Innovative radar interferometry
Radar interferometry is an important tool in remote sensing, where two receive antennas separated by a baseline are used to determine the three-dimensional position of reflectors. The research described in this dissertation extends the technology of radar interferometry in several ways. The first requirement for radar interferometry is appropriate data-collection hardware. This dissertation extends the usefulness of synthetic aperture radar (SAR) interferometry through the development of inexpensive instruments. BYU's airborne SAR YSAR demonstrates the feasibility of compact, low-cost SAR. This instrument is described, and data collected at several archaeological sites in Israel is presented. The low-cost radar concept is extended to interferometry by YINSAR, BYLJ's airborne interferometric SAR. The YINSAR instrument design is presented, along with SAR images and interferograms. These instruments are innovative in their use of off-the-shelf components and standard PC parts. The low data cost expands the field of SAR and interferometry to new areas where cost was previously prohibitive. A second crucial requirement for radar interferometry is that the data contain proper phase histories. For interferometric SAR, this often requires autofocus, or the correction of signal phase using the image data itself. A useful method of autofocus for spotlight SAR is the phase gradient autofocus (PGA) algorithm. This research extends the PGA algorithm by introducing a new method of applying it to stripmap SAR data. A replacement for the phase estimation step using the Kalman filter is proposed. The range dependence of the motion-induced phase error in low-altitude SAR is analyzed, and a range-dependent phase estimator is introduced. Another key requirement for useful radar interferometry is an understanding of the expected errors. This dissertation extends this aspect of interferometry through a detailed analysis of stationary ocean interferometry. This new technique is described, and the three-dimensional positioning errors due to parameter errors and thermal noise are derived. A simple approximation to the standard deviation of interferometric phase due to thermal noise is introduced. The mean and variance of the estimated wavenumber spectrum in the presence of height errors are derived, first for a one-dimensional slice of the spectrum and then for the full two-dimensional wavenumber spectrum.