In view of the sparse observation stations and low quality of the model analysis in the Tibet Plateau,the Ensemble-3DVAR hybrid data assimilation system(En-3DVAR) is used and both a single pressure observation data assimilation experiment and an assimilation analysis and prediction experiment based on real observation data are conducted.The experiments are to analyze and evaluate the horizontal and vertical structure and rationality of En-3DVAR hybrid assimilation analysis increment over the Tibetan Plateau and to compare them with the data in plain area in Eastern china in order to improve the quality of the data assimilation analysis field in the Tibet Plateau and to improve the model by using the En-3DVAR hybrid assimilation method.The single pressure observation data assimilation experiment shows that the background error covariance of En-3DVAR has the characteristic of flow-dependence.The real data En-3DVAR hybrid assimilation experiment shows that the analysis increment based on ensemble forecast has the characteristic of flow-dependence,especially in the Tibet Plateau and near trough and ridge systems.The absolute error of the analysis field in the Tibet Plateau is generally lower than that of 3DVAR system and the mean absolute error of the analysis field in the Tibet Plateau is lower than that in the plain area on the middle and high levels.The result shows that the improvement effect in the Tibet Plateau is better than that in plain area.What is worth our attention is that in the Tibet Plateau the analysis increment of the En-3DVAR hybrid assimilation is generally larger than that of 3DVAR.Especially on the surface layer,the u-wind component analysis increment is obviously much larger than 3DVAR,which may be related to the complex terrain of the Tibet Plateau.