Using the phase differences between satellite radar images recorded before and after an earthquake, interferometry allows mapping the projection along the line of sight (LOS) of the ground displacement. Acquisitions along multiple LOS theoretically allow deriving the complete deformation vector; however, due to the orbit inclination of current radar satellites, precision is poor in the north–south direction. Moreover, large deformation gradients (e.g., fault ruptures) prevent phase identification and unwrapping and cannot be measured directly by interferometry. Subpixel correlation techniques using the amplitude of the radar images allow measuring such gradients, both in slant-range and in azimuth. In this letter, we use a correlation technique based on the maximization of coherence for a radar pair in interferometric conditions, using the complex nature of the data. In the case of highly coherent areas, this technique allows estimating the relative distortion between images. Applied to ASAR images acquired before and after the December 26, 2003 Bam earthquake (Iran), we show that the near-field information retrieved by this technique is useful to constrain geophysical models. In particular, we confirm that the major gradients of ground displacement do not occur across the known fault scarp but approximately 3 km west of it, and we also estimate directly the amplitude of right lateral slip, while retrieving this value from interferometry requires passing through the use of a model for the earthquake fault and slip.