds of a severe hurricane landfall in a specific region. Considering that category 4 and 5 hurricanes can cause billions of dollars in damage, the future of these insurance companies may rest on the accuracy of their estimates.

Paleotempestology “is a nice scientific challenge, but it’s [also] got a very practical outcome,” notes Thompson Webb III of Brown University in Providence, R.I., who has conducted work similar to Liu’s.

A real killer

When the category 4 hurricane ripped through Galveston in 1900, wind and rain alone produced significant damage and some loss of life. But as in many such tempests, the real killer was the flooding by the storm surge. Hurricane winds blowing over shallows near a coastline can raise up a dome of salt water 50 to 100 miles across. This storm surge can send up to 25 feet of water into the region where a hurricane makes landfall.

If a lake or marsh sits not far from the coast, the storm surge may also leave an enduring imprint of the hurricane. Sand from the ocean floor or beach can be thrown inland with the water, eventually settling to the bottom of the lakes or marshes in a discernible sediment layer that records the storm’s impact.

In 1990, Liu and his graduate student Miriam L. Fearn began to look for such sand layers at Lake Shelby, just off the Alabama coast (SN: 9/18/93, p. 191). They drilled into the lake bottom, removing cores of sediment ranging from 1 to 10 meters deep.

A category 3 hurricane called Frederic had run through that region in 1979. In cores taken on the edge of the lake nearest the coast, the two investigators found a surface stratum of sand, which they attributed to Frederic. Cores taken in the middle of the lake didn’t contain this recent layer, indicating that the hurricane wasn’t fierce enough to send sand that far into the lake. Those cores did, however, hold 11 other sand layers, 0.1 to 1 centimeter thick, which Liu and Fearn concluded could only have resulted from the storm surges of earlier category 4 or 5 hurricanes.

“There are no other high-energy events that would cause sand to be transported out into the middle of a lake,” contends Liu.

Through dating methods such as radiocarbon analysis of organic matter within the cores, the scientists were able to determine that nine of the lake’s sand layers originated between 2,200 and 3,300 years ago. The two more recent millennia each had one sand layer.

Over the past decade, Liu and his colleagues have drilled similar cores in 16 lakes and marshes along the Gulf Coast from Texas to Florida. For four of the sites—Pearl River Marsh in Louisiana, Pascagoula Marsh in Mississippi, Lake Shelby in Alabama, and Western Lake in Florida—the investigators obtained sediment cores going back about 5,000 years.

The historical record of the past 150 years clearly indicates that total hurricane activity along the U.S. coast varies decade by decade. When Liu’s team lined up their cores from the four sites, they realized that catastrophic storm activity rises and falls over longer periods, as well.

“There are millennial-scale variations in hurricane activity. Our data suggest that there are much longer cycles superimposed on the decadal cycles,” says Liu. “We’ve had a quiet period, an active period, and for the past 1,000 years, we’re back to a relatively quiet period.”

Indeed, the data indicate that catastrophic hurricanes struck the Gulf Coast much more frequently 1,000 to 3,500 years ago than they do now. During that hyperactive period, such storms hit the area from four to five times more often than they have in the past 1,000 years.

Combining the data on the past 3,500 years from several of their sites, the investigators conclude that a category 4 or 5 hurricane batters the Gulf Coast every 300 or so years. In other words, there’s about a 0.3 percent chance each year that the region will see a storm like the one that razed Galveston in 1900—or something even stronger.

This is the first time that scientists can provide insurance companies with an estimate of severe hurricane frequency that has any compelling data to back it up, says Liu.

There was some initial skepticism that the sand layers observed in Liu’s coastal lake cores were truly from storm surges caused by ancient hurricanes. As his team has studied additional sites, other scientists have grown more comfortable with the strategy.

“I feel pretty confident that they’re seeing a storm record. There’s no other way to explain how you get a big sand layer,” says David L. Malmquist of the Bermuda Biological Station for Research in St. George’s, who heads the Risk Prediction Initiative.

It also helps that another research team has tried Liu’s strategy and is having similar success. During the past few years, Webb and Jeff Donnelly, also at Brown University, have led an effort to draw cores from coastal salt marshes in Rhode Island, New Jersey, Massachusetts, and Connecticut. While their cores push only 600 to 1,000 years into the past, the scientists have matched many of their sand layers to hurricanes documented in the historical record.

“It was key to prove the method with known storms,” says Webb.

