Abstract: The coupled analysis of Computational Fluid Dynamics (CFD) and Discrete Element Method (DEM) has emerged as a critical approach to revealing the mesoscopic mechanisms of soil-water coupling. The volume fractions of the particle phase and fluid phase are key parameters for determining the two-phase interaction in unresolved CFD-DEM coupling analysis, significantly influencing both computational efficiency and the accuracy of calculation results. For widely graded soils with remarkably different particle sizes, it is difficult to balance the calculation of volume fractions for both large and small particles, leading to computational instability and low accuracy. This study proposes an improved distributed point volume fraction algorithm. By allocating volumes to effective distributed points within the particle influence radius, the algorithm transforms grid search into distributed point search, substantially expanding the applicable range of fluid grid size L relative to particle diameter D. Additionally, the introduction of the particle mirror method and satellite point method achieves compensatory volume allocation for large particle boundaries and precise capture of small particle spatial distribution. Static particle arrangement numerical tests were conducted. Based on the calibration of algorithm parameters, the coupling accuracy and efficiency advantages of the distributed point method under different grid conditions were verified. Through turbid water infiltration numerical tests, the calculation accuracy of the distributed point method for particle volumes was further validated in scenarios where both (D ~ L) and (D << L) coexist. Numerical tests indicate that the improved distributed point method shows broad application prospects in large-scale fluid-solid coupling calculations for widely graded soils.