How Well Can We Estimate River Discharge Based on the Surface Water and Ocean Topography (SWOT) Mission Observations? The Ungaged Basin Scenario.

With its Ka-band Radar interferometer, SWOT will collect spatially continuous observations of rivers wider than 100 m, possibly as narrow as 50 m, over two 50 km wide swaths located on both sides of its flight path. In anticipation of the SWOT mission, a series of algorithms were designed to estimate river discharge based on the satellite's measurements of water surface height, slope, and inundation extents. Here, we evaluate how well such algorithms will perform when provided only with synthetic SWOT observations and an initial estimate of discharge coming from globally available hydrological simulations in the absence of in-situ measurements. For this experiment, we created synthetic SWOT observations based on hydraulic simulations of 22 rivers, which yielded 32 test cases, covering a range of discharges from 1 m³/s to 80,000 m³/s. In order to isolate the impact of inversion uncertainty from the effects of limited temporal sampling and measurement uncertainty, we created an experiment with three phases: 1- Daily sampled heights, widths, and slopes with no measurement error; 2- Degraded time sampling assuming observations every 2, 3, 4, 5, 7, 10, 21 days; 3- Daily sampled measurements considering three levels of uncertainties: half the expected SWOT uncertainty, full uncertainty level, and 1.5 times the expected SWOT uncertainty. In phase 1, we identified four important factors for the performance of the algorithms: biases in the initial estimate of mean annual flow, hydraulic variability among reaches, violations of mass conservation, and flow law errors. Phase 2 showed that although performance degraded as time sampling became sparser, discharge errors did not increase substantially until revisit periods exceeded 7 to 10 days depending on the discharge algorithm. Finally, phase 3 showed encouraging robustness to measurement error, including simulated biases, even in the most pessimistic scenario. Our study gives an indication of the level of maturity of the algorithms as well as highlights the challenges that the Discharge Algorithm Working Group must overcome to produce consistently good discharge estimates over the SWOT observable rivers.