Apply Multi-baseline SAR Interferometry on Long Term Space-borne SAR Data for 3-D Reconstruction in Forest and Urban Areas

Abstract:

Multi Baseline Synthetic Aperture Radar (MB SAR) Tomography is a promising extension to traditional SAR interferometry. By coherently combining SAR images acquired from different baseline location, MB SAR Tomography can achieve unprecedentedly full 3-D imaging of volumetric and layover scatters for each SAR cell.Its capability of 3-D reflectivity reconstruction and multiple scatters separation is enormously helpful for different scientific applications in forestry, agriculture , glaciology etc.

However, in order to apply on repeat-pass space borne interferometric dataset, the Fourier Based MB SAR Tomography is generally affected by unsatisfactory imaging quality due to low number of baseline with unequal distribution, atmospheric phase disturbance and temporal decorrelation. In this paper, we propose different signal processing techniques for overcoming these limitations in oder for a better image quality. 1) we develop a robust interpolator to translate the nonuniform greed to uniform one, largely improved the image quality 2) we apply Robust Capon Spectrum Estimation method to improve the resolution and interference of uncertainty in steering matrix. 3) for atmosphere disturbance and radiometric , we select certain flat and known area from image as a estimation for atmospheric offset.

We first test our result in simulated SAR data. Comparing with Fourier based method, the result shows better sidelobe suppression and robustness to unknown multiplicative phase noise. Finally, we test the algorithm using real ALOS PALSAR L-band data, acquired between August 2009 to February 2011 near Harvard Forest Area, MA, USA.