Abstract:Total mass densities at 400 km altitude derived from the CHAMP(Challenging Minisatellite Payload) satellite are used to characterize the distribution and monthly variance of thermosphere density. Thermospheric density is affected by the solar activity. During the solar minimum, the density is mainly determined by the geomagnetic activity, and the maximum moves from 12 to 18 periodically. The period is about 4 to 5 months, similar with that of the Ap index. Comparison between the CHAMP dataset and the NRLMSISE00 empirical model shows that:the density ratio bias varies with latitude. An underestimation occurs in the south of S, while an overestimation in the north. When the solar activity is quite, the model is greater than the CHAMP measurements by 5% to 10%. If F10.7 is larger than 130, the model is smaller by 10%. Under the quiet and moderate geomagnetic activities conditions, an overestimation of 10% occurs for most of the MLT. The bias also exhibits monthly variation, which is positive in winter and spring and is negative in summer and autumn. The secular statistics provide a valuable reference to improve the empirical model.