Estimating river bathymetry from data assimilation of synthetic SWOT measurements

Abstract

Summary This paper focuses on estimating river bathymetry for retrieving river discharge from the upcoming Surface Water and Ocean Topography (SWOT) satellite mission using a data assimilation algorithm coupled with a hydrodynamic model. The SWOT observations will include water surface elevation (WSE), its spatial and temporal derivatives, and inundated area. We assimilated synthetic SWOT observations into the LISFLOOD-FP hydrodynamic model using a local ensemble batch smoother (LEnBS), simultaneously estimating river bathymetry and flow depth. SWOT observations were obtained by sampling a “true” LISFLOOD-FP simulation based on the SWOT instrument design; the “true” discharge boundary condition was derived from USGS gages. The first-guess discharge boundary conditions were produced by the Variable Infiltration Capacity model, with discharge uncertainty controlled via precipitation uncertainty. First-guess estimates of bathymetry were derived from SWOT observations assuming a uniform spatial depth; bathymetric variability was modeled using an exponential correlation function. Thus, discharge and bathymetry errors were modeled realistically. The LEnBS recovered the bathymetry from SWOT observations with 0.52 m reach-average root mean square error (RMSE), which was 67.8% less than the first-guess RMSE. The RMSE of bathymetry estimates decreased sequentially as more SWOT observations were used in the estimate; we illustrate sequential processing of 6 months of SWOT observations. The better estimates of bathymetry lead to improved discharge estimates. The normalized RMSE of the river discharge estimates was 10.5%, 71.2% less than the first-guess error.