中文摘要：

介绍CHAMP星载加速仪数据的处理方法,通过对实测数据的分析证实仪器的z轴存在故障．为了研究太阳活动峰年大气模型的精度，处理了2001至2002年的加速仪数据，利用切向非引力加速度反算大气密度．然后从统计的角度分析DTM94和MSIS90大气模型的误差，得到一些定量的结论：太阳活动峰年DTM94的误差为30％-35％，MSIS90的误差为25％-30％，MSIS90比DTM94误差小，模型间最大差异约10％．

英文摘要：

Aerodynamic drag continues to be the largest uncertainty in determining orbits of satellites operating in earth＇s upper atmosphere below about 600 km. Critical precision orbit determination and tracking operations include collision avoidance warnings for the International Space Station, satellite lifetime estimates, laser communication and reentry prediction. Errors in neutral density are the major source of drag errors. Because empirical models are used to estimate satellite drag. Model accuracies have shown little improvement in the past 40 years. Density model errors on the order of 15%--20% have been recognized as a standard deviation for all empirical models developed since the mid 1960s. These large density standard deviations correspond to maximum density errors of approximately 40%--60% as observed in satellite drag data. This paper introduced how to conduct CHAMP/STAR accelerometer data. Then,the fault about instrument＇s x-axis was confirmed by analyzing its observation data. In order to study the accuracy of atmospheric density models during solar maximum year, total atmospheric densities were derived from accelerometer data by using along-track observation from 2001 to 2002. At last, the article evaluated the accuracy of models DTM94 and MSIS90 by statistical methods. The results indicated： during solar maximum year, DTM94＇s relative error is about 30%--35%, and MSIS90＇s relative error is about 25%--30%. MSIS90＇s error is less than DTM94＇s. The most difference between two models is about 10%.