Climatic changes

Paleotempestology may help insurance companies better set their rates, but scientists also desire the field’s data to learn how climate influences hurricane activity. For example, what climatic changes could have caused the millennial shifts in hurricane activity along the Gulf Coast revealed by the cores?

Liu hypothesizes that the shifting position of the Bermuda high explains the change in landfall frequency along the Gulf. A region of high atmospheric pressure in which air circulates clockwise, the Bermuda high has a strong influence on the climate of North America.

Scientists, for example, have attributed changing precipitation patterns in the United States over the past 10,000 years to shifts in the Bermuda high’s position. The central and the eastern parts of the country seesaw back and forth in receiving the greater amount of rain.

Liu suggests that his ancient hurricanes also follow an alternating pattern based on the Bermuda high. “If the high-pressure system is situated close to the continent and the Caribbean, then hurricanes tend to get steered to the Gulf Coast,” explains Liu. If the Bermuda high sits far from North America, hurricanes instead usually head for the Atlantic coastline.

He’s now compiling coring data from sites in Virginia and Massachusetts to determine if ancient hurricane activity rises along the Atlantic Coast when it declines in the Gulf.

“The data so far show potential for supporting the hypothesis, but we need to do much more work,” says Liu.

Documenting ancient storms

Beyond looking for sand layers caused by storm surges, scientists are investigating novel ways to document ancient hurricanes. Liu, for example, has traveled across the Pacific Ocean—where hurricanes are usually called typhoons or cyclones—in an attempt to extend the historical record.

“In China, there are very good records kept at the county level. County gazettes record every significant cultural and natural event in the area,” he notes.

In a pilot project using the records of a province near Hong Kong, Liu has compiled a 1,000-year record of typhoon landfalls. It reveals variations in hurricane frequency on the scale of centuries as well as decades.

That success has prompted the National Science Foundation to fund a project that will explore the gazettes from provinces all along the China coast.

“This will be the longest tropical cyclone record in the world,” says Liu, who further hopes to take cores from coastal lakes in China to corroborate the hurricanes recorded in the gazettes.

There are also options other than sand layers and paper records. Scientists are considering coral, tree rings, pollen, caves, and even clam shells as indicators of cyclones.

A research team led by Bjorn Malmgren at the University of Göteborg in Sweden has begun to study the humic-acid content of coral from sites in the Caribbean. Humic acids in soil run off into the ocean during the heavy rainfalls of severe hurricanes, and corals incorporate the acids into their skeletons. Malmgren says that by measuring how the humic-acid content of corals varies with time, his group can develop a record of ancient tropical cyclones.

The fierce winds of major hurricanes can also shear off the crowns of tall trees, reducing their photosynthetic ability. The resulting reduction in growth rate shows up as thin tree rings.

Similarly, the destruction caused by catastrophic hurricanes can change the vegetation within a region for years, altering the kinds and quantities of pollen trapped in sediments of lakes, ponds, and marshes. Other phenomena that affect the pollen record, such as fires or human deforestation, complicate its use in identifying prehistoric hurricanes, however.

One of the more unusual, and unexplained, features of a hurricane is that its rain is isotopically lighter than normal, with a lower concentration of oxygen 18 than typical rainwater has. James R. Lawrence of the University of Houston made this discovery in 1989 while studying groundwater after a Texas hurricane. He and other scientists propose that hurricane rainfall seeping into caves may be incorporated as detectable layers within deposits such as stalagmites or stalactites. Lawrence has also suggested studying clams to see if the carbonate in their shells holds an isotopic record of hurricane rain.

Despite these efforts, the search for signs of ancient hurricanes remains in its infancy. “One of the problems with paleotempestology is getting it jump-started,” says Emanuel. “There’s just a handful of people doing this.”

That’s in large part, he adds, because federal agencies haven’t embraced the interdisciplinary field, leaving the Risk Prediction Initiative as almost the only source of funding.

Yet to understand how a changing climate influences the number and intensity of hurricanes, scientists must turn to paleotempestology, argues Emanuel. He notes that predictions made by climate models on this issue don’t inspire confidence. One model suggests that global warming will increase hurricane frequency, while another foretells the exact opposite.

“I don’t think there’s any other way we’re going to get a handle on the relationship between hurricanes and climate except by looking back in time,” concludes Emanuel.

More Stories from Science News on Earth

From the Nature Index

Paid Content