Flow resistance estimation in mountain rivers.

Abstract

Three models of flow resistance for granular boundary have been calibrated, validated, and compared, these being the power law and two other models developed for large-scale roughness conditions (one based on a modification of the Prandtl-von Karman logarithmic law and another one based on a velocity profile split into two zones [a uniform one in the proximity of the roughness elements and another one situated above the former that follows a logarithmic distribution]). A large set (N = 1,533) of data taken from rivers and flumes was used, representative of a wide hydraulic and geomorphologic range in the field of gravel-bed rivers and mountain streams. The equations fitted to the coarser percentiles (d90 or d84) were preferable to those fitted to the median diameter (d50). This was because the higher the explanatory power reached by a given model, the lesser the difference in the goodness-of-fit between the different models and the lower the influence of the origin of the data (river or flume). The equations fitted in accordance with the models that envisage large-scale roughness conditions give similar predictions, with the exception of the interval of highest roughness (y/d90 < 1) in which the equation corresponding to the model based on the velocity profile configured into two zones is preferable. It is recommended to restrict the application of the equation fitted in accordance with the power law to the y/d90 interval between 1 and 20, since outside of this interval, it tends to underestimate the flow resistance notably.