IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE Stack of SAR Images or SLC Values at a Given Pixel or Tomographic Inversion Phase Model: Thermal Dilation Slow, Uniform Motion Height * Spatially Regularized Inversion * Sparse Inversion (Compressed Sensing) * Hierarchical Scatterer Detection * Beamforming, Capon * Subspace Methods (MUSIC, WSF, and so on) 2 FIGURE 3. A summary of the approaches for synthetic aperture radar tomography. WSF: weighted subspace fitting. * Subpixel Registration * Subtraction of the Phase of the Master Image * Compensation of the Atmospheric Phase 1 Preprocessing Stack of SAR Images Input Data Covariance Matrix at a Given Pixel Power h (Radar Coordinates) 3D Distribution h of the Reflectivity (Stack of or 1D Estimations) Surface or 3D Distribution of the Reflectivity y z h h Range Range 3D Point Cloud (Ground Coordinates) z Power * Geometrical Transformation → Ground Coordinates * Strong/Permanent Scatterer Extraction * Surface Extraction (Plane Fitting, Regularized Surface) 3 Postprocessing Sparse 1D Profile 1D Profile Power Output Data Elevation Elevation JUNE 2020 23 r r