Crop productivity depends mostly on water management and soil nutrients in the cultivated land. Taking Caoxinzhuang farmland in Yangling as the study area, this study aims to provide a sound basis for the layout of field nutrient monitoring facilities, in order to investigate farmland soil nutrients. Two sampling places were selected, concurrently named as test field 1 (farmland) and test field 2 (farmland), respectively. Specifically, test field 1 was newly reclaimed wasteland, whereas, test field 2 was the cultivated all year round, mainly wheat-corn rotation. The selected field was divided into 12 m×12 m nested 6 m×6 m plots, based on the soil nutrient samples collected in the topsoil (0-20 cm) of different fields during the growth stage of winter wheat. Classical statistical analysis and Geostatistics with Kriging method were employed to explore the characteristics of soil nutrient variability. SPSS22.0 software was used for the descriptive statistical analysis and normal distribution test. According to the Cochran optimal sampling quantity calculation formula, the optimum sampling number of each nutrient index in the field soil was determined.GS+ software (version 9.0, Gamma Design Software, USA) was used to perform a spatial semi-variogram analysis of soil nutrients, and further to adjust different model parameters for model fitting, including determination coefficient R2. Kriging interpolation and cross validation were carried out using the Geostatistics Analysis module in ArcGIS10.5. Sufer software (version 13.0, Golden Software, USA) was used to represent the spatial variation of parameters, including the soil organic matter (SOM), available phosphorus (AP), available potassium (AK), total nitrogen (TN), nitrate nitrogen (NO3--N), and ammonium nitrogen (NH4+-N). The results show that during the heading and ripening stages of winter wheat, the variation coefficient (CV) of total nitrogen (TN) <10% in the surface soil of farmland, indicating a weak variation, while, the CV of soil organic matter (SOM) and available phosphorus (AP) were between 10% and 100%, indicating a moderate variation. There was a strong variation coefficient (CV) >100% in the available potassium (AK) and the ammonium nitrogen (NH4+-N). The nitrate nitrogen (NO3⁻-N) changed from strong variation to moderate variation during the ripening stages of winter wheat. The optimal spherical model can be achieved in the semi-variable function model of soil nutrients. It infers that there were some differences in the spatial correlation of soil nutrients at different stages of crop growth. The nugget coefficient of soil organic matter (SOM) and the total nitrogen (TN) were less than 25% at two growth stages, indicating a strong spatial correlation that mainly affected by structural factors. There was a relatively large variability in the quick-acting nutrients, including the available phosphorus (AP), the available potassium (AK), the nitrate nitrogen (NO3--N), and the ammonium nitrogen (NH4+-N), where the nugget coefficient was between 25% and 75% at the heading stages of winter wheat, indicating the significant role of random factors. At the ripening stages, the nugget coefficient of quick-acting nutrients was less than 25%, indicating the enhanced spatial correlation. When the sampling interval was expanded from 6 m × 6 m to 12m × 12m, the degree of variation remained constant, while the variation coefficient difference of each index fluctuated within the range of 0.04%-59.48%, except for available potassium (398%) in the ripening stage of test field 2. In each index, the difference of nugget coefficient fluctuated within the range of 0.065%-34.177%, while the spatial variation distribution remained basically consistent. The 12 m×12 m grid can be recommended for the topsoil nutrient sampling.