Variationnal data assimilation of AirSWOT and SWOT data into the 2D shallow water model Dassflow, method and test case on the Garonne river (France)

摘要

For river hydraulic studies, water level measurements are fundamental information, yet they are currently mostly provided by gauging stations mostly located on the main river channel. That is why they are sparsely distributed in space and can have gaps in their time series (because of floods damages on sensors or sensors failures). These issues can be compensated by remote sensing data, which have considerably contributed to improve the observation of physical processes in hydrology and hydraulics in general and, in particular, in flood hydrodynamic. Indeed, the new generation of satellites is equipped with sensors of metric resolution. Remotely-sensed images from satellites such as SWOT (Surface Water and Ocean Topography) would give spatially distributed information on water elevations with a high accuracy (able to observe river wider than100m with a vertical precision ~dm) and periodic in time (revisiting ~week at mid-latitude). Gathering pre-mission data over specific and varied science targets is the purpose of the AirSWOT airborne campaign in order to implement and test SWOT products retrieval algorithms. A reach of the Garonne River, downstream of Toulouse (FRANCE), is a proposed study area for AirSWOT flights. This choice is motivated by previous hydraulic and thermal studies (Larnier et al., 2010) already performed on this section of 100km reach of the river. Moreover, on this highly instrumented and studied portion of river many typical free surface flow modelling issue has been encountered, and this river reach represents the limit of SWOT observation capability. The 2DH (vertically integrated) free surface flow model Dassflow (Honnorat et al., 2005; Honnorat et al., 2007; Honnorat et al., 2009; Hostache et al., 2010; Lai and Monnier, 2009) especially designed for variational data assimilation, will be used on this portion of the Garonne River. Mathematical methodologies such as twin experiments (Roux and Dartus, 2005; Roux and Dartus, 2006) will be performed on several modelling hypothesis in order to identify main characteristic of the river. An identification strategy would allow to retrieve spatial roughness along the main channel, variation of the local topographic slope or else temporal evolution of the streamflow.ud