Abstract: MODIS（moderate-resolution imaging spectroradiometer）data are widely used in large-scale，long-time，and dynamic drought monitoring for their high time resolution，high spectral resolution，moderate spatial resolution and other characteristics. In this paper，MODIS MOD11C3（monthly land surface temperature products）and MODIS MOD13C2（monthly vegetation index products）were used to study drought conditions of September and October in 2010 in Xilingol League，Inner Mongolia，China. Based on NDVI（normalized difference vegetation index），EVI（enhanced vegetation index）and LST（land surface temperature）indexes，the NDVI-LST and EVI-LST feature spaces were established，and from which the equations were obtained of dry and wet edges，then TVDI（temperature vegetation dryness index）was calculated. According to the criterion of drought levels of the arid and semi-arid regions，the spatial distribution maps of drought degree evaluation were obtained，then the map of drought evaluation comparison between September and October was produced，simultaneously，combining local meteorological information to take qualitative analysis of the drought monitoring result，and using surface soil moisture data with field survey method to take qualitative validation of TVDI. The results show as follows：（1）Even though the most study got TDVI from NDVI－LST feature space，TVDI calculated from EVI-LST feature space could also be suitable for drought monitoring. Under the condition of low vegetation cover，NDVI－LST feature space has better effects than that of EVI－LST，because TVDI obtained from NDVI－LST feature space has better correlation with surface soil moisture than that from EVI－LST feature space；（2）NDVI－LST and EVI－LST feature spaces established in this paper accorded with triangular relationship；（3）Previous studies have shown that [TVDI] can be used in one-day drought monitoring，but this study shows that TVDI is also fit for long-time drought monitoring．From the drought comparison between September and October，the conclusion can also be drawn that TVDI can be applied for dynamic drought monitoring；（4）TVDI has significant negative correlation with surface soil moisture and the correlation coefficient got through 0.05 level test，which means the lower the surface soil moisture，the higher the TVDI. The drought monitoring results reflected from spatial distribution maps of drought situation grades basically coincided with the actual situation in the study area．TVDI can be greatly used in drought early warning and monitoring of grassland area，this method can be used in long-term and large-scale drought monitor and forecast in real-time．

Key words: drought monitoring, MODIS, temperature vegetation drought index