Evaluation of atmospheric empirical model based on TIMED/SABER satellite temperature data

A total of 15 years of global temperature data from 2002–2016 as measured by TIMED/SABER were taken as standard to compare with those of USSA-76 and NRLMSISE-00. The accuracy of the two atmospheric models was analyzed by calculating daily mean and monthly mean temperature deviations and standard deviations of temperature deviations at different altitudes. The confidence of the model under different temperature deviations is calculated through setting the deviation between the models and the satellite observation temperature. The results show that the temperature data calculated by the two atmospheric models are in good agreement with the temperature observed by satellite. The daily mean and monthly mean accuracy of the NRLMSISE-00 are higher than those of the USSA-76. In the UMLT region, the two atmospheric models have yet to be improved to characterize the true atmosphere at the mid-high latitudes of upper mesosphere in summer and winter. The overall ability of the NRLMSISE-00 to represent the true atmosphere is superior to the USSA-76. In addition, the confidence decreases with the increase of altitude in the same latitude area and temperature deviation requirements. The confidence decreases with increasing latitude in the same altitude and temperature deviation requirements. By analyzing the differences between the atmospheric models and satellite observations. This paper provides a reference for the application of atmospheric models in engineering and provides the basis for the subsequent correction of atmospheric modes.