Interferometric Processing of Drone-Borne SAR Data Acquired with Highly Non-Linear Tracks

Abstract

Drone-based synthetic aperture radar (SAR) interferometry (InSAR) enables the generation of high-quality digital elevation models (DEMs) of local areas. However, the trajectories of low-altitude drones inevitably deviate from the ideal linear tracks due to, e.g., continuous platform stabilization and wind compensation maneuvers. These high-frequency deviations are larger relative to the flight height than those observed in airborne systems. As a result, the well-established airborne InSAR processing techniques are not directly applicable to drone-borne InSAR. In this paper, we discuss the use of a common-band Doppler filtering scheme that avoids Doppler decorrelation by leveraging a wide azimuth antenna pattern, enabling the use of frequency-domain SAR processing while effectively accounting for rapid Doppler centroid variations. We also introduce an alternative processing scheme to azimuth multi-looking that mitigates artifacts caused by time-variant residual motion errors by averaging DEMs generated from distinct azimuth sub-apertures. The proposed processing strategies are essential for producing very high-quality InSAR products using both repeat-pass and multi-platform drone-borne InSAR.