In this study,the 5th generation European Centre Hamburg general circulation model(ECHAM5) and global assimilation and prediction system(GRAPES),independently developed by the China Meteorological Administration,were used to simulate the stratospheric temperature during the period of January 1-6,2010.The results from the two models were then compared and analyzed,referring to the global final analysis(FNL) data.Through a series of comparisons,it was shown that the temperature variations at 50 hPa simulated by ECHAM5 and FNL varied little over time,whereas the simulations performed by GRAPES were significantly elevated in the Southern Hemisphere.The regional warming projected by GRAPES seemed too distinct and uncontrolled to be reasonable,although the warming pattern remained fairly stable with a value between 5 and 15 K relative to the start time.The temperature difference projected by ECHAM5 and FNL was small at the start time on the global scale.Over time,the values become larger in more locations,indicating a significant warming projected by GRAPES.In the Southern Hemisphere,where the warming by GRAPES was dominant,the maximal temperature difference was larger than 24 K at 30-60°S latitude,30-60°E longitude for the final two simulation days,then increased to more than 12 K over the entire Antarctic continent.The temperature initial condition comparison between ECHAM5 and GRAPES shows that they are significantly similar,especially in the Southern Hemisphere,with the difference values being about zero.Moreover,although the ozone profile of ECHAM5 was used in GRAPES,the warming in the Southern Hemisphere simulated by GRAPES still existed.Therefore,the errors produced by GRAPES were not significantly related to either factor.Atmospheric vapor and heating rate may also be factors leading to stratospheric warming,but in this preliminary study these factors could not be investigated,due to a number of technical problems.Further research focusing on the impact of factors such as vapor,heating rate,temperature,etc.will be conducted to further investigate the reasons for the simulated temperature errors produced by GRAPES.