Observation and prediction of flotsam

Abstract

Surface drift of eighty-nine undrogued RAFOS floats in the California Current System (CCS) has been studied to describe patterns of flotsam drift, its seasonal variability and predictability limitations. The floats were launched in the California Undercurrent during 1992–2010 and were tracked by the ARGOS system when they surfaced at the end of their subsurface missions. The duration of surface trajectories varied from as short as 11 to as long as 280 days. The surface drift of these floats was typically equatorward in the California Current. However, some floats moved poleward into the Subpolar Gyre, and others drifted westward into the North Equatorial Current. Usually, observations of surface currents use drifters which are coupled to the surface layer by drogues located at 15 m depth. While drogued observations are useful for studies of circulation of the upper layer of the ocean, a more typical operational problem involves trying to find flotsam that has fallen off the deck of a ship or to predict the path of an abandoned vessel. To better understand the behavior of these surface drifting objects, observations of the surface drifts of RAFOS floats in the California Current system were used to compare the floats’ motions to wind-induced drift and to evaluate the drift prediction by three ocean models: Ocean Surface Current Simulator (OSCURS), Global Navy Coastal Model (gNCOM) and Hybrid Coordinate Ocean Model (HYCOM). Cross-Calibrated Multi-Platform Ocean Surface Wind Vector L3.0 First-Look Analysis wind data were used to analyze the surface drift of the RAFOS floats. The best correlation between float drift speed and wind speed was observed during summer and fall months, when the wind regime over the Northeast Pacific is dominated by the North Pacific High. The predominant winds associated with this regime are steady and contribute to the surface drift in the same direction as the large-scale geostrophic component of the California Current. Evaluation of the drift prediction by three ocean models was conducted by comparing observed drifter trajectories with model-simulated trajectories at 7-day time scales. The model-simulated trajectories were initially collocated with RAFOS positions and restarted every 15 days. The OSCURS model was the only one of three analyzed models which allowed for tuning based on the surface RAFOS trajectories. Two main parameters, Wind Angle Deflection (WAD) and Geostrophic Current Factor (GCF), were adjusted to minimize the separation between observed and modeled trajectories. At shorter time scales, from weeks to several months, the tuned OSCURS model outperformed gNCOM and HYCOM models in reproducing the observed surface drift. For seasonal and longer time scales, both gNCOM and HYCOM can reproduce the seasonal variability, flow direction, and eddy-like structures in the California Current system, which may be useful for long-term forecasts of the Northeast Pacific circulation.