Assimilation of SWOT Altimetry Data for Riverine Flood Reanalysis: From Synthetic to Real Data

Floods are one of the most common and devastating natural disasters worldwide. The contribution of remote sensing is important for reducing the impact of flooding both during the event itself and for improving hydrodynamic models by reducing their associated uncertainties. This article presents the innovative capabilities of the Surface Water and Ocean Topography (SWOT) mission, especially its river node products, to enhance the accuracy of riverine flood reanalysis, performed on a 50-km stretch of the Garonne River. The experiments incorporate various data assimilation strategies, based on the ensemble Kalman filter (EnKF), which allows for sequential updates of model parameters based on available observations. The experimental results show that while SWOT data alone offers some improvements, combining it with in-situ water level measurements provides the most accurate representation of flood dynamics, both at gauge stations and along the river. The study also investigates the impact of different SWOT revisit frequencies on the models performance, revealing that assimilating more frequent SWOT observations leads to more reliable flood reanalyses. In the real event, it was demonstrated that the assimilation of SWOT and in-situ data accurately reproduces the water level dynamics, offering promising prospects for future flood monitoring systems. Overall, this study emphasizes the complementary strengths of Earth Observation data in improving the representation of the flood dynamics in the riverbed and the floodplains.

💡 Research Summary

This paper investigates how the Surface Water and Ocean Topography (SWOT) mission’s river node products can be combined with traditional in‑situ water‑level gauges to improve riverine flood reanalysis. The study area is a 50‑km reach of the Garonne River in southwest France, a region historically prone to flooding and equipped with a dense network of hydraulic structures. The authors employ the TELEMAC‑2D hydrodynamic model, discretized with a 41 000‑node mesh, to simulate depth‑averaged flow. Model parameters include a multiplicative factor μ applied to the upstream hydrograph and Strickler friction coefficients K_s for several river‑bed zones.

Data assimilation is performed using an Ensemble Kalman Filter (EnKF), which updates both the model state (water surface elevations) and the uncertain parameters at each assimilation step. Two observation streams are considered: (1) synthetic and real SWOT river‑node water‑surface‑elevation (WSE) observations, and (2) pointwise water‑level measurements from the VigiCrue and Vortex‑io gauge networks. SWOT provides high‑resolution (≈10 m along‑track) WSE profiles over a 120 km swath, with a nominal 21‑day repeat cycle; the authors also explore enhanced revisit scenarios (2–3 day sampling) in an Observing System Simulation Experiment (OSSE).

The OSSE is built on a reference simulation of the 2021 Garonne flood. Synthetic SWOT observations are generated at the times of three passes (passes 42, 113, 391) and are deliberately densified to assess the impact of more frequent satellite overpasses. Results show that even with the nominal 21‑day revisit, assimilating SWOT alone reduces the root‑mean‑square error (RMSE) of water‑level forecasts by roughly 15 %. When SWOT is combined with the in‑situ gauges, the RMSE reduction exceeds 30 %, and the EnKF is able to correct poorly constrained friction coefficients (especially in zones lacking gauge coverage). The experiments also demonstrate that shorter revisit intervals further improve the analysis, as the ensemble covariance matrix becomes better estimated and the filter can capture rapid water‑level changes.

A second set of experiments applies the same framework to the real 2024 flood event, which did not produce overtopping but still exhibited significant in‑channel dynamics. Real SWOT observations contain gaps and higher noise levels; the authors therefore inflate the observation‑error covariance accordingly. Despite these imperfections, the EnKF successfully integrates the satellite data, yielding water‑level time series that closely match gauge records across the reach. Importantly, the combined assimilation corrects the friction parameters in all five Strickler zones, even those without direct gauge data, confirming the spatial information value of SWOT.

The discussion highlights SWOT’s strengths—wide spatial coverage, high spatial resolution, and the ability to observe water‑surface elevations directly—against its limitations, such as relatively long revisit periods and sensitivity to DEM quality in complex riverbanks. The authors suggest that future work should explore multi‑satellite fusion (e.g., Sentinel‑6 SAR altimetry) and the incorporation of updated high‑resolution DEMs to mitigate bias introduced by DEM‑derived observations. They also note that the performance of EnKF is strongly dependent on the observation schedule; therefore, optimal design of satellite overpass timing and error modeling is crucial for operational flood forecasting.

In conclusion, the study demonstrates that assimilating SWOT river‑node altimetry with conventional gauge data via an EnKF framework markedly improves flood reanalysis for the Garonne River. More frequent SWOT observations further enhance model skill, and the combined approach effectively reduces uncertainties in both state variables and hydraulic parameters, especially in poorly observed river sections. These findings provide a solid scientific basis for integrating SWOT data into real‑time flood monitoring and early‑warning systems.