<p>As a kind of geophysical flow in high and cold region, rock-ice avalanches increase in volume and potential impact by eroding and entraining bed material during their movement, thereby posing significant risks to human lives and infrastructure located downstream. This granular process is regulated by the basal stress and its variations at the flow-bed interface. It is imperative to offer a comprehensive understanding of the basal stresses produced by granular flows in order to enhance hazard risk management strategies. In this study, we conducted a series of discrete element method (DEM) simulations of rock-ice avalanches under steady-state conditions to enhance our microscopic understanding of particle-bed interactions. The quantitative indices of basal stress fluctuation, specifically the maximum stress and the standard deviation of stress, as well as the microscopic indices of particle interaction, including the Savage number, granular temperature, and particle free space, are assessed through numerical simulation. The results indicate that as the Savage number increases, the mode of particle interaction with the bed shifts from continuous contact to random collisions, leading to significant fluctuations in basal stress. Furthermore, variations in stress fluctuation are correlated with granular temperature, indicating a dependence on random motion of particles. In conclusion, a microscopic mechanism underlying stress fluctuations is proposed based on particle dynamics. As the macroscopic flow intensifies, the available free space surrounding the particles increases, resulting in an elevated local velocity due to the random motion of the particles, which generate a greater impact force on the bed.</p> Graphical Abstract