Faraday Rotation Estimation from Unfocussed Raw data:
Analysis using ALOS PALSAR Data

Marco Lavalle, Domenico Solimini, Eric Pottier, and Nuno Miranda

Abstract

A radio wave propagating through the ionosphere undergoes Faraday rotation (FR), i.e. the rotation of its polarization plane. Faraday rotation is caused by the anisotropy of the ionospheric tenuous plasma in presence of a persistent magnetic field and can significantly impact the performance of a Synthetic Aperture Radar (SAR). Propagation in an anisotropic medium is not reciprocal, hence sigma-HV deviates from sigma-VH, causes errors in the estimation of polarimetric calibration parameters and therefore impacts current PolSAR/PolInSAR applications. SAR sensors such as ALOS-PALSAR that operate at L-band are more affected by FR than higher frequency SAR systems.

Once detected and estimated, FR must be compensated over the SAR scene. Apart from the use of reference point targets, such as trihedral corner reflectors, Faraday Rotation can be estimated by three different approaches:

- from full-pol single-look complex (SLC) data by considering the difference between the cross-polarized acquisitions HV and VH;

- from full-pol SLC data by simulating the SAR return of a circularly polarized wave;

- from model prediction using real measurements of total electron content (TEC).

The first two approaches have high spatial accuracy because are applied to each SLC sample, but they rely on the phase and cross-talk calibration of the SLC data. The second approach makes use of an external information (TEC maps) that usually has lower spatial resolution. The estimation of FR for PALSAR indicates fairly low values (lower than 8 deg), acceptable for the present polarimetric applications. However, this value might increase in the next years as a consequence of the cyclic solar activity that will increase the average TEC in the ionosphere.

In this paper, we argue that the estimation of the Faraday rotation from SAR products is more appropriate using unfocussed raw data than SLC data. Since the FR is estimated from a single received echo, this approach is more physical with respect to the usual approach. We have carried out an extensive analysis over about 30 polarimetric PALSAR raw data products (L1.0 products) acquired between 7 March 2007 and 7 June 2007. We estimated the Faraday rotation before and after focussing, using the first two approaches above, i.e. the difference between cross-polarized returns and circular basis change. The results of our analysis are finally compared with FR rotation estimated by TEC maps

Index Terms—Faraday rotation. L-band SAR. Polarimetry. SLC. ALOS. PalSAR

POLinSAR 2009, the 4th International Workshop on Science and Applications of SAR Polarimetry and Polarimetric Interferometry, ESA/ESRIN, Frascati, Italy, 26-30 January 2009.

© Authors, 2009