Abstract: Drought is a common natural disaster, which seriously impacts on China's agricultural production. As a complex natural phenomena commonly worldwide occurrence with wide spread and long duration, it is one of the most serious natural disasters for agricultural production and human life, also a direct result in huge economic, social and environmental losses, causing famine, epidemics and large population movements. In recent years in China there has been more large-scale, prolonged drought on the region, which has a serious impact on social-economy and agricultural production. Therefore, it is necessary to build a real-time, dynamic drought monitoring method. The traditional method was using drought monitoring ground point data through statistical analysis of drought monitoring, such method could neither get timely information on drought quickly nor accurately estimate the lag defects. But the space monitoring method, with satellite remote sensing technology which constantly evolved and matured, can be used as a long-term dynamic monitoring with a wide range of fast, low cost. So remote sensing technology and its application became the primary means of regional-scale drought monitoring. Combined with the Moderate Resolution Imaging Spectrometer data MOD11A2 and MOD13A2, the paper uses digital elevation model(DEM)for correcting Ts, extracting normalized difference vegetation index(NDVI)and surface temperature(Ts)to construct Ts-NDVI feature space, and adopts the calculated temperature vegetation dryness index(TVDI)as indicators of soil moisture monitoring based on the feature space to have inversion calculation of every 16 d soil moisture of Xinjiang on May, June, July, 2013, which reflects the surface layer soil moisture perfectely. With analysis of spatial and temporal characteristics of soil moisture in Xinjiang, the paper found that soil moisture in northern part of Xinjiang is higher than that of southern, western higher than eastern, and soil moisture decreases gradually from northwest to southeast, followed by performance as humid> normal> light dry> in dry> severe drought> extremely dry; From May to July soil moisture is increasing, which is very consistent with the distribution of precipitation in Xinjiang, monitoring results credible, able to provide strong information support for drought and drought decision-making departments. As the satellite sensor FOV affects the information receiving, thus affecting the image quality to a certain extent, a reflection of the vegetation will produce some error message, this paper does not consider the impact on the NDVI and Ts perspective, a certain error will exist, on the other hand, if the NDVI is small or large, the linear relationship between NDVI and vegetation coverage is not good, it will reduce the accuracy of NDVI inversion method, we can consider using other vegetation index instead of NDVI, or integrated multi-vegetation index inversion NDVI, such as in different seasons with different vegetation index,that the monitoring study for further improvement needs to be further studied on this issue.

Key words: normalized difference vegetation Index(NDVI), land surface temperature(LST), Temperature-vegetation dryness index(TVDI), Remote sensing monitoring