Estimating Sensor Motion in Airborne SAR

Research output: Book/ReportPh.D. thesis

3176 Downloads (Pure)

Abstract

In airborne synthetic aperture radar (SAR), accurate knowledge of the SAR sensor
motion is necessary to achieve an acceptable image quality with respect to resolution, image artifacts and geometrical distortions. For repeat-track interferometry, where SAR images acquired from different passes are combined, the absolute accuracy requirements are even more stringent, as small residual errors will be misinterpreted as scene displacements or topography.

This thesis deals with two aspects of motion estimation in airborne SAR. The first
part is an examination of the impact of propeller aircraft vibrations on SAR focusing. Uncompensated high-frequency motion leads to sidelobes (false echoes) in SAR images. Motion measurements from two propeller aircraft are presented and evaluated with respect to their high-frequency content, and the impact of the measured motion on the image quality is predicted. The motion measurements are compared to similar measurements on a jet aircraft. It is shown that narrowband vibrations at harmonics of the propeller frequency appear in the motion data for the propeller aircraft, and that vibration levels for the jet are significantly lower. It is predicted that the levels of the measured physical vibrations do not lead to unacceptable sidelobes if left uncompensated, but aliasing of the vibrations in the INU can induce spurious low-frequency vibrations at the aliased frequencies in the navigation data. This leads to incorrect motion compensation, and since the aliased vibrations can have higher levels than the actual vibrations due to the integration processes in the INU, unacceptable sidelobes may result.

The second part of the thesis deals with correction of differential motion errors in airborne repeat track SAR interferometry. The work extends an existing differential motion estimation algorithm that integrates the azimuth misregistration to obtain an estimate of the differential motion error between the images. In this thesis, the algorithm is analysed, and potential error sources are identified and examined. One error source is azimuth misregistrations from performing motion compensation with unknown topography. A simple method for reducing topography-induced errors using an external DEM is proposed, and it seems to slightly improve the motion estimate on the L-band EMISAR data for which it was tested.

Another error source in the residual motion algorithm is along-track residual motion errors, which lead to azimuth misregistrations that can be interpreted wrongly as vertical velocity errors. A method is proposed for estimating and correcting these errrors using an external DEM or - as is possible in some cases - the residual range misregistration. The proposed method is tested on a 100 km C-band scene, for which the observed residual cross-track error was still on the order of 15 cm after the initial correction. This variation was reduced to below 2 cm over a 60 km long strip in azimuth.
Original languageEnglish
Place of PublicationKgs. Lyngby
PublisherTechnical University of Denmark
Number of pages115
ISBN (Print)87-91184-59-2
Publication statusPublished - Sep 2006

Cite this

Kusk, A. (2006). Estimating Sensor Motion in Airborne SAR. Technical University of Denmark.