Tomographic SAR Profiling for Buried Target Detection using Multilayer Physical Modelling

Document Type : Original Article

Authors

1 University of Tehran

2 School of Surveying and Geospatial Information Engineering , University of Tehran

10.22059/eoge.2025.403540.1191

Abstract

Tomographic SAR offers a fuller description of volumetric scattering using voxels instead of pixels in SAR. The physics-based formulation incorporates multilayer propagation with Snell refraction and Fresnel transmission, moisture- and frequency-dependent dielectric behavior and attenuation per Hallikainen, and explicit surface and volume terms via the integral equation method (IEM) and Rayleigh theory. Data are focused along a sliding sub-aperture using a kernel that compensates refracted optical path length (OPL) while applying Fresnel and attenuation weights; an FMCW forward model with matched-filter/back-projection completes the chain. We evaluate four experiment classes: TP versus SAR, ideal versus realistic scenes, and controlled sweeps of soil moisture and soil texture. Performance is quantified by full width at half maximum (FWHM) in range and cross-range, peak sidelobe ratio (PSLR), and signal-to-background ratio (SBR). Relative to SAR, TP delivers narrower, more stable peaks and improved PSLR for buried targets. Under realistic conditions, IEM surface roughness elevates sidelobes and Rayleigh volume scattering raises the depth background, yet target localization remains stable. Increasing moisture reduces penetration and contrast, while texture primarily modulates peak width and amplitude through refractive index n and attenuation α. Overall, TP offers a practical middle ground between SAR and TomoSAR: with a single scan and appropriate windowing/sub-aperture design, it recovers an x–z depth profile that mitigates surface/volume ambiguity and improves FWHM, PSLR, and SBR compared with SAR.

Keywords

Main Subjects