To find out what makes a hurricane tick, you have to fly through it. To find out how often hurricanes strike land, you need to dig for them.
As more condos, office towers, and housing tracts spring up in cities along America's East and Gulf coasts, researchers are scanning mud they bring up from coastal marshes and ponds for clues about the history of major hurricanes.
The approach is relatively new, scientists say. With it, they hope to give residents, emergency planners, building and zoning boards, and insurers a better handle on the risk densely populated areas face from these storms. When combined with other emerging techniques, these "paleo-tempestologists" add, information they glean might help determine whether global warming could generate more- frequent or more-intense storms.
"The motivation is to extend the historical record" for major hurricanes, says Richard Murnane, who heads the Risk Prediction Initiative at the Bermuda Biological Station for Research.
The insurance industry is interested in the most intense storms, he says, but these are rare. Only three Category 5 storms - the most destructive - have hit the US in the past century. The latest was Hurricane Andrew, which struck Florida and Louisiana in 1992, killing 23 people and causing $26.5 billion in damage.
The potential of mud coring became apparent in 1993, when Kam-biu Liu and Miriam Fearn at Louisiana State University showed that paleo-hurricanes left their mark as sand layers in mud cores taken from the bottom of coastal lakes along the Gulf. But could the technique work for hurricanes along the East Coast?
"Initially, I was somewhat dubious," says Jeff Donnelly, a geologist at the Woods Hole Oceanographic Institution. He had spotted sand layers in mud cores he pulled from coastal marshes to study effects of changing sea levels on ancient ecosystems. The marshes sat behind barrier beaches, which can be overswept by tidal surges during storms. The surges drive sand into the marshes. But when powerful nor'easters clobber the East Coast they can generate surges of their own. Telling the two apart might be tough.
Despite his misgivings, Donnelly and his colleagues picked sites where winds from nor'easters came from the wrong direction to add sand to marshes. They punched cores from the marshes, dated the organic material, then compared the results with the historical record. "The only thing these sites were recording in the historic interval was hurricanes," he says. "I was amazed at how well it came out."
Dr. Donnelly published the results of his first hurricane-related study, based on cores pulled on Succotash Marsh in Rhode Island in 2001. He's also looking at sites along the Connecticut and New Jersey coasts and in wetlands near New York City. There, a hurricane in 1821 sent a 10- to 13-foot storm surge over Manhattan.
Yet as successful as coring has been, it can only yield dates of events to within a decade or two. When historical records aren't available as a back-up, a sand layer could appear to be a hurricane but actually mask several storms.
Woods Hole geochemist Anne Cohen is looking at corals to solve the problem. Coral layers can be dated within months and their chemical composition reflects that of the water. Rainfall from hurricanes carries a chemical signature. So by examining coral growth-bands, researchers can pin down dates for hurricanes during the broader periods that coring unveils. In 2000, Dr. Cohen tested the technique in the Virgin Islands and found records of several hurricanes written in coral. Cohen and Donnelly hope to study cores and coral samples from Puerto Rico next year.
"We really want to know how hurricane frequency and intensity will change with global warming," Cohen says. "Right now, the scientific community is divided."