Abstract: Soil moisture, as an important component of soil, is a key parameter of climate, agriculture and ecosystems. The rapid, large-scale and real-time monitoring of soil moisture plays an important role in drought forecasting, farmland irrigation and crop estimation. In this study, Landsat 8 data was utilized to improve the accuracy of surface soil moisture retrieval. The feature spaces of surface temperature and vegetation index could be formed to calculate the temperature vegetation drought index (TVDI) and the improved temperature vegetation drought index (ITVDI). The measured soil moisture data obtained from the fields were used to compare the estimated soil moisture calculated from TVDI and ITVDI. The results showed that the most of study area were in drought state during the jointing stage of wheat. The light drought regions of the study area were mainly distributed in the high vegetation-covered regions of western, north and central. The heavy drought regions were mainly located in the cities, bare soils and wheat-planted regions. In order to analyze the relationship between TVDI/ITVDI and soil moisture content, the least squares method was used to produce the scatterplots of TVDI/ITVDI and soil moisture content, respectively. The TVDI and ITVDI had a significant linear relationship with surface soil moisture, respectively, presenting an actual situation of drought in the study area. In addition, the ITVDI was introduced into the fractional vegetation cover to avoid the drought index that was limited by the type of land cover. For ITVDI, the correlation coefficient and the accuracy of estimated soil moisture were slightly higher than that of TVDI. Therefore, the ITVDI could effectively reflect the soil moisture status in the study area and will be more conducive to soil moisture retrieval in the high vegetation-covered areas.