Abstract:
This study examined the surface current dynamics around Madagascar and the Mozambique Channel. Dividing
these regions into the Eastern Madagascar Coast, Northern, Central, and Southern Mozambique Channel
subregions allowed for the assessment of the role of geostrophic-only and combined geostrophic and Ekman
currents in the dynamics. Moreover, calculated virtual drifters were compared with 151 hourly Surface Velocity
Program drifters (SVP drifters) from the Global Drifter Program spanning from 2000 to 2019. In addition,
while some SVP drifters retained their drogues, others lost them; this status was considered in the analysis.
The virtual drifter’s trajectories were calculated using the basic Lagrangian modelling equation, which requires
velocity fields. The velocity fields employed are based on the CNES-CLS18 mean dynamic topography model,
which comprises satellite altimeter observations, wind from numerical models and in-situ observations from
SVP drifters. These velocity fields are compiled at four levels of processing by the European Union-Copernicus
Marine Service. The virtual drifters from the two velocity fields were designated as Virtual Geostrophic Drifter
(VGD) and Virtual Geostrophic-Ekman Drifter (VGED), while the SVP drifters were classified as real drifters.
The results reveal substantial underestimations, averaging 29% with VGDs and improving to 21% when Ekman
drift is incorporated in the VGEDs compared to real drifters. This shows improvement in the CNES-CLS18
mean dynamic topography model. The Eastern Madagascar Coast exhibited the highest velocity discrepancies
due to the formed dipole in the southern tip of Madagascar, which prevented the movement of specific VGD
to the channel. However, the Northern Mozambique Channel showed lower underestimation due to strong
western boundary currents, local winds, and tropical cyclones, highlighting the region’s dominance of wind
seasonality. Due to the eddy’s linearity and non-linearity, virtual drifters have substantial position errors in
the Central and Southern Mozambique Channel. Moreover, virtual drifters failed to identify key submesoscale
structures because the velocity field’s 25 km resolution was a significant issue. The dispersion of virtual drifters
in the offshore Cabo Delgado coast of the Northern Mozambique Channel was predominantly influenced
by seasonal Ekman-driven currents. This highlights the necessity of incorporating these dynamics into future
simulations as winds modulate this region. A considerable part of these virtual drifters ran aground shortly
after being released due to the strong winds from the south that influence the ocean surface, displacing them
45° to the left in relation to the wind’s direction. This phenomenon was observed predominantly in winter.
