Nadir Echo Suppression in Staggered SAR

Abstract

Conventional SAR sensors employ a constant pulse repetition interval (PRI), which results in blind ranges that limit the achievable swath width: a larger PRI yields a wider swath, but also a worse azimuth resolution. The technique of digital beamforming (DBF) allows for simultaneous imaging of multiple sub-swaths, but these sub-swaths are still separated by blind ranges. A staggered SAR employs a continuous variation of the PRI causing transmission events to no longer line up on the same ranges and allowing the consequent missing samples to be interpolated over, provided that data are sufficiently oversampled in azimuth. As a result, a staggered SAR system with DBF is capable of imaging a continuous wide swath with high azimuth resolution. Furthermore, the increased amount of data to be downlinked resulting from the azimuth oversampling can be reduced through onboard processing. Staggered SAR is the baseline acquisition mode of Tandem-L and under consideration for NISAR and the future missions of ESA’s Copernicus Programme. In a conventional constant PRI SAR nadir echoes also line up on the same ranges and appear in the focused image as bright stripes. However, these bright stripes can be avoided by selecting a PRI that causes the nadir echoes to align with the blind ranges. A drawback of the staggered SAR is that, since the PRI is varying and there are no blind ranges, the nadir echoes cannot be aligned with the blind ranges and will contaminate wider range intervals in the focused data. This raises the questions of how the nadir echo affects a staggered SAR image and what can be done to suppress nadir echoes. To answer the first question, we start by presenting an analytical description of the positioning of the nadir echo in the raw staggered SAR data. We also present a simplified nadir echo model based on a dedicated TerraSAR-X acquisition. With these in hands, a simulator is built capable of generating raw staggered SAR data from a given backscatterer distribution, e.g., a satellite SAR image acquired by an existing SAR sensor, contaminate it with nadir echo and process it into a focused image. The simulator can be run for different nadir echo parameters showing the impact on real scenes and for the constant PRI case, which is considered for comparison. For the second question, we propose a post-processing step based on identifying the samples affected by nadir echo in the range-compressed data and recovering them through interpolation of neighboring azimuth samples similarly as done in the raw data for the samples missing due to transmission. This step is especially challenging, if onboard processing has been performed, as nadir echoes are smeared along azimuth. The results of this work extend the staggered SAR theory and can be used to evaluate the impact of nadir echo in a specific staggered SAR system or to impose requirements on the nadir attenuation provided by the antenna pattern.