Transfer of solutes into and out of streambeds

Abstract

Laboratory experiments were conducted to determine the mechanisms and rates of bed/stream exchange of non-reactive solutes for beds of medium and fine sand. Experiments were conducted under steady flow conditions with and without sediment transport in a recirculating flume. Flat beds and beds covered with ripples or triangular bedforms were studied. The net mass exchange was determined by measuring concentration changes in the main flow. The penetration of the solute (fluorescent dye) into the bed was also observed visually. Two key exchange mechanisms, 'pumping' and 'turnover', were identified. Pumping is the movement of pore water into and out of the bed due to flows induced by pressure variations over bedforms (ripples and dunes). Turnover occurs as moving bedforms trap and release interstitial fluid. Predictive models based on the details of the exchange processes were developed. A residence time distribution approach was used. The models do not require calibration. Appropriate scaling variables were identified. With stationary bedforms the exchange is strongly influenced by pumping. The predictions of net mass exchange based on models of pumping with periodic bedforms show good agreement with the measured exchange in the initial stages of the experiments (hours to days). The models under-predict the exchange later in the experiments. The deviation is associated with the large-scale and somewhat random features in the penetration of the dye cloud. Such features are expected to influence the net exchange for large time in natural streams. When the bedforms move slowly in relation to the characteristic pore water velocity, turnover can be neglected and pumping dominates. A model based on a random distribution of bedform sizes provides a good prediction of the mass exchange with slowly-moving bedforms. With rapidly-moving bedforms, turnover dominates the exchange at the start of the experiments, when the solute penetration is limited to the maximum bedform scour depth. The scour depth can be predicted well. Later the depth of penetration is greater than the scour depth and the model predictions.

Files

Additional details