This work studies the links between the Lagrangian transport in the Southern Benguela upwelling system and the ocean circulation through modelling experimentation. More specifically, it intends to show that the wind-induced circulation at short-time scales impacts the drift of Lagrangian particles released in the model. Three ocean model (ROMS) simulations are set-up. Simulation A is forced with a6 hourly atmospheric forcing (surface heat and fresh-water fluxes and wind stress). In simulation B and C, the atmospheric forcing is low-pass filtered with 5 days and 30 days cut-off periods. The ocean model outputs are averaged and saved at deferent temporal frequency: frequencies of 6 hours and 3 days. Particles released in the Lagrangian tracking tool are transported by the velocity vectors produced by the ocean model into the nursery area located at Saint Helena Bay. The presence of short-term fluctuations result in higher inner shelf transport and amore elongated plume dispersion pattern with much refine spatial scales. This finding contrasts with previous studies where the inner shelf transport success was always low compared to the outer shelf nursery. When the high frequency signal is filtered out, the inner shelf transport is greatly reduced especially during the upwelling season. The archiving frequency of the ROMS output is shown to impact Lagrangian studies, especially when the atmospheric forcing has variability at short time scales (less than 5 days).Monthly mean forcing results in dynamics variable with periods of not less than a month and 5 daily averaged forcing with sub-weekly dynamics. Therefore, in simulations B and C the less than 3 days archiving of the model outputs is not necessary, because their forcing does not generate dynamics with such periods. The latter must be adapted so that it does not filter out the ocean response to the high frequency atmospheric forcing. To explain the Lagrangian transports, as a first approach, the transport success was linked to the variability of the ocean circulation.

The Agulhas Current System is a vital element of the global ocean-climate system by virtue of its role in the transfer of energy, nutrients and organic material. In the context of working towards better climate change projections, it is necessary to develop a robust understanding of the complex dynamical mechanisms which facilitate this transfer. Mesoscale cyclonic and anticyclonic eddies transport heat, salt, organic matter and nutrients from the Indian Ocean into the South Atlantic Ocean. In so doing, they are key drivers of the Atlantic Meridional Overturning Circulation (AMOC). As such, it is important that they are adequately simulated by numerical models in order to advance the accuracy of climate prediction. In the absence of spatially and temporally coherent observing systems, numerical models provide the capacity to describe the oceanographic conditions of the region. Given the complexity of the regional dynamics, and the challenges it presents to free-running numerical models, data assimilation is a valuable tool in improving simulation quality. An important step in this continuing process is the objective, quantitative evaluation of model configurations, such that they can be continuously refined. In this study, the impact of assimilating along-track sea level anomaly (SLA) data is investigated with regard to the simulation of mesoscale eddies in the Agulhas System. Two configurations of a Hybrid Coordinate Ocean Model (HYCOM) configuration are analysed; one free run (hereafter 'Free') and one with along-track SLA data from satellite altimetry assimilated (hereafter 'Assim.') via an Ensemble Optimal Interpolation (EnOI) data assimilation scheme. The results of these two configurations are compared with each other, and against a set of corresponding observational data from satellite altimetry (hereafter 'Aviso'). To this end, an automatic eddy detection and tracking algorithm is implemented, in order to quantify eddy characteristics in a coherent and consistent manner.