One of the challenges in ocean surface current retrieval from synthetic aperture radar (SAR) data is the estimation and removal of the wave-induced Doppler centroid (DC). This article demonstrates empirically the relationship between the dc derived from spaceborne X-band InSAR data and the ocean surface wind and waves. In this study, we analyzed over 300 TanDEM-X image pairs. It is found that the general characteristics of the estimated dc follow the theoretically expected variation with

more »

... ce angle, wind speed, and wind direction. An empirical geophysical model function (GMF) is fit to the estimated dc and compared to existing models and previous experiments. Our GMF is in good agreement (within 0.2 m/s) with other models and data sets. It is found that the wind-induced Doppler velocity contributes to the total Doppler velocity with about 15% of the radial wind speed. This is much larger than the sum of the contributions from the Bragg waves (∼0.2 m/s) and the wind-induced drift current (∼3% of wind speed). This indicates a significant (dominant) contribution of the long wind waves to the SAR dc. Moreover, analysis of dual-polarized data shows that the backscatter polarization ratio (PR = σ 0 VV /σ 0 HH ) and the dc polarization difference (PD = |dc VV | − |dc HH |) are systematically larger than 1 and smaller than 0 Hz, respectively, and both increase in magnitude with incidence angle. The estimated PR and PD are compared to other theoretical and empirical models. The Bragg scattering theory-based (pure Bragg and composite surface) models overestimate both PR and PD, suggesting that other scattering mechanisms, e.g., wave breaking, are involved. In general, it is found that empirical models are more consistent with both backscatter and Doppler data than theory-based models. This motivates a further improvement of SAR dc GMFs. Index Terms-Along-track interferometry (ATI), Doppler centroid (DC), Doppler oceanography, ocean surface winds, ocean surface currents, synthetic aperture radar (SAR), TanDEM-X.