Lagrangian Predictive Skill Assessment for the Deepwater Horizon Spill

The explosion and sinking of the Deepwater Horizon drilling platform produced enormous human, ecological, and economic impacts. At the same time this disaster provided an unprecedented amount of Lagrangian information on ocean processes, including a large number of surface and near-surface drifters deployed in the northeastern Gulf of Mexico as well as remotely sensed images of the surface oil slick. In addition several global and regional ocean model predictions were used to forecast the spill movements. These models generally exhibited large variations in the mesoscale flow near the Deepwater Horizon site, even though they all assimilated similar sets of ocean observations. This provides a unique opportunity to thoroughly assess model Lagrangian predictive skill. Here, the predictive skill of one model, a regional implementation of the Hybrid Coordinate Ocean Model (HYCOM), is evaluated using data from more than 80 drifter trajectories in the northern Gulf of Mexico. These trajectories are compared with maps of Lagrangian coherent structures, computed from near-surface model velocities, to determine whether the observations are consistent with the larger scale transport structure predicted by the model. We also discuss new metrics to assess model Lagrangian predictive skill of the plume movement.