R. Lanari zyxwvutsrq A S. Hensley P.A. Rosen zyxwvutsrq DCBA Indexing terms: ScanSAR mode, SAR processing, SPECAN algorithm, Chirp z-transform zyxwvu Abstract: zyxwvutsrq DCBA The scan mode synthetic aperture radar (ScanSAR) image impulse response is derived in the time domain, and particular attention is given to the analysis of the phase, which is important for several applications, and especially in interferometric ScanSAR systems. A new algorithm for phase-preserving azimuth focusing of ScanSAR data, that extendsthe basic SPECAN procedure, is presented. The proposed algorithm avoids the interpolatioil step needed to achieve a constant azimuth pixel spacing by replacing the standard Fourier transform used in the SPECAN procedure with an appropriate chirp z-transform. The relationship between the modified SPECAN algorithm and the standard range-Doppler approach is also discussed. Experiments on real and simulated data are carried out to validate the theory. zyxwvu IHGFED 1 Introduction Synthetic aperture radar (SAR) is a coherent imaging sensor, operating on both airborne and spaceborne platforms, that allows the generation of high resolution microwave images. The most common operating mode for a SAR sensor is strip mode, where the radar antenna pointing direc- tion is fixed with respect to the platform flight track (azimuth), and the illumination footprint covers a strip on the ground as the platform moves. In this case the finest azimuth resolution is independent of the sensor- target distance and is achieved by transmitting pulsed signals and coherently adding the backscattered echoes (synthetic antenna formation). In principle, for the strip mode SAR configuration, the extent of the map- ping swath in the along-track direction is arbitrarily long, while the across-track (range) swath is fundamen- tally constrained by the length of time between succes- sive pulses [ 1, 21. If images with large range swaths are required it is possible to overcome the strip mode limitation by using zyxwvu CBA 0 IEE, 1998 IEE Proceedings onhne no. 19982218 Paper first received 18th September 1997 and in revised form 15th June 1998 R. Lanari is with IRECE-CNR, via Diocleziano 328,80124 Napoli, Italy S. Hensley and P.A. Rose11 are with the Jet Propulsion Laboratory, 4800 Oak Grove Drive, Pasadena, CA 91109, USA the scan SAR (ScanSAR) mode [3, 41. In this case the sensor antenna beam is periodically stepped in range to neighbouring swaths, referred to as sub-swaths. As a consequence the overallrangeswath dimension is increased; however, this result is achieved at the expense of the azimuth resolution since full antenna synthesis is no longer possible [3, 41. In this work we derive, in the time domain, the Scan- SAR image impulse response and investigate its charac- teristics. Particular attention is given to the analysis of the phase, which is important for several applications, including interferometry, polarimetry, etc. An immedi- ate application of ScanSAR interferometry is the Shut- tle Radar Topography Mission (SRTM), scheduled to fly in 1999. In this case, to map the entire landmass between 260" of latitude within a 10 day mission, it is necessary to operate the shuttle radar in ScanSAR mode [5]. We also present a new phase-preserving algorithm for azimuth focusing of ScanSAR data that extends the SPECAN procedure [6] and represents a candidate for SRTM. Theproposed approach avoids the range- dependent scaling of the azimuth pixel dimension of the ScanSAR images obtained when applying the basic SPECAN procedure. This result, achieved in the SPE- CAN approach via a post-processing interpolation step, is obtained in our case by simply replacing the standard Fourier transform (FT) used in the SPECAN algorithm with one whose kernel includes a range- dependent correction factor. This transform is imple- mented via a chirp z-transform 171 that can be carried out efficiently via fast Fourier transform (FFT) codes. ScanSAR images generated using the modified SPE- CAN algorithm have a constant azimuth pixel spacing whosedimension can be selected according to the application requirements. T=(x,r,!J) Fig. 1 Reference co-ordinate system S = (x' = zyxwvu vt,, 0) and T = zyxwvu (x. zyxwvu r, zyxwvu I?) are sensor and target locations, respectively 254 IEE Proc.-Radar, Sonar Navig., Vol. 145, No. 5, October 1998