Weather roulette: why good forecasting and a bit of luck avoided a hurricane catastrophe in Florida
Now that Hurricane Irma is dead, the post-mortem has begun. How bad was the storm? How much worse could it have been? And how accurate was the forecast? The answers might surprise you.
MIAMI, Fla. - After a week when Florida residents found themselves running from one end of the state to the other to avoid Hurricane Irma, some have criticized the meteorologists - and the media- for leading them on a merry dance.
But in fact the forecast was remarkably accurate, possibly one of the best projections ever, meteorologists say.
And while Irma caused as much as much as $62 billion in damage in Florida, it could have been more than three times that amount, experts add.
"The average error in a two-day forecast by the U.S. National Hurricane Center (NHC) is 90 miles, and 123 miles for three days. For Irma, it was only 59 miles two days out and 87 miles three days out," according to Jeff Masters, meteorology director at the private forecaster Weather Underground. "That's 30-35 percent better than the NHC's five-year average," he added, noting it represented huge improvement over Miami's famous 1992 storm Hurricane Andrew.
Forecasters attribute this to better modeling, thanks to advances in data gathering and new supercomputers able to crunch massive amounts of weather data.
Forecasters these days also have the benefit of the multiplicity of computer models from half a dozen sources which they factor into a consensus forecasting approach that can sometimes look like spaghetti.
In the case of Irma, a European model known as the ECMWF, (European Centre for Medium-Range Weather Forecasts), has consistently proven to be more accurate than the most trusted of the American models, known as the GFS, the Global Forecasting System, run by the National Oceanic and Atmospheric Administration (NOAA).
"The European model identified Florida as the target on Aug 31, while NOAA's own model only reached that conclusion four days later, on Sept 3, a week before it made landfall," said Judith Curry, president of Nevada-based Climate Forecast Applications Network (CFAN), a weather and climate services company.
Starting in 2009, NOAA began a project to improve hurricane forecasts by 50 percent by 2019 leading to significant advances in real-time observations of atmospheric and oceanic conditions, including the use of more sophisticated satellites and sensors dropped by parachute from special Hurricane Hunter planes into the storm, called 'dropsondes,' that collect vital thermodynamic data, such as pressure, temperature, wind strength and humidity.
In the past, forecasters relied heavily on statistical models of historical data. "That's fallen by the wayside with the new technology," Masters said.
Satellites and Supercomputers
In 2016, new satellites known as GOES-R began to feed back images more quickly and with four times the resolution of earlier generations. The GOES-R system transmits infrared thermal imagery to detect sea temperature, which helps determine the intensity of storms, which draw energy from the water’s heat.
NOAA and NASA are also experimenting with an unmanned aircraft, Global Hawk, to provide better weather readings, while drones are now being used to get closer to the interior core of a storm than traditional Hurricane Hunter planes can.
Congress also funded an upgrade of the National Weather Service’s IBM supercomputer system after Superstorm Sandy in 2012, bolstering capacity to crunch data by 25 times.
However, even with this technology, forecasting extreme weather like hurricanes still leaves a lot to chance.
"It's really weather roulette. There are limits to how precise we can be," said Curry.
"The job of a meteorologist is thankless. People want to know if their houses will be destroyed and science cannot predict that," said Univision's Channel 23 weatherman in Miami, Eduardo Rodríguez, who has a sign on his desk that reads "Science is never settled."
The atmosphere "is a chaotic system," added Albert Martínez, Univision's chief meteorologist. The chaos theory, or 'butterfly effect' applies equally to the weather where a small storm in Asia can trigger weather effects around the globe. In the case of Irma, Martínez noted it had one hurricane - Katia - moving ahead of it in the Gulf of Mexico, another hurricane behind it - Jose - in the Atlantic, and a high-pressure system over the U.S. steering it as well.
Saved by 50 miles
And then there's the element of luck.
Had Irma tracked 50 miles further north along Cuba's coast, the results could have been dramatically different, meteorologists say.
"It would have been a Category 5 (156 mph or more) instead of a Category 3 (111-129 mph)," said Masters, causing devastation to the densely populated Greater Miami region. In general, damage rises by about a factor of four for every category increase.
Also by tracking up Florida's west coast close to the shoreline deprived Irma of the warm Gulf water that fuels storms. "I'm convinced it would have been at least a $200 billion catastrophe," he added.
As it was, the more violent winds on Irma's back side (lower-east quadrant) passed over the barely-populated Everglades. Had Irma passed a little further west over the Gulf, flooding in Naples, Fort Myers and Tampa would have been far more severe, Masters said.
More money
Meteorologists say they could still do with better tools. "We have to invest in our talent", said Masters. While predicting the track of storms has improved greatly, forecasters say they have less understanding of hurricane intensity, which is of growing concern due to warmer global sea temperatures that not only fuel wind speed but also produce greater storm surge.
In 2006 the National Science Board recommended spending $300 million a year on hurricane research funding - 10 times current levels. Both Florida's U.S. Senators proposed legislation at the time to increase funding. The bill was referred to the Committee on Commerce, Science, and Transportation and died there.
As justification, science board noted that hurricane damage is increasing, with annual total losses averaging $1.3 billion from 1949-1989, $10.1 billion from 1990-1995, and $35.8 billion per year from 2000-2005. $168 billion in losses occurred in 2004 and 2005 alone. The value of infrastructure in the Gulf and Atlantic coast areas is over $3 trillion.
The Trump administration in June proposed a 16 percent cut to NOAA's budget, including losing $5 million for the development of advanced models for storm forecasting, as well as $26 million less for programs that monitor global environmental climate patterns related to droughts, floods and other extreme weather events.
The insurance industry is calculating about $25 billion in losses from Irma, more or less within its ability to cover claims.
But forecasters said Florida is not out of the woods. "There's an old saying that hurricanes are like bananas, they come in bunches", said Masters, recalling a string of expensive storms in 2004 and 2005 that put some insurers out of business.
"The 2017 hurricane season is only half over. Hopefully we've had our bunch."
