Hurricane Fabian, packing 135-mile-an-hour winds as it grazes the northern Leeward Islands and churns toward the Bahamas, is serving as a stark reminder that the 2003 Atlantic hurricane season has entered its peak period.
Now, federal and university forecasters say they could be poised for significant gains in their ability to give coastal residents, emergency planners, insurance companies, and others affected by tropical cyclones advanced seasonal and even monthly warnings of what an upcoming hurricane season is likely to hold.
The key, these scientists say, lies in what University of Maryland meteorology professor Eugenia Kalnay dubs the "genome project for the atmosphere"- a two-year effort undertaken in the late 1990s to take each day's raw weather data from 1948 to 1998 and run it through the latest weather-forecasting and analysis tools.
The results, researchers say, represent the most consistent picture yet of shifts in atmospheric conditions over a range of time periods. That consistency has allowed researchers to clearly spot global patterns that previous, smaller-scale studies had only hinted at and that could significantly improve seasonal forecasting of hurricanes.
The 50-year reanalysis comes from a joint effort of the National Centers for Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR). And it has yielded "extremely beautiful" data, says William Gray, a professor of atmospheric science at Colorado State University whose group has been issuing seasonal forecasts for 20 years. "We're going to see a lot of progress in seasonal prediction skill" as the reanalysis data are mined.
The reanalysis also appears to be paving the way toward shorter-term forecasts. Noting that tropical-cyclone activity can vary widely within a June 1 to Nov. 30 season, Dr. Gray's group has been devising methods to forecast such activity in the Atlantic during each of the peak months of August, September - and beginning this year - October.
For its part, the Climate Prediction Center (CPC) of the National Oceanic and Atmospheric Administration (NOAA) is a relative latecomer to the seasonal hurricane forecasting field. It began issuing its forecasts in 1998. While the CPC's methods and forecasts differ from Gray's in significant ways, CPC forecasters are gaining fresh insights from the NCEP/NCAR reanalysis on factors that can affect seasonal hurricane activity, notes Gerry Bell, lead forecaster for the agency's seasonal hurricane outlooks.
One of his group's major objectives, he says, has been to gain a better understanding of the atmospheric conditions that control the various time scales over which hurricane seasons become more or less intense than usual - especially time scales of multiple decades.
Indeed, researchers including Gray have noted that the Atlantic hurricane seasons appear to go through periods of several decades where seasonal activity is above or below normal. Since the mid-1990s, he notes, the Atlantic and Caribbean hurricane seasons appear to have moved into their stronger-than-normal phases.
After poring over the NCEP/NCAR data, for example, Dr. Bell and colleague Muthuvel Chelliah have identified what appears to be a 40- to 50-year cycle in convection and thunderstorm activity over three key regions - the central equatorial Pacific, the West African monsoon region, and tropical Central and South America. During the 1950s and '60s, the pattern shows greater-than-normal thunderstorm formation over the central equatorial Pacific and West African monsoon region but lower-than-normal thunderstorm activity over tropical Central and South America. On the downswing during the 1980s and '90s, the pattern reversed itself.
This apparent cycle can alter the course of the jet stream in the northern hemisphere and the strength and duration of El Niño and La Niña episodes. Thus, it can help set the stage for conditions that enhance or suppress Atlantic hurricane activity, the team suggests.
Other researchers had teased similar patterns out of data on a regional basis, Dr. Chelliah explains. But this study represents the first to go global and show that this "tropical multi-decadal mode" is the dominant factor in atmospheric variability seen so far in the data. The research has been accepted for publication in the American Meteorological Society's Journal of Climate.
Among other applications, the work is being used by Chris Landsea and colleagues at NOAA's Hurricane Research Division in a 150-year reanalysis they're conducting of Atlantic and Caribbean hurricanes for patterns in frequency and strength that can be useful in improving seasonal forecasts.
Given the quality of the NCEP/NCAR data, the CPC's Bell adds that it may be possible to tease out the effects of shorter-term atmospheric cycles. For example, he hopes to make forecasts a month ahead for the Atlantic and Caribbean based on the periodic 30- to 60-day swings in convection over a swath of the tropics from the Indian Ocean to the western tropical Pacific. Like its longer-term cousins, this cycle triggers changes on a monthly scale that can affect tropical cyclone intensity in the Atlantic and Caribbean regions.
"We'd like to isolate this climate pattern" in the data "and figure out its impact," Bell says.
For now, Gray's forecasts appear to be the only ones exhibiting more skill than relying merely on long-term seasonal averages to forecast a season. In an evaluation of the Gray team's work over a 20-year period, Brian Owens of Risk Management Solutions in Britain and Dr. Landsea (who earned his PhD under Gray) notes that Gray's forecasts are the only ones around long enough to evaluate. NOAA's forecasts, which represent the official government view, haven't been around long enough to apply the typical statistical tests of "skill," he notes in a paper published in February.
Still, armed with the NCEP/NCAR data, to which more-recent information is being added daily, both groups anticipate significant gains in seasonal hurricane forecasts as climatic history gives up its secrets.