Soil moisture is crucial in governing land surface processes that have an important influence on the yield and quality of crops. Therefore, it is essential to establish criteria for appropriate mean soil moisture monitoring strategy in order to obtain a representative mean soil moisture value of a farmer’s field. In this study, the plant available water capacity was introduced as an auxiliary variable and a stratified soil moisture sampling method based on the spatial autocorrelation of auxiliary variables (SSAV) was proposed by integrating classical statistics and geostatistics. The results of the proposed methods were compared with those of the international common simple random sampling (SRS) and stratified random sampling (STRS) methods at the field and regional scale. The results showed that the range of mean relative error and the standard deviation of the soil moisture obtained with the SSAV method were significantly lower than those of the soil moisture obtained with the SRS and STRS methods at both the field and regional scales. The root mean squared error between the observed and estimated soil moisture at the field and regional scales were found to be 0.0104 and 0.0125 cm³/cm³, respectively, with the SSAV method, which are significantly lower than those obtained with the SRS method (0.0124 and 0.0139 cm³/cm³, respectively) and STRS method (0.0116 and 0.0130 cm³/cm³, respectively). The standard deviation of the relative difference, mean absolute bias error, and root-mean-squared difference of the SSAV method, which were used as stability indices of the monitoring points, were all lower than those of the SRS and STRS methods. These results demonstrated that the SSAV could promote the monitoring accuracy and precision, and the soil moisture estimated based on the SSAV could represent the mean soil moisture for several years. The use of the SSAV is recommended as an effective method for the placement of soil moisture sampling points to estimate the mean soil moisture.

土壤水分对于控制对作物产量和质量有重要影响的地表过程至关重要。因此，有必要为适当的平均土壤水分监测策略建立标准，以获得具有代表性的农民田地平均土壤水分值。本研究引入植物有效水容量作为辅助变量，结合经典统计学和地质统计学，提出了一种基于辅助变量空间自相关(SSAV)的分层土壤水分采样方法。将所提出方法的结果与国际通用的简单随机抽样（SRS）和分层随机抽样（STRS）方法在现场和区域尺度上的结果进行了比较。结果表明，在田间和区域尺度上，SSAV方法获得的土壤水分平均相对误差范围和标准偏差均显着低于SRS和STRS方法获得的土壤水分。发现在田间和区域尺度上观测到的和估计的土壤水分之间的均方根误差为 0.0104 和 0.0125 cm³ /cm ³，分别使用SSAV方法，显着低于使用SRS方法（分别为0.0124和0.0139 cm ³ /cm ³）和STRS方法（0.0116和0.0130 cm ³ /cm ³， 分别）。SSAV方法作为监测点稳定性指标的相对差、平均绝对偏差误差和均方根差的标准差均低于SRS和STRS方法。这些结果表明 SSAV 可以提高监测的准确性和精度，并且基于 SSAV 估计的土壤水分可以代表几年的平均土壤水分。建议使用 SSAV 作为放置土壤水分采样点以估算平均土壤水分的有效方法。