Surface velocities measured by Synthetic Aperture Radar (SAR) contain contributions from both mean surface motion—referred to as total surface current (TSC)—and the often more prominent sea-state-induced wave motions. Most modern SAR systems cannot distinguish between these two phenomena, which has stifled TSC retrieval from SAR data for decades. We propose a new framework to separate TSC and wave-motion components by leveraging polarization diversity, exploiting the tendency of each phenomenon to imprint distinct signatures on orthogonal polarizations. Building on a foundational signal model, we derive four source-separation algorithms. To address the model’s theoretical limitations, we introduce empirical extensions via symbolic regression, guided by varying levels of theoretical insight. The developed algorithms are evaluated using simulated C-band SAR data, and benchmarked against a reference geophysical model function (GMF) implementation. Our methods demonstrate comparable overall performance, with errors on the order of O(0.1ms⁻¹), and notably outperform the GMF in resolving kilometer-scale spatial features—a domain where traditional GMFs generally struggle. Preliminary results obtained on TanDEM-X observations confirm the generalizability of our approach. These findings highlight the potential of future SAR missions with polarimetric capabilities, such as Harmony, to achieve high-resolution separation of surface-motion sources using polarization diversity.

The interactions between geometrical Doppler, beam pattern, and normalized radar cross section (NRCS) result in the unwanted coupling—leakage—of geometrical Doppler into the perceived geophysical Doppler. Starting from high-resolution synthetic aperture radar (SAR) data we model this leakage for synthesized real aperture radar (RAR) observations of ocean motion. The uncertainty introduced by leakage is ∼1ms−1. Corrections proposed in this work exploit the known interactions between beam pattern and geometric Doppler with NRCS gradients retrieved from simulated low-resolution RAR to estimate and correct for the incurred leakage, reducing the uncertainty to O(0.1ms−1). Further reduction of instantaneous leakage may be achieved through temporal averaging, since the NRCS gradients that cause leakage appear mostly atmosphere induced and decorrelate rapidly. The azimuth resolution and number of independent samples determine a system’s sensitivity to leakage. C-band systems are inherently prone to suffer from greater leakage and worse corrections than their Ku- and Ka-band counterparts. With DopSCA, the propagated effect of leakage is only secondary compared to the pulse-pair uncertainty. In similar systems where pulse-pair uncertainty is suppressed, leakage will dominate instead.