The spatial prediction method of soil particle composition in piedmont alluvial plain with flat terrain was explored.Taking Lingshou，Xingtang and Quyang counties in Hebei province as the study area，combined with environmental variables such as topographic factors，soil types，normalized vegetation index and surface temperature variables，the spatial distribution of surface soil particle composition of 114 sample points in the training set was predicted by three methods:ordinary Kriging method(SLR-OK)，regression Kriging method(SLR-RK)and random forest method(SLR-RF)based on symmetric logarithm ratio (SLR) transformation.The prediction accuracy of the three methods is compared through 50 sample points in the verification set.(1)According to the spatial prediction map，the sand particles show a spatial distribution trend of being low in the northwest and high in the southeast；The spatial distribution trend of silt and clay particles are opposite to that of sand.Compared with SLR-OK，SLR-RK and SLR-RF can better reflect local variation and reduce smoothing effect.(2)For sand and silt，the mean absolute error(MAE)and root mean square error(RMSE)of SLR-RF method for the content prediction of verification set are lower than the other two methods，and from the perspective of determination coefficient，they are higher than that of the other two methods，indicating that the prediction accuracy of SLR-RF is the highest；For clay particles，the MAE and RMSE of the SLR-OK method for the content prediction of the validation set are lower than the other two methods，and the coefficient of determination is the highest，indicating that the SLR-OK method has the highest prediction accuracy.(3)The auxiliary variables of the linear regression prediction model include elevation，soil type and wind effect index；Auxiliary variables for the random forest method model include elevation，soil type，normalized vegetation index，surface temperature，wind effect index，and flow accumulation，and for sand and silt，soil type and elevation are important auxiliary variables，and the normalized vegetation index，surface temperature，wind effect index and flow accumulation were relatively less important.The optimal method for spatial prediction of sand and silt in the study area is SLR-RF.In the piedmont alluvial fan area，the terrain is relatively gentle，and sand and silt are sensitive to topographic variables.The optimal method for spatial prediction of clay is SLR-OK.