Scaling up the assessment of the SWOT discharge inversion algorithm to thousands of gauges globall

One important goal of the Surface Water and Ocean Topography (SWOT) satellite mission is to estimate global river discharges from observations of river width, height, and slope. While a range of algorithms have been developed and intercompared for SWOT (e.g., Durand et al. 2016), our understanding of the algorithm accuracy has been confined to tens of rivers globally, due to the limited SWOT-like observations currently collected from hydraulic model outputs.

To scale up the assessment of SWOT discharge algorithms, this study will first collect discharge observations at thousands of global gauges whose river widths are wider than 50 m (i.e., observable by SWOT), to provide the most comprehensive observations to evaluate discharge estimations. Then at those gauges, all available Landsat images from 2010 to 2017 (8 years) will be collected to extract river widths with an automatic Google Earth Engine tool called RivWidthCloud (Yang et al. 2019). The extracted river width time series (temporally intermittent) will provide SWOT-like observations, which can be used to derive discharge using the Bayesian AMHG-Manning (BAM) algorithm (Hagemann et al. 2017; Feng et al. 2019). The prior discharge information needed by the BAM algorithm will come from an updated global discharge modeling database (Lin et al. 2019). These datasets, collectively, will provide a critical assessment of the SWOT discharge algorithms.

This study is expected to provide the first geographically explicit assessment of the BAM algorithm at thousands of global locations, and the insights gained may also help the global hydrologic modeling community with their data assimilation efforts.