【Objective】Remote sensing has been increasingly used to monitor soil moisture at large scales and the purpose of this paper is to improve its accuracy and reliability for vegetative lands.【Method】The Landsat 8 OLI imagery was used to construct the surface temperature (Ts)-vegetation index (NDVI) feature space, and the TemperatureVegetation Drought Index (TVDI) was calculated based on the wet and dry equation gotten from the Ts- NDVI space. The regressive models were verified against the soil moisture measured concurrently at different depths in a field.【Result】The TVDI obtained retrievably from the remote sensing was significantly correlated with the measured soil moisture with a=0.05. For three soil layers of 0~10 cm, 10~20 cm, 20~30 cm, the TVDI had the highest correlation with the soil moisture in 10~20 cm of soil with r=0.79. The spatiotemporal distribution of the soil moisture retrievably calculated from the remote sensing imagery was consistent with the distribution of crop and the climate change.【Conclusion】It is feasible to monitor the soil moisture using the temperature and vegetation drought index, especially for soil moisture in 10~20 cm of soil.