Abstract: The drought is one of the most influential natural disasters on a global scale.The soil desertification and salinization from drought can cause irreversible damage to the ecological environment.On basis of the preprocessing and cloud removal of MODIS data,this paper modifies the TVDI model through terrain correction and constructed an modified Temperature Vegetation Dryness Index （mTVDI） for drought monitoring in arid regions of Xinjiang,China,and compared it with the traditional TVDI and verified it using soil measured data.The results showed as follows: （1） mTVDI_E constructed by EVI（Enhanced Vegetation Index）and corrected LST（Land Surface Temperature）has the highest sensitivity to drought monitoring in arid area, and the correlation with measured soil moisture data is R²:0.74; （2） In 2015,the drought in Xinjiang was centered in the Tarim Basin and the Junggar Basin, and the situation was gradually decreased from the serious drought in the centers to a humid state in the mountainous areas. From the time point of view,the drought in Xinjiang was the most serious in June, July and August; （3） The spatial and temporal distribution characteristics of drought in Xinjiang based on mTVDI_E were compared and analyzed with TRMM precipitation data. The result shows that the spatial and temporal distribution of drought was consistent between the two methods.The correlation between precipitation and mTVDI_E in different time periods was high, and passed the P< 0.01 significance test.Above all,mTVDI_E based on TVDI can effectively carry out drought monitoring in arid regions of Xinjiang with high precision.Therefore,this will provide reference for the quantitative research on drought monitoring at large regional scale in the future.

Key words: arid area, drought monitoring, TVDI, TRMM