<p>Characteristics of the dynamic rock fragmentation and granular-flow entrainment that occurred in a heavily jointed landslide have been analyzed through a multidimensional framework consisting of the discrete element method (DEM), the landslide-induced field seismic signal, and the UAV image recognition. First, the results of the DEM were calibrated using geometric data of surface particles in the deposit derived from UAV image analysis, with particle size and roundness selected as the calibration factors. Additionally, the calibrated DEM model reveals that the acceleration of rock mass occurs in its initial stage and the first half of the entrainment process, consistent with the analysis result of the seismic signal. Second, it could be found that the characteristic diameter (<i>d</i>_<i>c</i>) and shape parameter (<i>β</i>) of rock fragments decreased rapidly during this period. On the contrary, the particle roundness (<i>R</i>), fractal dimension (<i>D</i>), unbalanced force (<i>Ubf</i>) and bond breakage energy (<i>E</i>_<i>b</i>) continuously increased in this period. These findings from landslide monitoring indicate that the fragmentation of the jointed source rock mass mainly occurs during the initial and first half of the entrainment stage. Furthermore, the evolution of the scraping intensity parameter indicates that two violent entrainment accidents occurred during the landslide movement. Meanwhile, the extremely high particle spin and roundness could be found in the middle depth of the granular flow during the entrainment process. Based on the comprehensive analysis above, we suggest that the drag reduction effect of the landslide movement is caused by two key factors, which include the rock fragmentation and the particle rotational grinding occurring near the contact surface between the sliding mass and the entrainment mass.</p> Graphical abstract <p></p>