Abstract
In the Tropical Atlantic Ocean, we assessed the accuracy of a Lagrangian model (Ichthyop) forced with velocity fields from a hydrodynamical model (CROCO) and two different remote sensing products (GlobCurrent and OSCAR) using trajectories of oceanographic drifters. Additionally, we evaluated the possibility to expand the drifters data using trajectories of GPS-buoy equipped drifting Fish Aggregating Devices (FADs). The observed and simulated trajectories were compared in terms of spatial distribution, velocity distribution and a nondimensional skill score. For the drifters and FADs, the GlobCurrent and OSCAR products lead to similar performances as the CROCO model-ouputs in the broad studied domain. In the Gulf of Guinea, however, the CROCO model performed significantly better than the other two because the parent solution of CROCO benefited from its communication with a child grid of finer resolution in this region. On average, the simulations lead to an underestimation of the drifter and FAD velocities, likely because the spatial resolutions of the forcing products were insufficient and the time frequency at which they were produced were too low to resolve the relevant oceanic processes properly. We found a low skill for all models to simulate FAD trajectories, possibly because of the devices vertical structure that prevent FADs from drifting like water parcels. Our results therefore suggest that in the Tropical Atlantic the FAD dataset may not be appropriate to use for corroborating Lagrangian simulations.