Comparing a multivariate global ocean state estimate with high-resolution in situ data: an anticyclonic intrathermocline eddy near the Canary Islands

摘要

The provision of high-resolution extit{in situ} oceanographic data is key for the ongoing verification, validation and assessment of operational products, such as those provided by the Copernicus Marine Core Service (CMEMS). Here we analyze the ability of ARMOR3D - a multivariate global ocean state estimate that is available from CMEMS - to reconstruct a mesoscale anticyclonic intrathermocline eddy that was previously sampled with high-resolution independent extit{in situ} observations. ARMOR3D is constructed by merging remote sensing observations with extit{in situ} vertical profiles of temperature and salinity obtained primarily from the Argo network. extit{In situ} data from CTDs and an Acoustic Doppler Current Profiler were obtained during an oceanographic cruise near the Canary Islands (Atlantic ocean). The analysis of the ARMOR3D product using the extit{in situ} data is done over (i) a high-resolution meridional transect crossing the eddy center and (ii) a three-dimensional grid centered on the eddy center. An evaluation of the hydrographic eddy signature and derived dynamical variables, namely geostrophic velocity, vertical vorticity and quasi-geostrophic (QG) vertical velocity, demonstrates that the ARMOR3D product is able to reproduce the vertical hydrographic structure of the independently sampled eddy below the seasonal pycnocline, with the caveat that the flow is surface intensified and the seasonal pycnocline remains flat. Maps of ARMOR3D density show the signature of the eddy, and agreement with the elliptical eddy shape seen in the extit{in situ} data. The major eddy axes are oriented NW-SE in both data sets. The estimated radius for the extit{in situ} eddy is $sim$ 46 km; the ARMOR3D radius is significantly larger at $sim$ 92 km and is considered an overestimation that is inherited from an across-track altimetry sampling issue. The ARMOR3D geostrophic flow is underestimated by a factor of 2, with maxima of 0.11 (-0.19) m s$^{-1}$ at the surface, which implies an underestimation of the local Rossby number by a factor of 3. Both the extit{in situ} and ARMOR3D eddies have decelerating flows at their northern edges. The ARMOR3D QG vertical velocity distribution has upwelling/downwelling cells located along the eddy periphery and similar magnitudes to the extit{in situ}-derived QG vertical velocity.