Estimating geostrophic and total velocities from CTD and ADCP data: Intercomparison of different methods

Inferring geostrophic velocity fields from CTD data distributions can be handicapped by the impossibility of referring dynamic height to a no-motion level. This is often the case over the continental shelf, but also at open sea, even when velocity measurements (e.g. from a vessel mounted ADCP) are available. In this paper we test and compare four different methods aimed to estimate the geostrophic and total velocity fields from hydrodynamical data. Two of them can either use only CTD data (then relying on the election of a no-motion level) or incorporate ADCP data (through a multivariate interpolation); the other two methods always combine CTD and ADCP data. A 3D primitive equation model is used to reproduce realistic scenarios that provide control velocity fields and typical CTD and ADCP data profiles. The chosen scenarios represent different dynamic situations (in terms of data quality, bathymetric constrictions and dynamical characteristics such as the relative ageostrophic/geostrophic velocity variance) and make possible a broad discussion on the capabilities and limitations of the examined methods. Results show that the performance of the methods is highly dependent on the dynamics to be resolved. The combination of CTD and ADCP data constitutes the best approach for most of the analyzed situations, though special attention has to be paid when dealing with low quality ADCP data and when the circulation is characterized by intense non-divergent ageostrophic velocities.