When tropical cyclone Haiyan slammed into the Philippines late last week, residents of the city of Tacloban, which was devastated by the storm, said they were prepared for the winds but not the wall of water that swept through the city.
That deadly swell was Haiyan's storm surge. While high hurricane wind speeds are one cause of death and destruction, storm surges and flooding kill the most people during tropical cyclones.
Louisiana State University's Hal Needham was one of the U.S. hurricane researchers who closely watched Haiyan make landfall. For the past five years, he's been working to gather information about past storm surges that might help researchers improve their prediction of future ones.
Tacloban's surge was possibly the highest recorded in the Philippines, rivaling the 24-foot surge record set in 1897, Needham noted. But in that surge, the researcher saw more than a record. He saw a comparison.
Needham is a self-proclaimed "extreme weather" enthusiast, whose blog, Hurricane Hal's Storm Surge Blog, analyzes present and past cyclones. His ability to run down historic storm surges is similar to how a farmer can rattle off past bumper crops and famine years.
That encyclopedic knowledge, though, was gained in pursuit of a greater purpose. For the past five years, Needham has been mining government data, ancient newspaper clippings and academic journals for information on hurricane high-water marks.
The database he's created, called SURGEDAT, includes 560 storm surges since 1880. It's the world's most comprehensive storm surge database, with historic storm surges from the U.S. and around the world. The project is part of the National Oceanic and Atmospheric Administration-funded Southern Climate Impacts Planning Program, where researchers at Louisiana State and the University of Oklahoma conduct research on climate impacts in the south-central United States.
Parallels with the past
In Tacloban's storm surge, Needham immediately saw a parallel with the Galveston, Texas, hurricane of 1900, which was the deadliest weather disaster in U.S. history. "The reason why the water came in so quickly in the Philippines is very similar to the reason why the water came in so quickly in Galveston," Needham explained.
Haiyan's storm track meant its eye passed close to Tacloban. At first, this led to strong winds pushing water out of San Pedro and San Pablo Bay. As the eye passed near Tacloban, however, the winds reversed, blowing water up the bay and into the city.
"As soon as the winds changed direction, you got this really high water rise really quickly," Needham said. While reports from Galveston in 1900 didn't exactly mention a wall of water, the water rise residents reported at the time was incredibly rapid, he noted.
That storm track, which brought the eye close to the Texas city, was also akin to Haiyan's path near Tacloban. Isaac Cline, Galveston's chief meteorologist at the time, stood on his porch as the flood came and estimated water rose 4 feet in four seconds.
"And in that way, that's very similar to what happened in the Philippines," Needham said.
Even though the Gulf Coast shoreline is structured differently from the Philippines', drawing parallels in such storms can be instructive for surge predictions, Needham said. His recent research, which is currently in press, bears this out.
He took SURGEDAT's 130 years of storm surge and hurricane data for the Gulf of Mexico and ran a statistical analysis to see which hurricane conditions correlate with storm surge height. What he found was somewhat unexpected.
The factors that most strongly predicted storm surge height were the size and wind speed of the hurricane 18 hours before it made landfall. "The pre-landfall hurricane characteristics are very important," Needham said.
This finding rings true when looking at Hurricane Katrina, which had a devastating storm surge. While it was only a Category 3 hurricane when it made landfall, the wind speeds 18 hours before it hit were 175 miles per hour, Needham said.
He believes the connections found in this sort of data mining could help physicists learn more about how the physical characteristics of hurricanes relate to storm surge. "We are hoping that data-driven methods can work well with modeling. The beauty of modeling is that modeling can get a handle on [future] events."
Katrina, other past hurricanes spur research
It was the four hurricanes that made landfall in 2004, followed by Hurricane Dennis and then Katrina in 2005, that spurred additional research into improving storm surge predictions, said Scott Hagen, a professor of engineering at the University of Central Florida.
According to Hagen, a storm surge modeler, in order to develop a good hurricane storm surge model, researchers need an accurate understanding of ocean water depth, coastal topography and the characteristics of the land the water will flow over, since that can alter water and wind speed.
"So being able to distinguish between a sandy beach and a marshy area versus a mangrove swamp versus urban areas," Hagen said.
Advances in lidar technology have allowed coastal storm surge models to have better inputs in the categories of topography and land characteristics, Hagen said. "We now use lidar fundamentally in developing our topographic models."
Ocean depth remains an area of greater uncertainty, Hagen said, as mapping the seafloor is costly and difficult. But unmanned ocean vehicles like undersea gliders have the potential to improve seafloor mapping in the near future.
Yet even with good data in all those realms, entered into a good model, "you are still missing the most fundamental input, and that is the wind field itself," Hagen said.
Hermann Fritz, a professor of civil engineering at the Georgia Institute of Technology, noted that Katrina, which hit locations at the same angle as the 1969 Category 5 Hurricane Camille, had higher storm surge water marks than Camille, because its winds were more sustained farther out.
"It's very critical how large an area the hurricane is going to essentially blow towards the shoreline over shallow water," Fritz said. After Katrina, researchers worked to re-create the wind fields for that storm.
Their analysis found that even though the storm had decreased in intensity from a Category 5 to a Category 3, its wind field expanded. This meant it was able to push a lot of water onto land, leading to the storm surges that devastated New Orleans.
That effort, which Hagen called "unprecedented," advanced the science of storm surge modeling. "We benefited tremendously from having access to the Hurricane Katrina wind fields."
Yet even though knowledge of wind fields has greatly improved in recent years, that's still an active area of research with room for improvement, Hagen said.
Getting out ahead of storms
Andrew Kennedy, an engineer at the University of Notre Dame, is a storm chaser of a sort. He goes out in advance of storms, scrambles along the coast and attaches his handmade pressure sensors, encased in PVC pipe, to anything that seems solid.
Sometimes, when he can afford it, he uses helicopters to lower instruments into the ocean. Then he waits for the hurricane to hit.
What his research team is trying to do, Kennedy said, is measure storm surge and wave height. The U.S. Geological Survey is also involved in similar work, deploying a mobile storm surge monitoring network in advance of storms.
Because the researchers have to put their instruments out in advance of hurricanes, they only get good data for about one out of every three storms, Kennedy said. The rest of the time, the storm might shift its track or weaken. "You make your guesses as to whether it will make landfall."
After collecting the instruments that survived the storm, the researcher feeds his measurements of surge and wave height to Notre Dame storm surge modeler Joannes Westerink.
In the case of Hurricane Ike, which hit the Texas coast in 2008, Kennedy's data showed the storm surge models underestimated the surge. "They had the bottom friction wrong," Kennedy said. "The measurements helped to show what happened in great detail, and that helped improve the model predictions."
Translation to the developing world
Yet while the U.S., spurred by Katrina and other significant hurricanes that hit land, is steadily improving the science behind storm surge prediction, it remains to be seen whether less developed countries such as the Philippines will be able to capitalize on such advances.
"The unfortunate thing is there aren't always the resources there to do those types of studies," Hagen said. When storms such as Haiyan hit, he said, a positive result could be that international organizations fund research on storm surge and cyclones in the developing world.
Yet the biggest advances in saving lives may come down to improved warning and evacuation systems, said Georgia Tech's Fritz. He pointed to the recent cyclone Phailin, which went ashore in India in mid-October.
Close to a million people were evacuated there, and because of that, the death toll was very low.
"That's the thing with hurricanes or cyclones or typhoons. They are announced," Fritz said. "You do have time on the order of several hours to days to get the people out."