QUANTIFYING BEDLOAD TRANSPORT IN EPHEMERAL CHANNELS USING SEISMIC METHODS

The transport of sediment is one of the fundamental geomorphic processes governing the evolution of landscapes. Reliable sediment flux forecasts are necessary for a variety of applications such as sedimentation engineering, river restoration, and flood risk mitigation. Quantifying bedload driven by flood events in ephemeral channels is notoriously difficult because of the scarcity, irregular nature, and high intensity of flash floods. Seismic methods appear to be a promising tool to characterize such fluvial processes, as they continuously record the ground motions caused by bedload and water movement while located outside of the active channel. We evaluated the performance of the physics-based model estimates of bedload fluxes developed by Tsai et al. (2012) by comparison to continuous monitoring of bedload measurements. The model establishes a mathematical relationship for the power spectral density (PSD) of the Rayleigh waves produced by vertically impulsive impacts from saltating particles based on the rate of impacts of fluvial sediment for a given bedload flux and grain size distribution. As a test of this model, we collected seismic data during flow events and compared the seismically-estimated bedload flux with high-precision bedload flux observations. These data derive from a multi-year campaign of monitoring an ephemeral, sand-and-gravel bedded channel reach of the Arroyo de los Pinos, central New Mexico, USA. Based on seismic data analysis, we find that bedload transport correlates to signals in the 30-80 Hz frequency range, whereas rainfall correlates to signals above 100 Hz. Inverting seismic data for bedload fluxes using the vertical impact model results in overestimates of the observed bedload flux by ~2 orders of magnitude. We investigate three hypotheses that may explain this discrepancy. First, the process of rolling and/or sliding particles, as opposed to saltating particles, may be the predominant cause of model discrepancy. Rolling particles are perhaps a very significant contributor to bedload at this study site. Second, the fine-grained alluvial characteristics of this riverbed, as contrasted to a rigid bedrock substratum used in the model, lead to significant attenuation of seismic energy as a result of the inelastic impact of bedload particles. Third, the bedload impact frequency model may not fully depict the impact of particles onto the riverbed in this environment. By thoroughly examining bedload transport mechanisms and considering alternative impulse functions for seismic noise generation, we intend to construct a new physics-based model within the framework of the existing models to quantify bedload transport in the ephemeral environment.