Abstract: The arid zone of Inner Mongolia is an important part of the ecological barriers of plateaus in China, whose ecological environment is directly related to ecological security of northern China. To understand the responses of ecosystem evapotranspiration partitioning to drought, a SiB2 model was constructed using eddy covariance data to simulate vegetation transpiration (T) and surface evapotranspiration (ET). The ratio of T/ET was calculated based on eddy-related data from the flux station and the SiB2 model, respectively. Combined with the drought degree assessment index (SPEI), the changes of ET and T/ET ratio under different dry and wet conditions in the study area were analyzed. The results showed that the ET in the study area was significantly affected by drought, and that there was a negative correlation between ET and drought degree. The lag time of drought effect on ET v increased  when drought was severe or extremely severe. The T/ET ratios calculated by SiB2 model from 2003 to 2012 was strongly correlated with drought degree. In the humid year (2003), the T/ET ratio reached the highest value of 0.64, indicating that vegetation transpiration played a dominant role in evapotranspiration. In the relatively dry year (2005), T/ET value was 0.39, indicating that vegetation transpiration decreased under drought stress and that soil evaporation played a dominant role in evapotranspiration. In the case of continuous drought, the fluctuation range of T/ET and ET was weak, indicating that vegetation could self-regulate under drought, which plays a certain role in maintaining ecosystem stability. The results have significance for promoting rational allocation of regional water resources and improving water use efficiency in arid areas.

Key words: evapotranspiration partitioning, eddy covariance technique, SiB2 model, drought.