Direct Comparison of Sea Surface Velocity Estimated From Sentinel-1 and TanDEM-X SAR Data

This article presents the first direct comparison of the sea surface radial velocity (RVL) derived from the two satellite SAR systems Sentinel-1 and TanDEM-X, operating at different frequencies and imaging modes. The RVL is derived from the Doppler centroid (Dc) provided in the Sentinel-1 OCN product and from the along-track interferometric phase of the TanDEM-X. The comparison is carried out using unique opportunistic acquisitions, collocated in space and time, over three different sites located in the Iceland Sea, the Pentland Firth, and the Kattegat Sea. First, it is observed that the RVL derived from both satellites is biased, thus calibration is applied using the land as a reference. The comparison shows that the correlation and the mean bias between the two datasets depend on the differences in acquisition time, incidence angle, and azimuth angle, and on wind and surface velocities. It is found that, given a time difference of less than or similar to 20 min, the spatial correlation coefficient is relatively high (between 0.70 and 0.93), which indicates that the two SAR systems observe similar sea surface current fields. The spatial correlation degrades primarily due to increasing time difference and decreasing velocity magnitudes. It is also found that the mean RVL bias increases primarily with the radial wind speed, which suggests that the bias is mainly due to the wave-induced Doppler shift. This article shows that under certain conditions, i.e., similar acquisition geometry and short time delay, a good agreement between the two independently derived RVL is achieved, both in the spatial variation and absolute mean value. This encourages a synergistic use of the sea surface velocity estimated from different C- and X-band SAR systems.

Automated summary

This article presents the first direct comparison of sea surface radial velocity derived from two different satellite SAR systems. The study finds that while there are biases between the two datasets, under certain conditions, a good agreement can be achieved, encouraging the synergistic use of different SAR systems for estimating sea surface velocity.