Abstract: Based on the Global Land Data Assimilation System as the data source, the temperature vegetation drought index was improved. On this basis, the drought characteristics were analyzed. The four assimilation data including evapotranspiration, groundwater runoff, precipitation and soil water storage were selected as water cycle elements, and the spatio-temporal evolution of water cycle elements was analyzed using wavelet transform. The results showed that the correlation between the improved temperature vegetation drought index（TVDI） and soil moisture data had been improved. When the soil depth was between 0 cm and 10 cm, the correlation between TVDI and soil moisture was the highest except for individual climatic regions. The correlation between TVDI and soil moisture in the warm season was better than that in the cold season except for the Northwest Desert region and the Inner Mongolia grassland areas. The annual precipitation in China was concentrated between 600 mm and 700 mm, with evapotranspiration below 500 mm. The annual groundwater runoff in China ranged from 100 mm to 250 mm, and the annual soil water storage capacity ranged from 460 mm to 540 mm. The annual evapotranspiration and annual precipitation in the Northwest Desert region,Tibetan Plateau and Northeast China showed a significant increase trend, and the annual soil water storage and annual groundwater runoff in the Northwest Desert region and Northeast China showed a significant downward trend.

Key words: drought characteristics, water cycle elements, spatio-temporal evolution, remote sensing, assimilating data, temperature vegetation drought index