Abstract's details
Monitoring smaller water structures using SWOT L2 HR Pixel Cloud
Event: 2025 SWOT Science Team Meeting
Session: Hydrology: HR SWOT Data (Data Validation & Enhancement)
Presentation type: Oral
While the SWOT mission's scientific goals for hydrology aimed for rivers wider than 50 meters and water bodies larger than 100 meters by 100 meters, the first data showed that much smaller structures were visible and that it was possible to estimate their height. This has provided new opportunities, both for monitoring a larger number of small rivers and natural lakes and for monitoring human-operated water infrastructure. This last application responds to particularly critical issues in areas where water is a scarce resource and where the available data is insufficient on the volumes of water stored in small (less than a few hectares) reservoirs and the flow circulation through canals.
This ability to estimate the water elevation for small objects in water still has limitations related to signal level (dark water due to very smooth water, particularly in sheltered areas) and the presence of man-made structures with high reflectivity in the environment. The resolution of SWOT images also does not allow for precise estimation of the extent of the water in very small objects. For applications that require this information, it must be obtained from another source or derived from the water height.
As these very small objects are typically absent from the SWOT prior Lake and River databases, their monitoring requires using the Level-2 HR Pixel Cloud product. To this end, we have developed pre-processing methods to filter the pixel cloud points corresponding to the object of interest. These methods are based on geometric information and pixel properties. The points thus extracted can, for example, be used to derive the storage change time series for a small reservoir, or the temporal evolution of the slope in a small channel.
An implementation of the Pixel Cloud preprocessing and filtering methods, as well as some case studies, are now available in the PixieDust library : https://github.com/SWOT-community/PixCDust
These filtration methods could still be improved by a better understanding of the phenomenology of the observation and, in particular, of the signal coming from the surrounding areas. Another promising direction toward the extraction of water elevation in even smaller objects may be to improve the processing from the SLC data, for example through an early detection of the small water structures and a dedicated phase unwrapping step.
Finally, a key priority for the adoption of Wide Swath Altimetry data for monitoring small water structures, both for scientific or public water management applications, will be to improve the propagation of uncertainties from pixel cloud data to the final information.
Back to the list of abstractThis ability to estimate the water elevation for small objects in water still has limitations related to signal level (dark water due to very smooth water, particularly in sheltered areas) and the presence of man-made structures with high reflectivity in the environment. The resolution of SWOT images also does not allow for precise estimation of the extent of the water in very small objects. For applications that require this information, it must be obtained from another source or derived from the water height.
As these very small objects are typically absent from the SWOT prior Lake and River databases, their monitoring requires using the Level-2 HR Pixel Cloud product. To this end, we have developed pre-processing methods to filter the pixel cloud points corresponding to the object of interest. These methods are based on geometric information and pixel properties. The points thus extracted can, for example, be used to derive the storage change time series for a small reservoir, or the temporal evolution of the slope in a small channel.
An implementation of the Pixel Cloud preprocessing and filtering methods, as well as some case studies, are now available in the PixieDust library : https://github.com/SWOT-community/PixCDust
These filtration methods could still be improved by a better understanding of the phenomenology of the observation and, in particular, of the signal coming from the surrounding areas. Another promising direction toward the extraction of water elevation in even smaller objects may be to improve the processing from the SLC data, for example through an early detection of the small water structures and a dedicated phase unwrapping step.
Finally, a key priority for the adoption of Wide Swath Altimetry data for monitoring small water structures, both for scientific or public water management applications, will be to improve the propagation of uncertainties from pixel cloud data to the final information.