Abstract
We carried out a virtual experiment of data assimilation, which assimilates virtual SWOT satellite observation data into global river model CaMa-Flood with LETKF, at the Amazon basin. Assimilating virtual SWOT observed water surface height improved reproducibility of land surface water dynamics such as river discharge, even for simulations using bias corrupted runoff as external forcing. Unlike previous researches which set relatively small scale rivers as target, we set the whole basin of large scale river as target, allowing correction effect to flow down rivers in this research. This caused reproducibility improvement of river discharge, especially at the downstream where catchment area is large. For locations where correction is difficult because of large river discharge comparing with local runoff, discharge was improved by correction effect propagated from river upstream. This research showed that data assimilation method is valid to improve estimation of spatio-temporal land surface water dynamics variability even for large scale rivers.
