River Discharge Estimation Solely from Satellite Imagery and at-Many-Stations Hydraulic Geometry (AMHG)

Abstract:

Knowledge of river discharge is critically important for water resource management, climate modeling, and improved understanding of the global water cycle, yet discharge is poorly known in much of the world. Remote sensing holds promise to mitigate this gap, yet many current approaches for quantitative retrievals of river discharge require in situ calibration or a priori knowledge of river hydraulics, limiting their utility in unmonitored regions. This talk demonstrates a novel remotely sensed discharge retrieval method that requires no prior knowledge and no ancillary data whatsoever. The approach is enabled by a newly discovered river-specific geomorphic scaling phenomenon, termed at-many-stations hydraulic geometry (AMHG), which holds that a river’s paired at-a-station hydraulic geometry (AHG) parameters (a and b, c and f, k and m) are log-linearly related along a river. An associated AMHG discharge retrieval method uses only remotely sensed cross sectional river top width as an input to an unconstrained optimization of width-AHG via a genetic algorithm. Using the AMHG approach, we demonstrate successful retrieval of river discharge to within 20-30% of in situ gauge observations for the Mississippi, Athabasca, and Yangtze rivers. Expanding the method to 34 rivers globally, we find that the AMHG discharge retrieval method is sensitive to river morphology, cross sectional geometry, the quality of input widths, and genetic algorithm optimization parameters. These results suggest that the AMHG discharge retrieval method can meaningfully address global discharge knowledge gaps solely from repeat satellite imagery.