Diagnosing Frontal Dynamics From Observations Using a Variational Approach

Intensive hydrographic and horizontal velocity measurements collected in the Alboran Sea enabled us to diagnose the three-dimensional dynamics of a frontal system. The sampled domain was characterized by a 40 km diameter anticyclonic eddy, with an intense front on its eastern side, separating the Atlantic and Mediterranean waters. Here, we implemented a multi-variate variational analysis (VA) to reconstruct the hydrographic fields, combining the 1-km horizontal resolution of the Underway Conductivity-Temperature-Depth (CTD) system with information on the flow shape from the Acoustic Doppler Current Profiler velocities. One advantage of the VA is given by the physical constraint, which preserves fine-scale gradients better than the classical optimal interpolation (OI). A comparison between real drifter trajectories and virtual particles advected in the mapping quantified the improvements in the VA over the OI, with a 15% larger skill score. Quasi-geostrophic (QG) and semi-geostrophic (SG) omega equations enabled us to estimate the vertical velocity (w) which reached 40 m/day on the dense side of the front. How nutrients and other passive tracers leave the mixed-layer and subduct is estimated with 3D advection from the VA, which agreed with biological sampling from traditional CTD casts at two eddy locations. Downwelling warm filaments are further evidence of subduction, in line with the w from SG, but not with QG. SG better accounted for the along-isopycnal component of w in agreement with another analysis made on isopycnal coordinates. The multi-platform approach of this work and the use of variational methods improved the characterization and understanding of (sub)-mesoscale frontal dynamics.

Automated summary

Intensive hydrographic and horizontal velocity measurements in the Alboran Sea were used to study the three-dimensional dynamics of a frontal system. A multi-variate variational analysis method was applied to reconstruct the hydrographic fields, and the results showed that this method improved the understanding of (sub)-mesoscale frontal dynamics compared to classical optimal interpolation. The estimation of vertical velocity and the analysis of nutrient distribution further revealed the characteristics and motion of the frontal system.