Abstract: study of landslide kinematics and impact intensity is of great significance for quantitative assessment of landslide risk. In this paper, the basic characteristics of Mabian landslide(occurred on May 5, 2018) in Leshan City, Sichuan Province were investigated. The support vector machine(SVM) model and particle flow method(PFC) were used to calibrate the meso-strength parameters of landslide rock and soil. Combined with the UAV data, the high-precision DEM of landslide area was generated. On this basis, the three-dimensional PFC model of Mabian landslide was reconstructed and simulated. The movement, accumulation and impact process of landslide are studied. The results show that: the movement of Mabian landslide lasted 32 seconds, main sliding time is 16 seconds, and the peak velocity is 10.2 m/s after 5 seconds; the movement trace of rock and soil in the middle and rear of the landslide is linear, while the middle and front of the landslide is in a diffusion state, and finally in a fan-shaped accumulation; the impact force of the landslide at the slope foot can reach 1.5×10⁹ N, and the impact force presents the exponential attenuation characteristics with the increase of the movement distance. The results are basically consistent with the actual video interpretation results of landslide movement and accumulation, and the related research methods could provide references for quantitative risk assessment of landslide.