Abstract: The vertical distribution of flow velocity in debris flow section is the key to study the flow,impact force and sediment entrainment process. However,in the field measurement of debris flow and flume test,the commonly used stratified current meter only collects the velocity data of a small number of sampling points in the section due to the limitation of the arrangement of measuring devices,which makes it difficult for the linear distribution model based on the regression of measured results to accurately describe the velocity distribution law of debris flow. The paper uses a large-scale flume test of debris flow and a smoothed particle hydrodynamics(SPH)numerical model based on HBP constitutive,and inverts the three-dimensional dynamic process of debris flow. Large amounts of particle velocity data are analyzed and processed by hierarchical statistical algorithm. The vertical distribution characteristics of velocity in the section are obtained. A nonlinear vertical distribution model of flow velocity of debris flow based on logarithmic function is proposed. In order to verify the accuracy of the proposed model,several sets of flume test data are used for comparative analysis. The results show that the proposed logarithmic distribution model can fit the velocity profile more accurately than the traditional linear distribution model,and has better robustness in terms of model parameter sensitivity.