Bedload transport fluxes are known to widely fluctuate in both time and space (Ma et al., 2014). The typical scales of fluctuations are broad and often overlap: from random particle motions at the grain scale to large bed morphologies evolving at the river reach scale. Thus, the individual contribution of each physical mechanisms to the total variation of the sediment load is generally hard to evaluate. In this work, we question the utility of Principal Component Analysis (PCA), also called Proper Orthogonal Decomposition (POD), to analyse the sedimentary response of an erodible experimental channel subjected to periodic floods. The method has already demonstrated its great capacities for studying the coherent structures developping in turbulent flows (Berkooz et al., 1993). Its potential to reveal the patterns emerging in sediment transport systems has received much less attention.