Parameter optimization of sediment runoff simulation employing acoustic bedload observation

We developed an automatic parameter optimization system that calibrates sediment runoff simulations to match bedload transport rates measured by A Japanese pipe-microphone in a mountainous catchment in Japan. The sediment runoff simulation incorporates both rainfall-runoff processes and sediment production from off-channel sources (i.e., sediment from hillslopes), combining bedload and suspended load transport modeling. The optimization was conducted in two stages. First, rainfall-runoff parameters, such as soil layer thickness and hydraulic conductivity, were optimized to reproduce observed water discharge. Subsequently, sediment transport-related parameters, including the spatial distribution of riverbed grain size, sediment supply volume from hillslopes, grain size distribution of supplied sediment, and channel width, were optimized to improve the accuracy of sediment discharge predictions. Using the optimized parameter sets, the simulation achieved high predictive performance. The Nash-Sutcliffe Efficiency (NSE) for the water discharge was 0.82, and the NSE for sediment discharge was 0.54, indicating that the model successfully captured both the hydrological and sediment transport dynamics of the target basin. This study demonstrates the feasibility of combining surrogate sediment monitoring with process-based sediment runoff modeling to achieve efficient and accurate parameter calibration, even in complex mountainous environments where sediment production from off-channel sources plays a significant role.