Diagnostic analysis of the 3D ageostrophic circulation from a multivariate spatial interpolation of CTD and ADCP data

A multivariate (MV) optimal statistical interpolation method is applied to conductivity-temperature-depth (CTD) and ship-mounted acoustic doppler current profiler (ADCP) data from quasi-synoptic oceanographic surveys. MV analysis aims to improve the spatial interpolation of any particular variable (e.g., dynamic height) by including in the analysis observations of other physically related variables (e.g., current). The version used in this work also provides estimates of the non-divergent and irrotational components of the flow. The method is tested in a sharp frontal region to the north of the Western Alboran gyre. After deriving the optimal analysis parameters, we first show that MV statistical dynamic height analysis errors are significantly smaller than those derived from univariate (UV) analysis. In our region, this translates in a more realistic shape for the geostrophic relative vorticity and the vertical velocity field. The latter peaks at about 45 m/day (as given by the quasi-geostrophic omega equation), with a tendency for light water to be upwelled upstream of the gyre while denser water is downwelled downstream of the gyre. For the horizontal velocity we show the existence of large (up to 40 cm/s) ageostrophic velocities, These are mainly non-divergent and can be explained by the cyclostrophic acceleration induced by the anticyclonic gyre. The irrotational velocity component is of the order of 10 cm/s towards the dense side of the front. The robustness of the method is checked by means of several tests that evaluate the sensitivity of results with respect to the synopticity of the data, the analysis parameters, the reference level and the presence of tidal or inertial currents. (C) 2000 Elsevier Science Ltd. All rights reserved.