Abstract's details

Using image processing techniques to detect inland water bodies in Sentinel-6MF Fully Focused SAR radargrams and improve water surface height retrieval: A case study over Garonne River.

Jean-Alexis Daguzé (CLS, France)


Sophie Le Gac (CNES, France); Gabriel Calassou (CLS, France); François Boy (CNES, France)

Event: 2023 Ocean Surface Topography Science Team Meeting

Session: Instrument Processing: Measurement and Retracking

Presentation type: Type Forum only

Contribution: PDF file


Satellite altimetry has emerged as a powerful system for monitoring and understanding Earth's water resources. This technology has revolutionized our ability to monitor not only the oceans but also inland water bodies, including rivers, lakes, and reservoirs worldwide. In recent years, there has been a growing need to retrieve river water levels using altimetry data. Water surface height (WSH) retrieval is usually done by performing waveform retracking algorithms that can be empirical or physically based. This process becomes very critical over inland waters due to complex surface characteristics with respect to the ocean. Indeed, multiple scattering can disturb the radar echoes as the signal interacts with surrounding land areas and water bodies of different sizes, relative altitudes and backscattering properties, making the river water level retrieval a real challenge and sometimes with low accuracy.

Recently, an important step forward has been made with the Sentinel-6MF mission and its on-board Poseidon-4 radar nadir altimeter. Its numerical radar architecture, operating in Open-Burst mode makes it now possible to exploit the full capabilities of the Fully Focused SAR (FFSAR) processing providing very high azimuth resolution (with very low grating lobes) in comparison to the conventional pulse-limited radar and delay/Doppler altimetry [1] which appears to be very useful in differentiating small targets along the satellite track. By combining neighboring waveforms and looking at 2D radar scenes it appears to be possible to see the river features emerging as bright curves from the darker surroundings and local ponds. Considering image processing techniques, this work aims to locate the river signal in the radargrams if the signal-to-noise ratio allows it, depending on the river’s roughness and position from nadir. This useful information is then used in the 1D retracking step to improve water level accuracy. For this purpose, well-known edge operators (e.g., Sobel) are performed on FFSAR pseudo-image and compared. Moreover, a new physically based retracker algorithm, inspired from current studies overs lakes [2], is developed to account for this a priori information and to determine a more accurate water level. Preliminary results and performances analysis will be presented over the Garonne case study for which in-situ observations are also available.

This study shows innovative methods using nadir altimetry and how to exploit the full potential of this technique over hydrology. Our results will also be used in the frame of the SWOT mission Cal/Val and we believe it will be an added-value dataset to validate the river products from the swath altimeter KaRin.

[1] A. Egido and W. H. F. Smith, "Fully Focused SAR Altimetry: Theory and Applications," in IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 1, pp. 392-406, Jan. 2017, doi: 10.1109/TGRS.2016.2607122.

[2] Boy, F. et al, 2021. Improving Sentinel-3 SAR mode processing over lake using numerical simulations. IEEE Transactions on Geoscience and Remote Sensing, 60, pp.1-18.
Jean-Alexis Daguzé