In September 1813, a major hurricane destroyed US gunboats and ships that were defending St Mary’s, Georgia, from the British. Fifty sailors drowned.
In a letter to the US secretary of the navy, the commodore of the naval task force wrote that a privateer named Saucy Jack had been deposited so high on the marshes that the sea must have risen nearly 6 metres above its low water mark. Other letters of the time accurately described the passage of the eye of the storm and the timing of events.
Such historical records are proving invaluable to researchers trying to understand long-term patterns in the frequency and intensity of the hurricanes that lash the US mainland. Last week, Isabel became the 15th hurricane to strike since 1990, feeding dire predictions that hurricanes are on the increase.
But a historical analysis completed six weeks ago, combined with sophisticated computer models and modern hurricane-monitoring technology, has confirmed there is no such trend. Instead, hurricanes in the Atlantic come and go in cycles lasting a few decades.
While that is good news for some, it is also a warning for coastal communities that may have seen no storms for a generation or more.
To forecast hurricane intensities, predict seasonal and climatic patterns of storms, and issue risk assessments that influence building regulations and house insurance premiums, scientists rely on a database of tropical storms and hurricanes maintained by the National Hurricane Center in Miami, Florida – part of the National Oceanic and Atmospheric Administration (NOAA).
But the database, which contains details of storms blowing across the Atlantic Ocean, the Gulf of Mexico and the Caribbean Sea, is full of both systematic and random errors, says Chris Landsea, the NOAA researcher who is leading the hurricane re-analysis project. “When this database was put together in the 1960s, they didn’t spend a lot of time getting details right on the track or the intensity [of hurricanes].”
The first efforts to rectify these errors were led by Jose Fernandez-Partagas, a researcher at the University of Miami. He worked in near-total obscurity, poring over newspaper reports and shipping logs to uncover previously undocumented tropical cyclones from the mid-1800s to the early 1900s.
After his death in 1997, Landsea and his colleagues took over the task, instituting a systematic search of historical records, including diaries, newspapers, letters, county records, and popular news magazines of the time, such as Niles’ Weekly Register. Such records provide data mostly about a storm’s location and time of landfall, and the extent of damage.
But crucial data about wind speeds and atmospheric pressure can be gleaned from shipping logs. Sailors used the systematic method known as the Beaufort scale to estimate wind speeds by looking at the impact of wind on their sails. And by the late 1800s, many ships were carrying barometers. From the pressure readings, researchers can estimate a storm’s maximum wind speed. “We are able to use methods that are common today for storms 100 years old,” says Landsea.
Another crucial aspect of hurricanes that hit land is the storm surge – the exceptionally high water levels on the coast that can cause flooding inland. Reports such as the location of the Saucy Jack high on the marshes allow researchers to estimate the storm surge values.
“Sometimes watermarks are available,” says Landsea’s colleague Al Sandrik of the NOAA’s National Weather Service in Jacksonville, Florida. “For example, we have pictures of the storm surge on Newcastle Street in Brunswick, Georgia, which, I believe, is the earliest picture of a storm surge ever taken.” That was on 2 October 1898.
Based on such observations, the researchers have made over 5000 changes to the hurricane database for the years 1851 to 1910. The new data shows a cyclical pattern of hurricane activity that repeats every few decades.
The 1850s to mid-1860s were quiet, followed by an intense period from the late 1860s to 1900, with five seasons of 10 or more hurricanes. The year 1886 was the busiest season on record in the US, with seven hurricanes hitting the coast. And after a lull, the storms raged again in the 1930s, 40s and 50s.
“It does confirm there are cycles of activity, rather than long-term trends towards more or stronger storms,” says Landsea. That database also reveals that states such as Georgia that were largely spared during the 20th century remain at risk.
“If we take a closer look at the 19th century, we had major landfalls in 1854, 1893, and 1898,” says Sandrik. “So this is significant for the people in the area.”
Article continues below
The project is also changing our thinking about recent hurricanes. For instance, hurricane Andrew, which devastated Florida in 1992, was thought to be a category 4 hurricane, defined as having a maximum wind speed of between 210 and 249 kilometres per hour at sea level.
To gauge the strength of Andrew, researchers assumed maximum wind speeds at sea level were 75 per cent those measured by planes flying through the storm at 3000 metres. But since 1997, researchers have been able to get more reliable measurements by parachuting a device called a dropwindsonde into the windiest parts of the hurricane.
The instrument is fitted with a GPS receiver and records pressure, temperature and wind data every five metres. Such measurements have revealed that the wind speed at sea level is actually 90 per cent of the speeds recorded at 3000 metres. So Andrew’s maximum wind speed has been raised from 230 to 264 kilometres per hour, making it one of a handful of category 5 hurricanes to have struck the US.
NOAA researchers are not the only ones re-analysing hurricanes. Kwok Fai Cheung of the University of Hawaii at Manoa, in Honolulu, has started using newly developed hurricane-forecasting models to simulate storms that have hit the coast of New England in the past 500 years.
Cheung uses historical records to get the location of each hurricane, its size as indicated by the damage it caused on the ground, and the storm surge values. He plugs the data into his computer, literally running the model backwards. “I adjust the central pressure, which is the intensity of the hurricane, until my calculated storm surge matches the recorded value,” says Cheung. That way, he is able to recreate the hurricane that caused the flooding.
These re-analysis projects confirm that hurricanes are not increasing in numbers and intensity. But that is no cause for complacency – in fact, quite the opposite.
“The multi-decadal cycle is reaching a point where we are going to be seeing an increase in hurricane activity,” says Sandrik. “People wanting to live and build in coastal areas need to take into account the fact that we are returning to a more active period.”