From 1,000 barrels a day at the outset to as much as 40,000 barrels a day today, estimates of the amount of oil that poured from BP's Deepwater Horizon blow-out during the event's initial 44 days have climbed.
Early jumps in the flow rate estimates came as the scope of the disaster expanded, but more recent increases reflect the sophisticated tools scientists are now using on the flow-estimation problem.
These tools have long been used for basic science, but this may well mark the first time some of them have been applied to a real-world environmental problem.
In describing a sonar-based approach she and her colleagues are using to take the measure of the blow-out's flow, University of Georgia marine scientist Daniela Di Iorio says, "I never in my mind would have dreamed of using this to measure an oil plume." Typically, she says, she has used the technique to study underwater hydrothermal vents.
The national incident command, which is coordinating the response to the blow-out, set up its flow-rate technical group roughly a month after the blast. Within the group, four teams have emerged, each using a different approach for estimating the flow.
One of those approaches, for instance, involves analyzing high-resolution video of the plume to estimate the pace at which material is flowing. The title is a mouthful: particle image velocimetry. But the principle is simple, explains Peter Cornillon, a researcher at the University of Rhode Island's Graduate School of Oceanography who has worked with the flow-rate technical group team using the approach.
He likens it to estimating a race car's speed by using a video camera aimed at a fixed spot along the track. The difference in the car's location from one video frame to the next yields an estimate of the distance traveled. And the speed at which the camera records each individual frame provides a way to calculate the time it took the car to travel that distance.
Typically, the well-known technique is a lab tool in which the fluid is clear, and tiny particles made of anything from glass beads to soil can be used as tracers that allow scientists to see the flow patterns in the fluid.
"We don't have little glass beads in this oil plume, and it's opaque," Dr. Cornillon continues. So the team using this technique has focused its attention on individual billows in the oil plume as markers for tracking the flow.
Billows change quickly with time, but with the software packages scientists use to analyze the videos, "you can come up with a pretty reasonable estimate of how far it moved," he says.
Help also has come from outside the core technical group. Scientists from four academic institutions, led by the Woods Hole Oceanographic Institution's Richard Camilli, have been pinging the plume with two types of sonar mounted on a small remotely operated undersea vehicle.
One sonar aims its signals horizontally. By timing the echoes as sound scatters back from different portions of the plume, researchers can estimate the plume's width. Another beam, aimed upward at an angle, compares the outbound pitch or frequency of each ping against the pitch of the pings coming back after scattering off the receding plume.
The return ping will have a lower pitch as it scatters off the fast rising plume, just as a train horn appears to fall in pitch as it speeds away from someone near the track.
Using the angle at which the upward-looking sonar is aimed, and the estimate of the plume's width from the horizontal measurement, researchers can calculate the volume of material flowing out of the well-head, as well as the pace of the flow, explains Dr. Di Iorio, who is a member of a team of scientists applying the technique to the BP blow-out.
Additional teams are using images of the ocean taken from satellites and aircraft to estimate the oil on the surface. Others are modeling conditions in the well's reservoir itself to derive additional, independent estimates of flow rates.
For all their high-tech trappings, the techniques researchers are using have common sources of uncertainty, as well as sources unique to each approach.
To begin with, "there's no guarantee the flow rate stays the same over time," says Steven Wereley, an associate professor of engineering at Purdue University and a member of the team analyzing video of the plume.
Moreover, the oil-flow estimates depend heavily on a good estimate of the amount of oil in the plume, which is a mixture of oil, gas, brine, and mud. Over time, BP has supplied three different estimates of that so-called oil-to-gas ratio, Wereley says.
With that number in hand, "we'll go back to all of these groups, take a look at what their estimates were," she says, and see what adjustments need to be made in using these techniques.
"We will learn so much more about measuring oil in the ocean that we will be able to do a much better job next time measuring the release of oil," she says.