A physical model for seismic noise generation by turbulent flow in rivers

Previous studies suggest that the seismic noise induced by rivers may be used to infer river transport properties, and previous theoretical work showed that bedload sediment flux can be inverted from seismic data. However, the lack of a theoretical framework relating water flow to seismic noise prevents these studies from providing accurate bedload fluxes and quantitative information on flow processes. Here we propose a forward model of seismic noise caused by turbulent flow. In agreement with previous observations, modeled turbulent flow-induced noise operates at lower frequencies than bedload-induced noise. Moreover, the differences in the spectral signatures of turbulent flow-induced and bedload-induced forces at the riverbed are significant enough that these two processes can be characterized independently using seismic records acquired at various distances from the river. In cases with isolated turbulent flow noise, we suggest that riverbed stress can be inverted. Finally, we validate our model by comparing predictions to previously reported observations. We show that our model captures the spectral peak located around 6-7Hz and previously attributed to water flow at Hance Rapids in the Colorado River (United States); we also show that turbulent flow causes a significant part of the seismic noise recorded at the Trisuli River in Nepal, which reveals that the hysteresis curve previously reported there does not solely include bedload, but is also largely influenced by turbulent flow-induced noise. We expect the framework presented here to be useful to invert realistic bedload fluxes by enabling the removal of the turbulent flow contribution from seismic data.