Error propagation for geopositioning from airborne spotlight synthetic aperture radar using the stereo method

Ade Komara Mulyana, Purdue University


The stereo SAR (Synthetic Aperture Radar) technique computes three-dimensional coordinates of ground objects by making use of two quantities that can be derived from a SAR image: range and doppler angle. Due to the flexibility in the image collection requirements, this is an attractive alternative to interferometric SAR. Applying stereo SAR to stripmap SAR images, however, is known to produce coordinates with only modest accuracy. Another mode of SAR, the spotlight mode, produces SAR images with a superior resolution (at the expense of smaller coverage). Together with the more advanced navigation systems available today, this makes applying the stereo technique to airborne spotlight SAR images an interesting topic of study. An error model for stereo spotlight SAR in the form of the precision for the observations is developed. The precision of the navigation data is derived directly from the performance description of available navigation systems. The error in the range and doppler angle is derived from the analysis of the image formation process applied to real spotlight SAR data, including the autofocus process. An error analysis for stereo SAR is performed based on a covariance analysis study. The impact of navigation data quality, different flight trajectories, and different distances to the scene, on the precision of the computed ground coordinates are evaluated. The analysis is done using simulated spotlight SAR images of discrete point objects. Ground coordinates with Circular (Horizontal) Error at 0.9 probability (CE90) on the order of 1–2 m and Linear (Vertical) Error at 0.9 probability (LE90) of 1 m are possible to achieve from a medium distance of about 6 km. From a longer distance of about 50 km, CE90 of about 6 m is obtained from a reasonable flight configuration. The results also reveal the necessity to determine the direction of the velocity vector precisely. Each component of the velocity vector must be determined to better than 10 cm/sec. These results show that high quality SAR images produced by the spotlight mode combined with high accuracy navigation data available from the current system can indeed produce high precision ground coordinates.




Bethel, Purdue University.

Subject Area

Civil engineering|Remote sensing

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server