Measures of flow intensity as predictions of bed load

The relative merits of tractive stress, vertically averaged velocity, stream power and unit stream as predictors of bed-load transport rates were assessed. The experiments of Gilbert, Meyer-Peter and G.P. Williams were analysed. Tractive stress, when based on depth and slope, was a poor predictor especially in narrow channels. Elimination of wall and bedform roughness gave grain tractive stress, in good correspondence with the experimental bed-load rates. Vertically averaged velocity was directly comparable with grain tractive stress when corrected for depth and particle size. Stream power was limited as a predictor because it was the product of vertically averaged velocity and bed tractive stress. Unit stream power was a good predictor of concentration in some circumstances. Its poor performance in the data set of Williams is discussed and reasons given.