This study elucidated the spatial variability of heterogeneous soil texture in the soil water zone combining the classical statistics, geostatistics and Voxler method for a real-world field in the Yanlou Township of the middle of Lankao County, Kaifeng City, Henan Province. The results showed that the spatial variability of different soil particles in the soil water zone in the study area was significantly different, and the variability intensity was sand>clay>silt, and the spatial variability intensity of soil texture increased with depth of soil water zone increasing. The spatial correlation degree of sand, silt and clay in soil water zone was different, and the spatial correlation strength was clay>sand>silt. In the soil water zone, the optimal distribution model of sand in the shallow layer is Gaussian model, and the optimal distribution model in the middle and deep layer is spherical model. The optimal models of particle spatial distribution in shallow and middle layers are Gaussian models. The optimal distribution model of clay particles is the spherical model in the shallow layer and deep layer, Gaussian model in the middle layer.