Weather forecasters were relying on limited radar data on the day of a fatal water taxi accident last year in Baltimore harbor, undermining their ability to issue timely, lifesaving warnings of an approaching storm, the National Weather Service found in an internal review set for release today.
The storm's 40- to 50-mph gusts capsized Seaport Taxi's 36-foot pontoon boat Lady D, pitching 25 people into the frigid water. Five of them died.
Had forecasters on duty March 6, 2004, in the National Weather Service's forecast office in Sterling, Va., called up radar data from Baltimore or Dover, Del., they would have spotted the true threat of the storm bearing down on the harbor.
Wind velocity was the clue that was missed.
"It is likely a more aggressive and more efficient approach by forecasters to analyze evolving weather conditions could have increased concern regarding potential for high winds," the study concluded. "This could have resulted in more timely and more complete information ... in forecast and warning services."
By the time the storm's severity was apparent from instruments on the ground, it was too late. Forecasters issued a special marine warning at 4:05 p.m., urging boats to seek safe harbor immediately. But by then, passengers from the Lady D were already fighting for their lives in the frigid water.
"What we issued was behind the actual events. We were in catch-up, reaction mode," said Dennis H. McCarthy, director of the weather service's Office of Climate, Water and Weather Services.
The findings are part of a "service assessment" - an internal NWS review conducted after any weather event that causes a loss of life or significant property damage.
Weather Service Director David L. Johnson said the review's goal is "to improve products and services, to look at the things we did well, and the things where we need to improve."
The agency has already taken steps to fix the problems. Earlier this summer, the Sterling forecast office became the first in the nation to have Federal Aviation Administration airport radar data integrated directly into forecasters' computer work stations.
David R. Manning, Sterling's warning coordination meteorologist, welcomed the advance:
"Any time we have more data, it helps us to get a better understanding of what the situation is, meteorologically speaking."
The National Transportation Safety Board has not concluded its investigation into the accident. But in a preliminary report in December, the board said the Lady D, operated by the Living Classrooms Foundation, was carrying 700 pounds too much weight.
Although the boat's captain was operating within the 25-person limit set by the Coast Guard, that limit was based on outdated estimates of the average passenger's weight, the agency found. In a pontoon boat, extra weight raises the center of gravity, making it more unstable in high winds.
Johnson said the purpose of the NWS report is not to assign blame for the tragedy and it does not directly fault forecasters. "I think they did the right thing," he said.
But, he continued, "I think that we can provide them tools to help them do it quicker."
A small-craft advisory had been issued on Friday, March 5. A cold front and showers were expected Saturday afternoon. The advisory was still in effect.
Forecasters on the day shift March 6 believed that conditions were unfavorable for the development of dangerous thunderstorms. But skies cleared enough to heat the atmosphere, and the air became unexpectedly unstable, the report found. Radar spotted a line of thunderstorms moving northwest to southeast across the region, toward Baltimore, at 40 to 50 mph.
On their computers in Sterling, forecasters could see the storms on Sterling's own Doppler radar. Calling up radar images from would have required forecasters to move to another computer or drop what was on their work station monitor to seek out radar data from the Internet.
Often, forecasters don't take such steps.
"In a weather situation, a forecaster has to digest a lot of information very quickly, and the more time you spend trying to hunt for it, the less time you have to analyze it," said Manning, the warning coordinator.
Unlike the BWI radar, data from Dover Air Force Base in Delaware were available on the forecasters' work station computers. But Dover was much farther from the storms than Sterling.
McCarthy said there is a "natural tendency for a forecaster to use what feels like 'his' radar," especially when it is closer to the storms.
But as one of the cells in the storm front took aim on Baltimore harbor, that tendency kept Sterling forecasters in the dark about the danger.
Doppler radar measures wind speed by bouncing microwaves off raindrops or ice crystals moving in the wind. The wind speed changes the frequency of the returned radar signal. By measuring the change, Doppler radar can calculate how fast the wind is moving toward or away from the radar.
The problem, according to the weather service report, was that the storms were moving across Sterling's radar beam - perpendicular to the beam - underestimating the winds' velocity.
But the storms were headed directly toward radar beams from BWI and Dover. Had forecasters looked, those radars would have revealed a large area of stronger winds in the same thunderstorms, the report said.
Forecasters were also misled by the absence of reports of wind damage from communities in the storm's path. Some private weather instruments, including the WeatherBug network, were seeing strong gusts, but the data were not available to forecasters on a timely basis.
The severity of the approaching storm became clear shortly after 3:30 p.m., McCarthy said, when automated sensors at airports began to pick up evidence of lightning and winds that called for stepped-up warnings.
Forecasters "reacted appropriately," the report concludes, issuing short-term "NOWcasts" at 3:49 p.m. and 4 p.m., alerting the public to thunderstorms with gusts to 35 mph and 45 mph respectively.
But by then they were in a "reactive mode," the report concludes. The "special marine warning" urging mariners to seek safe harbor from winds up to 50 knots (58 mph) was issued at 4:05 p.m., after the Lady D had capsized.