Using Ensemble Ocean Currents for Drift Predictions

Predicting the drift of objects and substances in the ocean is relevant for several important applications, such as assisting search and rescue and oil cleanup operations, and for understanding and analysing the drift and distribution of plastics or fish eggs and larvae in the ocean. The largest uncertainty of such simulations is normally due to currents obtained from ocean models, in particular for the short-term applications such as oil spill accidents and search and rescue operations. Assimilation of observations help make the ocean models more precise, but unfortunately only a limited amount of observations with high spatial resolution are available. Trajectory models compensate for the imprecise forcing data by adding horizontal diffusivity, providing a spatial spread to encompass the most likely drift. However, a more realistic spread can be obtained by running an ensemble ocean model, where current fields are perturbed in a more physically sound way. In this study, we are analysing a set of ocean surface drifters released in the Fram Strait and Barents Sea, within the domain of an 24 member ensemble setup of ROMS with 2.5km pixel size, run operationally by MET Norway. We explore and demonstrate methods to combine the ensemble current fields to improve the predictability of the drifter trajectories. Also, we demonstrate a method for further improvement of future predictability for situations where a recent part of the drift trajectory is known, e.g. for an object with GPS tracking that has lost connection.