Implementation of TOPS mode on RADARSAT-2 in Support of the Sentinel-1 Mission
Davidson, Gordon1; Mantle, Vince1; Rabus, Bernhard1; Williams, Dan1; Geudtner, Dirk2

The European Space Agency (ESA) is developing the Sentinel-1 (S-1) constellation of two C-band Synthetic Aperture Radar (SAR) satellites that will provide imagery for the Global Monitoring for Environment and Security (GMES) and other national user services. Interferometric Wide Swath (IW) is the main mode of operations for Sentinel-1, and will operate in the new TOPS (Terrain Observation with Progressive Scans in azimuth) mode, providing a swath width of 250 km. This mode will support the use of SAR interferometry (InSAR) techniques for GMES land applications such as monitoring of surface deformations caused by tectonic activities, glacier flow, and water extraction and mining activities.

Prior to the S-1 launch, there is a need for realistic C-band TOPS image data provided in the correct S-1 product format to support testing of the S-1 Image Processing Facility (IPF) as well as application development. The Canadian RADARSAT-2 mission operates at the same C-band frequency (5.405 GHz) as the Sentinel-1 mission. An experimental RADARSAT-2 TOPS mode, designed to be as similar as possible to the S-1 IW mode, is being developed and tested. The design of this mode makes use of the RADARSAT-2 SAR instrumentís phased array antenna that supports multiple SAR imaging modes, including ScanSAR modes, as well as quad-polarisation and repeat-pass InSAR capabilities. At the symposium, the RADARSAT-2 TOPS mode design, the data acquisitions, the processing, and the analysis of preliminary results will be described.

TOPS mode requires a forward steering of the antenna during the acquisition of a burst. S-1 will be capable of steering in very small angular steps in azimuth time, whereas the RADARSAT-2 TOPS mode has coarser angular steps. A limitation of the step size is the time required to switch from one azimuth angle to the next. Thus, to determine the step size that can be used for RADARSAT-2 TOPS, an in-orbit experiment was conducted to determine how rapidly the beams can be switched. The results indicated an angular step size of approximately 0.02 degrees. During the design phase, simulations showed that with this step size, the net effect of the steps on point target response and radiometry appears to be small to negligible.

The design for the experimental TOPS mode on RADARSAT-2 was based on the existing 'ScanSAR Narrow B' (SCNB) mode, which uses three sub-swaths and covers a comparable range of incidence angles. The following points summarize the design of the RADARSAT-2 TOPS mode from SCNB:

  • To approach the S-1 slant range resolution, the 11.58 MHz pulse is replaced with a 30 MHz pulse, for a nominal 5.0 m slant range resolution.
  • The nominal azimuth scan rate is approximately 1.1 degrees per second, and the nominal burst duration time is approximately 1 second.
  • To maintain azimuth ambiguity ratios, the processed target azimuth bandwidth is approximately 230 Hz, giving an azimuth resolution of about 30 m, and the beam steering angle limits are kept to within +/- 0.6 degrees per burst.
  • The along-track spatial overlap between focused bursts is over 10%.

    Properties of the acquisition of TOPS data that are important for interferometric applications are: 1) the synchronization between bursts of repeated passes; 2) the difference in the Doppler centroid between bursts of repeated passes; and 3) the interferometric baseline between the passes. In order to estimate the expected coherence of bursts due to these factors, a study of existing InSAR stacks of RADARSAT-2 data was made during the design phase. In summary, the 90 percentile values of these properties were found to be: 41 ms burst synchronization, 92 Hz Doppler centroid difference, and 435 m perpendicular baseline. The combination of these effects gives a 90 percentile value of coherence of 0.66. Thus, although RADARSAT-2 was not originally designed for ScanSAR interferometry in terms of burst synchronization, a high proportion of acquired TOPS images are expected to be useful for InSAR analysis. In addition to the design of the mode, modifications to the RADARSAT-2 Ground Segment were made to support the ingest of raw data from the experimental TOPS mode, and the output of range compressed data in a format that can be input to a processor for TOPS data. The TOPS data is processed by a prototype MATLAB processor, as well as the IPF.

    Test scenes of RADARSAT-2 TOPS data for image quality measurement and interferometric analysis are described. These include scenes over rainforest for radiometric measurement, transponders for point target analysis, and calm water for noise floor estimation, as well as InSAR pairs over arid areas of low topography and surface displacement for testing of coherence and interferometric phase. Acquisitions over areas of surface displacement will be continued to form an InSAR stack of sufficient thickness to assess point scatterer phase quality.

    Preliminary results are presented of point target analysis, radiometric analysis, noise floor estimation, coherence, and interferometric phase. The analysis of interferometric phase includes an estimate of phase noise, and an analysis of phase ramps in azimuth and range as well as quadratic terms. The analysis of phase trends attempts to separate the effect of orbit errors from the effect of fine registration errors in TOPS mode. The scanning antenna in TOPS introduces an azimuth-varying Doppler centroid, which results in a phase variation in the presence of azimuth mis-registration. Spectral diversity techniques applied to repeat pass burst interferograms and differential interferograms for the overlap regions of successive bursts are investigated to correct phase trends. Compensating with spectral diversity the more subtle phase artifacts stemming from local variations of azimuth registration due to ground movement in range is also investigated using RADARSAT-2 TOPS repeat data acquired over Antarctic ice streams.