Predicting The Flow: Math Helps Track Oil Spills

Instead of random guessing about where the oil from a spill might end up, scientists have now created a complex model to track exactly where it (or ash from volcanoes) will go next.

In the days after the 2010 Deepwater Horizon disaster, there was a lot of conjecture about which way the spill might move. The worst-case scenario, said scientists, was for the crude to become part of the Loop Current, which would potentially take it south into the Florida Keys. But nobody could be sure. Apart from watching from the air, decision-makers had to prepare for the worst, and hope for the best.

George Haller, a mechanical engineer at McGill, hopes to be able to give more warning next time—and not only in the case of oil spills. Collaborating with another researcher at the University of Miami, Josefina Olascoaga, Haller has come up with a method of understanding the intricate internal patterns of large-scale contaminations, including oil spills, volcanic ash clouds, and tracts of oceanic plastic waste.

Haller describes the ocean as a complex swirl of "roads" and "intersections." When flows come from opposite directions, they eject a mass of water, some of which are powerful and persistent enough to propel an oil slick in a particular direction.

By analyzing "velocity data" from thousands of points in the ocean, Haller and Olascoaga’s model is able to predict what will happen next—in the case of oil spills, up to five days in advance.

"If you look at it, it’s not obvious what’s going to happen," says Haller. "But the research has shown there are hidden material structures in the mass that create coherence in the flow. The little fluid particles that were moving without a pattern suddenly fall in order, and start marching in the same direction."

Haller and Olascoaga took historical data from the Gulf spill, and accurately forecast two major "instabilities" in the mass: the "tiger’s tail" that drifted toward the Loop because of shifts within the spill, and "coastal drift" that occurred because of flows coming in from the outside.

Most importantly, Haller says the research should help responders manage spills, for example by knowing where to use dispersant, and which parts of the coastline should be evacuated.

Similarly, the research could be used to predict the movement of ash clouds, and help planners to know what airports and flights might need to be shut down. In 2010, fallout from the Eyjafjallajokull volcano, in Iceland, led to the grounding of most of Europe’s air traffic—unnecessarily according to many people. Haller says a better model could lead to a more rational and science-based response.

Part of the funding for the research came from the BP-funded Gulf of Mexico Research Initiative. Haller says one silver lining from disasters such as Deepwater Horizon is that they can sometimes lead to better science, though it’s hardly a pleasant way forward.

"Technology improves because there is so much data to look at afterwards," he says. "In the big scheme of things, it probably advances science, although [the disaster] is clearly not a blessing to society or the environment."

Add New Comment


  • Chris Barker

    The professors probably do know what is new and what is not, but the journalists need to see through the hype a bit: As Bill Lehr says, there are many (and have been for decades) models used for predicting where oil spills (and lots of other things) go -- didn't you see the forecasts put out by NOAA (and others) during theDeepwater Horizon event?

    Yes, there was a lot of speculation about the potential for Loop Current transport, but that's because no one can predict what the Loop Current (or any of the currents in the Gulf) will look like more than a few days in advance. If you read the cited article, it says:"""...provides a model-independent forecasting scheme that relies only on
    already observed or validated flow velocities at the time the prediction is made"""So this "new" approach requires "validated flow velocities" -- that's the whole problem -- longer term forecasts could not be done because those flow velocities are not available. Period.I'm not saying there isn't anything new or useful in this work, but it's not new to be able to forecast the movement of oil a few days out -- so what is new?

  • George Haller

    Thanks to Chris Baker and Jim Lehr for their comments.Our mathematical algorithm relies on observed or validated data(now-casts) up to the present time, and uncovers hidden instabilities that have already been developing  in the flow. We detect their exact location, direction and strength. At no point do we run model trajectories intothe future, as the above two commenters seem to have assumed.
    As Chris Barker correctly notes, nobody knows what the Loop Current velocity field will exactly look like in a few days. However, that turns out to be of secondary concern, once a massive stream of water and oil is known to be gaining momentum, heading towards the perimeter of the oil spill. We would neednothing short of a hurricane to stop that. Granted, our methodscannot predict hurricanes.The resulting instabilities in the spill shape are driven by  Lagrangian Coherent Structures (LCSs). These are hidden flowfeatures that the fluid dynamics and geophysics communitiesare just beginning to understand and exploit. This wheel has certainly not been spinning for three decades; it is being put on the axle as we speak. Please check the literature, if needed.One source to check could be the final report of a 2011 NOAA workshop( http://www.crrc.unh.edu/public... this meeting, a discussion group  led by Chris Barker was in charge of reviewing  'research needs for improving trajectory models."  One  of their main recommendations was to "Improve Lagrangian coherent structures". We've done just that. If there are remaining scientific questions about this work, I'm happy to take those off-line.

  • Bill Lehr

    This opinion is my own, not that of any agency of the U.S. government. Note to the professors: The wheel and oil spill trajectory models already exist. There is no need to reinvent them. Quite sophisticated trajectory models were used in coordination with extensive remote sensing data to accurately track and predict the surface oil movement during Deepwater Horizon.There are many areas of oil spill science that needs more research, such as subsurface oil behavior and emulsification formation. On the other hand, basic oil surface transport has been modelled successfully for three decades. I applaud your efforts but encourage you to work closely with existing experts in the field.

  • Rduran

    Actually state of the art Hazardous material transport has many shortcomings, and this I have heard directly from the people in charge in NOAA. I don't mean to be rude but is is obvious that comments suggesting the contribution described in the article is nothing new or no significant contribution are not well informed.
    As a PhD candidate in Physical Oceanography I can recognize that the paper by Olascoaga and Haller is a significant contribution in a very promising and relatively new research tool that is expected to make a big difference in the very difficult task (in fact we already know it is impossible for long time scales) of predicting trajectories in a chaotic system.