Estimating sand concentrations using ADCP-based acoustic inversion in a large fluvial system characterized by bi-modal suspended-sediment distributions

Quantifying sediment flux within rivers is a challenge for many disciplines due, mainly, to difficulties inherent to traditional sediment sampling methods. These methods are operationally complex, high cost, and high risk. Additionally, the resulting data provide a low spatial and temporal resolution estimate of the total sediment flux, which has impeded advances in the understanding of the hydro-geomorphic characteristics of rivers. Acoustic technologies have been recognized as a leading tool for increasing the resolution of sediment data by relating their echo intensity level measurements to suspended sediment. Further effort is required to robustly test and develop these techniques across a wide range of conditions found in natural river systems. This article aims to evaluate the application of acoustic inversion techniques using commercially available, down-looking acoustic Doppler current profilers (ADCPs) in quantifying suspended sediment in a large sand bed river with varying bi-modal particle size distributions, wash load and suspended-sand ratios, and water stages. To achieve this objective, suspended sediment was physically sampled along the Paraná River, Argentina, under various hydro-sedimentological regimes. Two ADCPs emitting different sound frequencies were used to simultaneously profile echo intensity level within the water column. Using the sonar equation, calibrations were determined between suspended-sand concentrations and acoustic backscatter to solve the inverse problem. The study also analyzed the roles played by each term of the sonar equation, such as ADCP frequency, power supply, instrument constants, and particle size distributions typically found in sand bed rivers, on sediment attenuation and backscatter. Calibrations were successfully developed between corrected backscatter and suspended-sand concentrations for all sites and ADCP frequencies, resulting in mean suspendedsand concentration estimates within about 40% of the mean sampled concentrations. Noise values, calculated using the sonar equation and sediment sample characteristics, were fairly constant across evaluations, suggesting that they could be applied to other sand bed rivers.