Multi-Target-Multi-Image Technique for SAR Antenna Pattern Estimation
Fabio, Rocca¹; Paola, Rizzoli¹; Christian, Rossi²; Davide, Giudici^1
1Politecnico di Milano; ²aresys

In paper we approach the problem of estimating the Azimuth Antenna Pattern (AAP) form multiple single look complex SAR images, aiming at monitoring the status of active antenna arrays, like ENVISATīs. Accurate estimate of the pattern can be provided by transponders over a very wide angular aperture. However these are quite expensive to be deployed and maintained, (currently only one transponder is maintained by ESA); moreover, they can retrieve only the one-way pattern.

In this paper we investigate the capabilities of deriving the AAP directly from the imaged data, by means of blind deconvolution approach. An estimate of AAP is usual achieved by Power Spectrum analysis on focused data. However this estimate is useless as it is affected by alias and moreover it covers a very small angular interval (depending on the PRF).

In the paper we propose an approach that selects the brightest point scatterers, that have wide angular aperture and are less influenced by clutter. The pattern is estimated by averaging over multiple targets, where the effect of ambiguities are removed by a frequency adaptive weighting, that selects - for each angular interval - the best targets to exploit. A wide angular estimate of the AAP up to 5 lobes is retrieved by processing data in super-resolution mode (like for SPOT). Although the average would provide an effective ambiguity reduction, the result would be biased by the angle-dependent behaviour of the targets cross-section, that could not be separated from the AAP of the sensor. This problem is however solved in a multi-image scenario, where repeat-pass interferometric acquisitions are exploited to derive an estimate of the targets radiation pattern, that is then removed in the final AAP estimate. This estimate is derived by either a debiased and weighted incoherent or coherent average: both are discussed and compared. Finally, some results of multi-target-multi-image AAP estimate based on real ENVISAT datasets are proposed.