| Popis: |
—To reduce uncertainties in a modelled system, data assimilation strongly relies on the availability of observations, and its performance depends directly on the spatial density, the frequency, and the quality of these observations. Yet, rivers are rather poorly observed. The SWOT (Surface Water and Ocean Topography) mission, to be launched in 2021, is expected to provide global water level observations at a high-resolution coverage for rivers down to 50-meter wide. In order to highlight the merits of these future observations, we compare the performance of an Ensemble Kalman Filter on a 50-kilometre reach of the Garonne (South of France) when only hourly water height gauge measurements are available in the middle of the reach, and when complementary SWOT-like observations are available. A 10-kilometre spatial average with frequencies of 3 and 1 days are tested in the framework of twin experiments. Results show that assimilating the SWOT-like observations allows the ensemble size to be reduced without losing accuracy. With a better correction of the friction coefficients and the upstream discharge, the water height systematic bias is cancelled out and the root mean square error is decreased, i.e. the deviation to the reference is reduced. The beneficial impact of the SWOT-like observations holds in the 12 first hours of the forecast. |
