Céciles Cazals (CS Group, France); Igor Gejadze (INRAE, France); Isadora Rezende (INRAE, France); Pierre-Olivier Malaterre (INRAE, France); Quittard Dylan (INRAE, France)

SWOT provides, for the first time, dense spatial coverage that enables the observation of continuous water surface profiles along rivers. However, SWOT observations are subject to errors caused by factors such as dark water, specular ringing, and others. As a result, the use of quality flags is necessary to filter the data, which can significantly reduce the volume of usable observations and limit their applicability in various contexts. In this work, we propose an approach for processing SWOT RiverSP observations of water surface elevation and river width. Our method involves several steps tailored to the nature of SWOT data and its revisiting swaths. Initially, a data pre-processing step is applied to remove inconsistencies through outlier detection and smoothing. This is followed by a densification step to fill in missing observations in areas such as the nadir gap or regions affected by dark water, provided that swath overlaps are available. The densification relies on a Bayesian framework that leverages the spatial and temporal variability of water surface profiles, accounting for different flow regimes. A similar approach is also used to densify river width observations. The proposed methodology is validated over CALVAL sites.