Updated

After several missteps, the federal government finally got it right, accurately estimating how much oil spilled into the Gulf of Mexico, an independent scientific study found.

Nearly 185 million gallons of oil spilled from the broken BP well into the Gulf of Mexico this summer, according to a study by two Columbia University researchers who made their estimates based on video of the oil spewing from the well.

The federal government's final estimate was a shade more than 172 million gallons. The Columbia researchers' estimate is 12.6 million gallons more than the federal figure. However, because it's so difficult to get a precise estimate, there is a large margin of error for both the government figure and the Columbia number. The margin is so large that the two estimates essentially overlap, the researchers said. Their study was published online Thursday in the journal Science.

U.S. Geological Survey Director Marcia McNutt, who oversaw federal estimates of the spill size, called the Columbia study "a completely independent and unbiased verification of the government result."

Tim Crone of Columbia, the lead researcher who calculated his estimate based on detailed flow formulas determined by watching video of the leak, "Our numbers overlap, so sure, we agree."

But Crone said he is more confident in his estimate because it went through the rigorous independent peer review required to be published in a respected journal. He added that it is hard to compare in depth to federal numbers "because few details of their methods have been released."

Crone used a different technique to study video than most scientists who analyze flow rates from video. Usually, scientists track particles and calculate a speed as they travel across a screen, sort of like watching a car race down a highway. But in this case the particles were hard to track, so Crone used a technique he's been working on for a decade. Crone studies individual points in the video — all the points — and watches their changes in color and texture. For this study, Crone reviewed video from two dates, May 15 and June 3, and extrapolated for the spill estimate.

The Columbia estimate has a margin of error of 20 percent, so the spill would be somewhere between 148 million and 222 million gallons. The federal estimate had a 10 percent margin of error, so the spill would be somewhere between 155 million and 189 million gallons.

Federal estimates of how much oil spilled and where it has gone have been criticized by both scientists and Congress because the agencies involved have not provided detailed information on how they calculated their figures. But the government numbers were not based solely on traditional video analysis; they also factored in measurements of pressure and estimates from photos and infrared images taken from airplanes and satellites.

Purdue University engineering professor Steve Wereley, who at first was critical of federal efforts and then joined the government team, said he trusts the government work as more comprehensive than Columbia's.

The federal government struggled mightily to calculate the spill's size at first and was sharply criticized for clinging to a too-low estimate for weeks.

Ian MacDonald, a Florida State University oceanographer, said that by his count it took eight tries for the federal government finally to get its calculations right.

Initially, federal officials adopted BP's estimate that 42,000 gallons a day was gushing out. They upped it to 210,000 gallons a day and stuck with that number for weeks. Then the government set up a special team of experts to estimate the spill size; that group came up with a range of estimates that was blasted by independent scientists as still too low.

Finally in mid-June, about two months after the oil rig accident that caused the deepwater gusher, the federal government said the well could be leaking as much as 2.4 million gallons a day. And in August, following the shutdown of the spill the previous month, federal scientists estimated a total leak of 172.2 million.

___

Online:

Science: www.scienceexpress.org

Video of BP spill that Columbia used for the estimate: http://www.earth.columbia.edu/videos/watch/259