Statistical accuracy for estimations of large wood blockage in a reservoir environment

The blockage of weirs or bridges by in-stream wood can reduce the flood discharge capacity, leading to hazardous situations. To assess the related risk, blocking probabilities quantifications are needed. However, large wood has a random behaviour and it is challenging to accurately evaluate the blockage process and the influence of different large wood or hydraulic parameters properly. Investigations of large wood processes with physical models have been performed in the past but some contradictions regarding the coherence of the results has been found. Herein, a compromise between statistical accuracy and experimental repetitions is presented. The influence of a different number of experimental repetitions on the accuracy of blocking probabilities estimations of groups of stems in an ogee crested weir equipped with piers was systematically evaluated. Statistically justified numbers of repetitions for achieving maximum standard errors of 0.20 are presented for different semi-congested LW transport regimes. The bootstrap re-sampling technique was applied to generalize the results obtained experimentally. It was found that an increasing number of stems may decrease the randomness of the blockage process. This observation allowed to decrease the number of experimental repetitions needed to achieve equal levels of statistical accuracy compared to individual stem experiments.