Speckle Energy Spectrum in a Complex Radar Viewing Geometry From the Spaceborne SWIM Observations and From Model

The estimation of speckle noise or its suppression is a crucial need to estimate geophysical parameters from radar remote sensing. This is particularly important for the ocean wave spectra retrieval from measured radar backscatter coefficients, as speckle noise can completely dominate the signal in certain conditions. Such measurements are performed by the spaceborne SWIM radar, a near-nadir looking, conical scanning instrument carried by the CFOSAT satellite dedicated to the measurement of directional spectra of ocean waves from the analysis of backscattered fluctuations. In this study, we compare empirical results on the speckle spectrum obtained from the SWIM observations using three different types of acquisition and a theoretical model. We show that a cross-spectral method is efficient for estimating the speckle energy spectrum with a look separation of the order of 13 ms, but shows some limits when this time lag is increased to about 40ms because of the limited footprint overlap. In the case of an azimuthal scanning geometry like SWIM, the speckle energy increases by several order of magnitude in a sector of about ±15° close to the along-track direction. From the theoretical model, we conclude that this is due to the decrease of the Doppler bandwidth in this look geometry and that the variance of the surface scatterer velocities limits the speckle energy increase. It also explains the sensitivity to wind speed of the speckle energy in this azimuthal sector. The almost linear decrease of speckle energy with wavenumber is explained by the spectral response of the radar impulse, slightly modified by range decimation and resampling applied in the first steps of processing. Close to the along-track direction, the variation of speckle energy with latitude is mainly explained by the variation in the integration time imposed by the SWIM on-board command. The model results also show that the impact of Earth rotation on the speckle noise is small. Finally, we show that the speckle correction has an impact on the shape of the wave spectrum derived from the SWIM inversion.