Theory of data‐driven SAR autofocus to compensate for refraction effects

Abstract

There are challenges in attempting to acquire sufficient atmospheric measurements that are required in order to model the physical atmospheric refraction effects for all transmission pulses within a typical synthetic aperture radar (SAR) collection interval. In particular, atmospheric turbulence can induce spatial variations of the air density, temperature, and other physical variables along the synthetic aperture, which can have differing effects on the local refraction-induced bending and delay of each of the transmission pulses. The primary goal of the current investigation is to develop an alternate approach based on data-driven techniques in order to develop methods for automatically correcting such refraction-induced defocus effects on SAR image data. This alternate methodology applies two-dimensional maximum likelihood signal-theoretic techniques in order to estimate and compensate for the refraction-induced phase errors of the input defocused SAR imagery, thus yielding improved scene refocus wherein the majority of the refraction-induced defocusing effects have been corrected. The efficacy of this data-driven approach is demonstrated through the injection of known bending and delay effects applied to measured Ku-band SAR data.