Comparing methods to quantify grain-scale sediment structure in gravel-bed rivers

Understanding when gravel moves in river beds is essential for a range of different applications, but is still surprisingly hard to predict. The critical shear stress at which a grain will move depends on its relative size and structure within the bed, and spatial and temporal changes in grain-scale structure are likely to be a major driver of changes in critical shear stress. Consequently grain-structure metrics such as protrusion, pivot angle and contact with any surrounding fine grained matrix are used as parameters in models to predict critical shear stress, and so there is an increasing demand for measurements of these parameters in order to improve our predictive ability. However, we do not have established methods for measuring these parameters, nor do we know whether different methods provide consistent results. Here we present and compare new datasets of sediment structure metrics collected from eight locations in a small gravel-bed stream using three different methods: direct field-based measurements, terrestrial laser scanning (TLS), and computed tomography (CT) scanning. Using each method, we measure metrics including grain size distribution, grain protrusion and fine matrix content. We find that distributions of grain size are consistent between field-based and TLS data, but smaller in CT data. All three methods produce similar distributions of protrusion relative to grain size. There is also some consistency between field and CT measures of fine-grained matrix. However, the identification of similarity also depends on the type of analysis, and an alternative analysis shows less similarity in protrusion and fine-grained matrix between the different methods. Of the three methods, TLS-based approaches have potential to be most easily applied, and our analysis suggests that for grain-size and protrusion they perform as well as the alternative methods. However, they cannot currently be used for measuring fine-grained matrix content.