Estimating surface water flow speeds using time–frequency methods

Abstract

Synthetic aperture radar (SAR) image formation implicitly assumes that the backscattered returns arise from stationary isotropic targets. When objects move during the SAR integration time, signal anomalies appear that distort, displace and smear moving targets in the image. Although these anomalies degrade the image, they provide the information that allows analysis of the underlying target motion. Joint time-frequency analysis (JTFA) exploits these signal anomalies to estimate the motion of typical point targets. However, the weak, transient returns from water surface scatterers complicate standard JTFA estimates of water surface speeds. A time-frequency representation is applied based on the Capon-s spectral estimation technique that allows joint analysis of multiple azimuth lines, thereby increasing the signal-to-clutter ratio of weak scatterers. The authors compare the time-frequency estimate, employing single-phase-centre SAR data, to along-track interferometric SAR estimates of the same flow and show that both the methods produce comparable results. The authors derive the JTFA equations and estimate water surface speed for data collected at a specific imaging geometry. This study highlights the feasibility of using a single-phase-centre SAR system to determine the motion of slow moving distributed targets representative of water flow.