Uncertainty discussions about the bedload transport rate estimation using ADCP data and comparison with the physical samples

Measuring and assessing bedload data is crucial for successful and efficient river management. Hence, understanding bedload transport and its characteristics provides insights into river morphology dynamics and aids in evaluating the impacts on boat navigation, hydropower production, ecological systems and aquatic habitat.

Acoustic Doppler Current Profilers (ADCPs) have been widely utilized for measuring bedload characteristics, with reported correlations between apparent bedload velocity, backscatter strength, and physically measured samples. To estimate bedload transport rates, three primary approaches are employed: (i) the kinematic model, which utilizes semi-empirical equations; (ii) multi-regression calibration tailored to different grain sizes; and (iii) machine learning techniques that utilize multiple features derived from ADCP outputs.

Among these, the kinematic model is the most commonly used and exists in two variations: one utilizing virtual particle velocity and the other relying on volume mass conservation. Although ADCP-estimated bedload transport rates are widely used, they are often accompanied by substantial uncertainty and error. Furthermore, the potential errors introduced by physical sampling methods and their impact on comparisons with ADCP-derived estimates have received little attention.

This study addresses these gaps by examining errors associated with both ADCP-based approaches and physical sampling techniques used to estimate bedload transport rates. It further evaluates how these errors interact under varying sediment transport conditions, offering insights into the reliability and limitations of ADCP measurements in comparison to traditional sampling methods.

The initial results indicate that the primary sources of error in the kinematic model stem from secondary parameters that are empirically derived, such as bedload concentration and active layer thickness. Furthermore, weak transport conditions are significantly overestimated by ADCP measurements, highlighting the limitations of the method under conditions of low-intensity and highly non-homogeneous transport.

Furthermore, a kinematic approach, which relies on the average virtual velocity of a monogranular loose bed, also raises questions regarding its reliability considering the ACDCP measuring capabilities, particularly when direct comparisons are made with physical samples. Additionally, errors associated with physical bedload sampling (e.g., bedload rate underestimation due to trap misalignment to sediment flux) are considered.

This study critically examines these methodological issues, with specific attention to the discrepancies that arise when comparing ADCP-based to traditional measurement techniques eventually providing a comprehensive evaluation of the uncertainties inherent in both.