Abstract: Abstract: The purpose of this study is to investigate the temporal and spatial migration of sediment in the sand filter column. Three kinds of sand filter columns (0.90-1.25, >1.25-1.60, and>1.60-2.0mm) were selected as the research object, and the sand filter layer was used as the fractal resistance model. The maximum pore diameters of the sand filter layers were 0.99, 1.14, and 1.41, and the average pore diameters were 0.41, 0.68, and 0.83. The Yellow River sediment of the People's Victory Canal was used as the raw water sediment impurity particles, where 0.4‰ muddy water was configured. The filtration rate was 2m3/h. After that, the filter column was removed to separate from the sediment during filtration. The quality of the sediment was calculated by the trapped filter layer of the three sand filter columns at different depths. The particle size distribution of sediment and sediment trapped by the filter column was measured by laser diffraction instrument. The results showed that the ratio of sediment intercepted by three filter columns was 97.08%, 94.53%, and 90.5%, respectively. The sediment intercepted by the surface filter layer accounted for 97.88%, 96.36%, and 81.48% of the total sediment intercepted by the filter column. Therefore, the smaller the particle size range of the filter column was, the larger the sediment retention ratio was, and the better the filtration effect was, but the more serious the surface filtration phenomenon was. An analysis was also made on the particle size distribution of the three filter layers at different depths. The D3 of the 0.90-1.25 sand filter column from the upper layer to the lower layer was 4.47, 1.00, 0.90, and 0.91μm. D98 was 476.30, 111.40, 98.31, and 97.79 μm. Span was 1.82, 5.13, 5.47, and 5.56; From the upper layer to the lower layer, D3 of ＞1.25-1.60 sand filter column was 3.77, 1.54, 0.97, and 0.99μm. D98 was 460.20, 344.20, 107.60, and 298.20μm. The Span was 1.95, 3.13, 4.50, and 4.47, and the D3 of ＞1.60-2.0 mm sand filter column from the upper layer to lower layer was 4.33, 3.50, 2.43, and 1.45 μm. D98 was 463.90, 456.00, 415.40, and 215.20 μm. The Span was 1.82, 1.86, 2.38, and 2.59. Consequently, the larger the filter column toward the lower layer was, the smaller the sediment particle size was, where the particle size tended to be concentrated. Each filter column was in the same depth of the filter layer, the larger the filter particle size was, the larger the particle size was, and the larger the particle size span was. Finally, the loss of sediment mass and particle size distribution were calculated using the sediment mass and particle size distribution intercepted by the sand filter column. Specifically, about 50% of the clay and silt particles of the original soil flew away in the process of the three sand filter columns, whereas, only 3% of the fine, medium and coarse sand of the original soil was lost. According to the hydraulic test, the smaller the particle size of the sand filter material composing the filter column was, the larger the reduction of the filter speed was, the larger the particle size of the filter material was, the gentler the flow rate was, the lighter the degree of blockage inside the filter column was, and the more stable the flow field of the water flow in the filter layer was. The larger the particle size of the sand filter layer was, the larger the pore diameter was, the slower the pressure rise was, and the stronger the anti-clogging ability of the filter layer was. This finding can provide a strong reference to optimize the layer thickness and particle size of the sand filter.