Enhancement of turbulence and nutrient fluxes within an Eastern Boundary Upwelling Filament: a diapycnal entrainment approach

Abstract

The filaments of the African Eastern Boundary Upwelling System (EBUS) are responsible for feeding nutrients to the oligotrophic waters of the northeastern Atlantic. However, turbulent mixing associated with nutrient uplift in filaments is poorly documented and has been mainly evaluated numerically. Using microstructure profiler measurements, we detected enhanced turbulent kinetic energy dissipation rates (ε) within the Cape Ghir upwelling filament. In contrast to previous studies, this enhancement was not related to symmetrical instabilities induced by down-front winds but to an increase in vertical current shear at the base of the mixed layer (hρ). In order to quantify the impact of vertical shear and the influence of the active mixing layer depth (hε) in the filament, a simple one-dimensional (1D) turbulent entrainment approach was used. We found that the effect of turbulent enhancement, together with the isopycnal morphology of the filament front, drove the formation of local positive entrainment zones (Δh=hε−hρ), as hε was deeper than hρ. This provided suitable conditions for the entrainment of cold, nutrient-rich waters from below the filament pycnocline and the upward transport of biophysical properties to the upper boundary layer of the front. We also found that diapycnal nutrient fluxes in stations influenced by the filament (1.35 mmol m-2 d-1) were two orders of magnitude higher than those of stations not affected by the filament front (0.02 mmol m-2 d-1). Despite their importance, the effects of vertical shear and hε have often been neglected in entrainment parameterizations. Thus, a modified entrainment parameterization was adapted to include vertical shear and observed ε, which are overestimated by existing parameterizations. To account for the possible role of internal waves in the generation of vertical shear, we considered internal wave scaling to parameterize the observed dissipation. Using this adapted parameterization, the average entrainment velocities were six times (6 m d-1) higher than those obtained with the classic parameterization (1 m d-1).