Abstract's details
From unfocused to fully focused SAR processing: illustrations of potential benefits for different surfaces
Event: 2018 Ocean Surface Topography Science Team Meeting
Session: Instrument Processing: Measurement and Retracking
Presentation type: Oral
In the “unfocused” Delay-Doppler processing, as currently implemented in both CryoSat-2 and Sentinel-3 ground segments, an aperture synthesis (i.e. coherent processing of the echoes within a burst) is performed, allowing to narrow the altimeter along-track resolution down to 300m, and then a multi-looking process is applied to reduce speckle noise. In such an approach, the phase of the echoes is exploited over a relatively short period of time (a burst is only a few milliseconds) compared to the illumination time of the surface (typically more than 2 seconds). A SAR processing technique allowing to exploit the whole phase history of the scatterers, called Fully-Focused SAR (FFSAR), has been recently developed by Egido & Smith in [1]. Promising results have been presented by the authors in previous OSTST meetings, with two main improvements: for heterogeneous surfaces (sea-ice, hydrology, coastal), parameter estimation is improved thanks to a better along-track resolution and for homogeneous rough surfaces (ocean), it offers a better noise reduction capability than unfocused SAR.
Based on these findings obtained with CryoSat-2 data, it was deemed interesting to extend this method to Sentinel-3A data. Following the approach used by Egido & Smith, our processing was first tested and validated exploiting ESA Crete transponder overpasses. Then, a specific L1B processing has been implemented in the CNES Sentinel-3 Processing Prototype (S3PP), allowing us to assess its performance over different surfaces.
Preliminary results on hydrology convinced us that this surface could immediately benefit from the FFSAR processing using available S3A data, leading us to put special effort on this topic. Over specular water bodies, the water height was estimated by retracking FFSAR waveforms with a simplified sinc waveform model. Excellent results obtained over a large range of hydrological targets (rivers, lakes, channels) of various sizes will be presented. An assessment has been carried out by comparing water height estimates with in-situ measurement and an innovative lidar system embarked on a drone flying over the surface. This technique clearly opens new amazing perspectives for hydrology from space.
Over ocean, the FFSAR processing has also been investigated. It does bring more efficient noise reduction than unfocused SAR processing. However, many questions are still under investigation. Particularly, we are analyzing the impact of the coherent integration time on FFSAR waveforms (L1) and geophysical parameters (L2).
Based on these findings obtained with CryoSat-2 data, it was deemed interesting to extend this method to Sentinel-3A data. Following the approach used by Egido & Smith, our processing was first tested and validated exploiting ESA Crete transponder overpasses. Then, a specific L1B processing has been implemented in the CNES Sentinel-3 Processing Prototype (S3PP), allowing us to assess its performance over different surfaces.
Preliminary results on hydrology convinced us that this surface could immediately benefit from the FFSAR processing using available S3A data, leading us to put special effort on this topic. Over specular water bodies, the water height was estimated by retracking FFSAR waveforms with a simplified sinc waveform model. Excellent results obtained over a large range of hydrological targets (rivers, lakes, channels) of various sizes will be presented. An assessment has been carried out by comparing water height estimates with in-situ measurement and an innovative lidar system embarked on a drone flying over the surface. This technique clearly opens new amazing perspectives for hydrology from space.
Over ocean, the FFSAR processing has also been investigated. It does bring more efficient noise reduction than unfocused SAR processing. However, many questions are still under investigation. Particularly, we are analyzing the impact of the coherent integration time on FFSAR waveforms (L1) and geophysical parameters (L2).
Contribution: IPC_04_Rieu.pdf (pdf, 2998 ko)
Back to the list of abstract