Assimilation of virtual wide swath altimetry to improve Arctic river modeling

Abstract Global surface water variations are still difficult to monitor with current satellite measurements. The future Surface Water and Ocean Topography (SWOT) mission is designed to address this issue. Its main payload will be a wide swath altimeter which will provide maps of water surface elevations between 78°S and 78°N over a 120 km swath. This study aims to combine coupled hydrologic/hydraulic modeling of an Arctic river with virtual SWOT observations using a local ensemble Kalman smoother to characterize river water depth variations. We assumed that modeling errors are only due to uncertainties in atmospheric forcing fields (precipitation and air temperature) and different SWOT orbits were tested. First, we tested orbits that all have a three day repeat period but differ in terms of their spatial coverage of the study reach; these orbits correspond to the first three months of the mission, which will be dedicated to calibration and validation experiments. For these orbits, the mean spatial Root Mean Square Error (RMSE) in modeled channel water depth decreased by between 29% and 79% compared to the modeled RMSE with no assimilation, depending on the spatial coverage. The corresponding mean temporal RMSE decrease was between 54% and 91%. We then tested the nominal orbit with a twenty two day repeat period which will be used during the remaining lifetime of the mission. Unlike the three day repeat orbits, this orbit will observe all continental surfaces (except Antartica and the northern part of Greenland) during one repeat period. The assimilation of SWOT observations computed with this nominal orbit into the hydraulic model leads to a decrease of 59% and 66% in the mean spatial and temporal RMSE in modeled channel water depth, respectively. These results show the huge potential of the future SWOT mission for land surface hydrology, especially at high latitudes which will be very well sampled during one orbit repeat period. Still, further work is needed to reduce current modeling uncertainties and to better characterize SWOT measurement errors.