The effect of wind parametrizations in MEDSLIK-II oil spill simulations : A case study of the FPU P-53 incident in Brazilian waters

Abstract

Wind parameterization is fundamental in oil dispersion modeling in seawater, as it directly impacts the speed and direction of oil particle transport, influencing response strategies and environmental impact assessments. This study investigated the effects of Stokes drift, wind drift factor, and wind drift angle on oil transport using the MEDSLIK-II model, focusing on the 2019 P-53 platform accident in Brazilian waters. The model’s accuracy was assessed by the difference in distances between simulated and observed trajectories from Synthetic Aperture Radar. Sixteen simulations (REA01–REA16) were conducted, varying the wind drift angle (10° to 45°), drift factor (0%, 3% and 6%), and incorporating Stokes. Increasing the drift factor to 6% accelerated the transport to coastal, while larger angles (>25o) reduced the distance between the simulated and observed oil slicks. Two parameterization sets (45o angle with a 6% factor, with and without Stokes drift) showed the lowest distance errors (23.49 km and 21.16 km), correctly simulating the coastal impact region. Contrary to previous studies, incorporating Stokes drift slightly reduced the model’s performance. This effect and the high distance errors were likely due to (1) inconsistencies in the wind direction data used in the model and actual conditions during the first 21 hours, and (2) the influence of the high- velocity core of the Brazil Current (~0.8 m s -1) only on the simulated plume, intensifying the southwestward transport and deviating from the observed slicks. Additional simulations for the two best parameterization sets, using hydrodynamic data with higher temporal and spatial resolution and artificial northeast winds, reinforced this hypothesis, showing lower distance errors with the inclusion of Stokes drift compared to the model without drift (6 km vs 7 km). These results contribute to the discussion on the parameterization for oil dispersion simulations in MEDSLIK-II at mid-latitudes (~22oS) in the South Atlantic Ocean. The study emphasizes the importance of using hourly, high-resolution oceanic data as model input, supported by in situ measured wind data, to improve accuracy in validation experiments.