快速、无损监测农田土壤水分含量,是智慧农业的重要研究内容。以新疆南疆阿拉尔国家农业科技园区膜下滴灌棉田为研究对象,运用EM38-MK2大地电导率仪快速、高效的获取了4组不同时期的棉田土壤表观电导率数据,并同步采集表层土壤（0~20 cm）样品,通过构建表观电导率数据与室内测定含水量数据间的反演模型获取了测点的含水量数据,并按照土壤水分干旱分级标准对研究区土壤水分进行划分,综合利用GIS软件和地统计方法对土壤水分的时空变异性进行研究。结果表明：4个时期的土壤水分反演模型决定系数均大于0.80且均方根误差（RMSE）和平均绝对百分误差（MAPE）均较小,表明反演模型精度较高,土壤表观电导率与表层土壤水分相关性较好;不同时期土壤含水量数据表明土壤水分具有很强的时间变异性,变异性由中等变异转变为弱变异再转变为中等变异;受人为灌溉等因素的影响,变异函数模型也存在差异;半方差分析中4个时期的土壤水分块金值与基台值之比均大于75%,表明土壤水分在空间上趋近于弱空间相关;高程反距离权重（IDW）插值图及水分克里格插值图表明微地形是影响土壤水分分布的重要因素。本研究可为干旱区膜下滴灌棉田土壤水分动态监测提供重要的方法支撑,从而更好地指导农业灌溉。

Rapid and non-destructive monitoring of soil moisture content in farmland are essential contents of accurate irrigation. Traditional soil water measurement methods are applicable to small scale and difficult to obtain spatial continuous soil moisture information. Soil moisture is affected by microtopography and soil texture. It has strong spatial variability characteristics. Electromagnetic induction technology has been widely used in soil properties investigation owing to its advantages of rapid, efficient, and non-destructive. In this study, the cotton fields under mulched drip irrigation in Alar National Agricultural Science and Technology Park in southern Xinjiang are considered. EM38-MK2 is used to obtain soil apparent electrical conductivity data of four groups of cotton fields at different stages rapidly and effectively. Surface soil samples (0-20 cm) are collected simultaneously. The observation points’ water content is obtained by constructing an inversion model between the soil apparent electrical conductivity data and the indoor measured water content data. Then, the study area’s soil moisture is divided into the soil moisture and drought classification criteria. Finally, the soil moisture’s spatiotemporal variability is investigated using GIS software and geostatistical methods synthetically. The results show that the determination coefficients of the soil moisture inversion model in four different periods are greater than 0.80, and root mean square error (RMSE) and mean absolute percentage error (MAPE) are small, indicating that the inversion model has high precision and the correlation between soil apparent electrical conductivity and surface soil moisture is good. The data of soil moisture content in different periods show that moisture has strong time variability, which changes from medium variability to weak variability and then to medium variability. Moreover, the variation function model is different owing to the influence of artificial irrigation factors. In the semi-variance analysis, the ratios of Nugget and Sill of soil moisture in four different periods are more than 75%, indicating that soil moisture tends to be a weak spatial correlation. The maps of elevation inverse distance weight (IDW) interpolation and moisture Kriging interpolation show that microtopography is an essential factor affecting soil moisture distribution. This study can provide an important method of support for the dynamic monitoring of soil moisture in cotton fields under mulched drip irrigation in arid areas to better guide agricultural irrigation.