Atmospheric Density Model Calibration Using 2-dimension Kernel Regression Method

The errors of traditional empirical thermospheric density models often translate into orbit errors of the Low Earth Orbit (LEO) satellites,adversely affect applications such as re-entry operations,manoeuver planning,collision avoidance and precise orbit determination for geodetic missions.By using the data detected by Tiangong-1,the features of NRLMSISE-00 model's density errors is analyzed,and it is found that the errors of model at the similar local time and latitude can be considered approximately identical during the quiet geomagnetic field (Ap ≤ 30).In this paper, 2-dimension kernel regression method is used to estimate model's error based on the sample error data at the similar station.Finally,weighted calibration method is developed for the calibration of model's density according to the duration of time from the sample date to the prediction data,and the RMS level of the error could be reduced from 14.09% to 4.05% through calibration.The results indicate that thermosphere densities can be obtained in high precision with this method.