Vegetation index has different vegetation detection capability which could result in the different sensitivity on temperature vegetation dryness index. Therefore, two sets of image in April and August of 2011 were chosen to construct the Surface Temperature (T_s)/Vegetation Index (VI) feature space which used Normalized Difference Vegetation Index (NDVI), Ratio Vegetation Index (RVI) and LST in the Delta Oasis of the Weigan and Kuqa Rivers in Xinjiang. In the establishment of the T_s/VI space, linear regression functions were used to approximately fit the lower and upper boundaries of the T_s/VI space so that the dry and wet edges of the space could be determined. Two T_s/VI spaces were established based on the T_s/NDVI and T_s/RVI and the difference between the two spaces were investigated. Temperature Vegetation Dryness Index (TVDI) was calculated from the T_s/NDVI and the T_s/RVI spaces. Make use of TVDI and contemporaneous field measured soil gravimetric water moistures data carrying through regress analysis and establish experienced models of TVDI to assess soil moisture. The main results were as follows:(1) It was significantly associated among VI, T_s and soil moisture. TVDI-NDVI and TVDI-RVI could be used to predict the near surface soil water moisture. (2) TVDI-NDVI had a higher correlation (0.4299) with the soil surface moisture in April (low vegetation coverage month) but TVDI-RVI had a higher correlation (0.5791) with the soil surface moisture in August (high vegetation coverage month). (3) Analyzing and comparing NDVI and RVI, it was showed that NDVI was sensitive in the low vegetation coverage and RVI was sensitive in the high vegetation coverage. The conclusion was the T_s/RVI feature space was more suitable for inversing soil moisture in high vegetation coverage and the T_s/NDVI feature space was more suitable for inversing soil moisture in low vegetation coverage.