Satellite's malfunction casts new questions on accuracy of future weather forecasts
For a brief period on May 22, Dan Kottlowski’s view of the eastern Atlantic Ocean went dark, and as a senior meteorologist at AccuWeather, that’s a problem.
A satellite used for monitoring weather conditions had malfunctioned for the second time in a year, and suddenly Kottlowski’s ability to provide clear forecasts to hundreds of clients and the general public had been hampered.
"We really did not have any coverage for a period of time off the East Coast," Kottlowski said. "It put us in an awkward position."
The National Oceanic and Atmospheric Administration did not respond to repeated requests for comment, but a bulletin posted on its website shows the satellite was restored to service last week after engineers were able to correct the problem and another satellite was repositioned to make up the gap — ultimately, no harm was done.
But the incident has frayed nerves in the nation’s forecasting community, which worries that the federal government’s aging fleet of satellites is in danger of failing and more than $20 billion in planned replacements are not on the immediate horizon.
Were another Hurricane Sandy to strike in the coming years, they say, forecasters might not have the same ability to accurately warn of the threat.
"It shows that we’re on shaky ground," said David Robinson, the state climatologist at Rutgers University. "It illustrates the potentially dire straits our environmental satellite program is going to be in come 2015, 2016, when all of our satellites are getting to the end of their life expectancy."
Aside from the skilled eyes of meteorologists around the country, weather forecasting in the United States is anchored by two things: satellites and supercomputers.
Supercomputers maintained by the National Weather Service churn through trillions of mathematical calculations a day, providing the basis for North American forecast models heavily relied upon by the meteorological community.
After Sandy, the federal government recently announced plans to pump $25 million into upgrading the National Weather Service’s forecasting supercomputers, welcome news after European forecast models widely outperformed their North American counterparts during the storm.
But the massive computing systems are only as good as the data fed to them by satellites, and experts say that if an emphasis isn’t placed on upgrading the nation’s aging and ailing satellite fleet, it won’t matter and could imperil the accuracy of forecasts both in the United States and abroad.
"You can have the fastest car on the planet, but if you don’t put gas in it, what’s the point?" said Marshall Shepherd, the president of the American Meteorological Society. "If you take out the satellite data, all of a sudden (forecast) models become very average."
Experiments conducted following Sandy show that without satellite data, some forecast models would have completely misjudged the storm, predicting that it would float harmlessly out into the Atlantic Ocean rather than making a direct hit on New Jersey, as was forecast in the days ahead of the storm’s eventual landfall.
"They kind of take the pulse of the planet on a regular basis," Robinson said. "You remove any one of these and it’s going to degrade your forecast."
In February, the federal Government Accountability Office added the potential for gaps in satellite data to a list of the 30 largest risks facing our nation today. A failure to address the issue could not only imperil life and property, forecasters say, but jeopardize the private weather industry — estimated to be worth more than $200 million — that relies on the data satellites collect.
Funding delays, cutbacks and project mismanagement have deeply affected plans to replace two groups of satellites, which are expected to reach the end of their lifespan midway through the decade.
The satellite that failed last month was part of NOAA’s Geostationary Operational Environmental Satellite system. Those four satellites remain in orbit over the United States, providing a constant view of atmospheric and meteorological conditions.
The malfunction, the second since September, forced engineers to shut the satellite down for weeks while they corrected the problem. Another satellite was repositioned to fill the loss of coverage, but it left NOAA on tenuous ground without another U.S. satellite to fall back on if one of the others faltered.
"When it went down, it sort of forced us to run on our spare tire," Shepherd said. "What happens if our spare tire goes out?"
Replacements are planned to be launched later this decade but may not come as soon as needed. In a letter to Congress earlier this year, Secretary of Commerce Rebecca Blank, whose department oversees NOAA, warned that funding cuts due to sequestration would further delay the launch of new geostationary satellites.
"Sequestration will result in a 2-3-year launch delay for the first two next-generation geostationary weather satellites (currently planned to launch in 2015 and 2017), which track severe weather events such as hurricanes and tornadoes," Blank wrote.
Of equal or greater concern is an expected gap in polar-orbiting satellites, which constantly circle the Earth and provide meteorological data crucial for accurate forecasts.
In the days ahead of Sandy, the satellites picked up on key meteorological events occurring across North America, which ultimately allowed forecast models to accurately predict the storm would turn toward the coast.
NOAA currently operates a pair of those satellites, but they, too, are quickly nearing the end of their lifespan. With the next generation of satellites not expected to launch until 2017 at the earliest, the agency believes that starting sometime in 2016 there will be a period of up to 53 months in which it will only have one operational polar-orbiting satellite, eliminating 50 percent of the data it currently collects.
An experiment was conducted following Sandy, reproducing conditions several days ahead of the storm but eliminating the satellite data. It showed that, in the absence of robust data,the lauded European forecast computer model was left blind in places and errors cascaded to throw the forecast completely off.
"Garbage in, garbage out; that’s always the rule of thumb when it comes to these computers," Kottlowski said.