Vegetation index temperature mixed pixels affects the remote sensing monitoring of drought conditions, Water deficit index (WDI) was compared with surface minus air temperature (Ts-Ta) as a water stress indicator which can overcome the difficulty. The experimental field was located in Dalat Banner, Inner Mongolia, and the experimental object was silage summer maize. Normalized difference vegetation index (NDVI) and land surface mixing temperature (Ts) were extracted from UAV-acquired images (multispectral, thermal infrared). Maize physiological parameters and meteorological data were collected to establish WDI model, under three irrigation regimes. WDI model and remote sensing data (NDVI, Ts) were used to generate vegetation indextemperature trapezoidal space, WDI map, and Ts-Ta map. WDI and Ts-Ta were extracted in the sample area. WDI and Ts-Ta were extracted in sample areas. The relationships between WDI or Ts-Ta and soil water content/stomata conductance were analyzed. Results demonstrated that vegetation indextemperature trapezoidal space, WDI and Ts-Ta map were sensitive to shortterm drought response. WDI and Ts-Ta showed similarity, both showed strong correlation with soil water content and stomata conductance (R2=0.4～0.85). At scale of ten days, the correlation between WDI and soil water content/stomata conductance (R2>0.68) was significantly higher than the correlation between Ts-Ta and soil moisture content/stomata conductance (R2<0.6). At the scale of ten days, the effects of different water stress gradients on drought monitoring were analyzed. It was found that under sufficient irrigation, there was no significant correlation between WDI or Ts-Ta and soil water content/stomata conductance (R2<0.12). Under different deficit irrigation, the correlation between WDI and stomata conductance/soil water content was significant (R2=0.7283~0.82), the correlation between Ts-Ta and stomata conductance/soil water content showed large fluctuations (R2=0.3566~0.8074). WDI had greater practicability and stability when monitoring the continuous change of drought compared with the difference in land temperature, at the scale of ten days.