The Rouse equation is a well-known deterministic model for suspended concentrations. However, the transport of sediment particles is influenced by several random variables, such as non-uniform sediment size and turbulence structure. Experiments have demonstrated that the stochastic characteristics of turbulence structure, such as ejections and sweeps, can cause fluctuations in sediment concentrations. A new method is proposed to quantify the probabilistic sediment concentrations. In this study, the multiple-state discrete-time Markov chain and stochastic particle tracking model were used to simulate sediment transport with spatially and temporally varying probabilistic concentrations under the stochastic turbulence structure. Point estimate methods were adopted to estimate the variability of non-uniform sediment sizes. The proposed model was implemented for three cases. In the first case, the proposed model was validated against the experimental data. In the second case, spatial and temporal concentrations at high and low Rouse numbers with mean and non-uniform sediment sizes were compared. The result demonstrates that in the prediction with mean sediment sizes, the sediment concentration is overestimated near the bed, and the advection of the sediment concentration in the x-direction is underestimated. In the last case, higher-order statistical moments of the fluctuating concentrations were estimated through simulations using the proposed model. Simulation results conducted using the proposed method were compared with experimental data. The results revealed that the prediction results based on the proposed model are in good agreement with experimental data.

劳斯方程是悬浮浓度的众所周知的确定性模型。但是，沉积物颗粒的传输受几个随机变量的影响，例如不均匀的沉积物尺寸和湍流结构。实验表明，湍流结构的随机特征（例如喷射和掠掠）会引起沉积物浓度的波动。提出了一种新的方法来量化概率沉积物浓度。在这项研究中，使用多状态离散时间马尔可夫链和随机粒子跟踪模型来模拟在随机湍流结构下具有随时间和空间变化的概率浓度的泥沙运移。采用点估计方法来估计不均匀沉积物尺寸的变化。提出的模型针对三种情况实施。在第一种情况下，针对实验数据验证了所提出的模型。在第二种情况下，比较了在高和低Rouse数下的时空浓度以及平均和非均匀沉积物大小。结果表明，在平均沉积物大小的预测中，沉积物浓度在床附近被高估，而沉积物浓度在x方向上的平流被低估了。在最后一种情况下，通过使用所提出的模型进行仿真来估计波动浓度的高阶统计矩。使用该方法进行的仿真结果与实验数据进行了比较。结果表明，基于该模型的预测结果与实验数据吻合良好。