A comparison of in-situ float velocities with altimeter derived geostrophic velocities

Satellite borne altimetric data are of increasing prominence for assimilation in ocean circulation models and interpretation of localized in-situ measurements. Physically, geo-referenced sea-surface height (SSH) data products are mostly referenced relative to a long-term SSH mean, and consequently called SSH-anomalies. The Modular Ocean Data Assimilation System (MODAS) adds climatological SSH fields to provide space time interpolated absolute steric SSH fields. This, in theory, should provide realistic geostrophic surface velocities and flow patterns, including quasi-permanent features such as western boundary currents or free jets. This study compares such data for the wider Agulhas Retroflection region with co-located, simultaneous velocity measurements from Cape of Good Hope Experiments (KAPEX). KAPEX used ship-borne Acoustic Doppler Current Profiles (ADCP) and neutrally buoyant RAFOS (Ranging and Fixing of Sound) floats at intermediate depths to obtain in-situ velocity data. Correlation coefficients of MODAS-2D geostrophic and RAFOS subsurface flow directions fall between 0.8 and 0.9 with a typical error less than 0.05. The high correlation suggests that MODAS-2D provides a correct depiction of anticyclonic/cyclonic flow patterns in this region, making it a useful tool to describe the Agulhas Retrollection. Root-mean-square differences between velocities as measured by the various data sources rage between 20 and 30 cm s(-1), lying between the natural variability observed for the intermediate and surface layers. Decreasing slope parameters of linear regressions between MODAS, RAFOS and ADCP velocities reflect the baroclinic velocity shear. Slope equals I at surface and decreases to 0.4 at depths below 1000 m. Offsets of linear regression of these fits are not significantly different from zero, except for the zonal component in the Agulhas Return Current (5-10 cm s(-1)). This discrepancy suggests a missing meridional gradient in this region's climatological signal that is added to the SSH anomaly field within MODAS. (C) 2002 Elsevier Science Ltd. All rights reserved.