科研之友
微信
新浪微博
Facebook
分享链接
摘要
Computation of a new Earth's gravity field model is difficult or even impossible if without a priori information about the gravity field. It is important to understand how incorporation of this information at different stages of gravity field modeling could influence the final results. The focus of the paper is on a technique based on satellite accelerations; a version of this technique was used earlier to derive the gravity field model DEOS_CHAMP-01C-70 from a 1-year set of CHAMP data. The technique makes use of both the original kinematic satellite orbit to derive the observed accelerations, and a smoothed orbit to compute the reference accelerations and to detect outliers. Now, using the same data set, we have analyzed how sensitive the obtained results are on the utilization of the a priori knowledge of the Earth's gravity field at different data processing steps. We have found that the final model is not very sensitive on how the smoothed satellite orbit is obtained, though it is advisable that this orbit accounts for a realistic gravity field model. On the contrary, it is crucial that the orbit used to derive the observed satellite accelerations should not be smoothed. Furthermore, it is important to built an accurate stochastic model of data noise. A reasonable way to do so is to analyze the posterior residuals, which are obtained as the difference between the observations and simulations based on a certain gravity field model. The quality of this model is not crucial; it can be obtained from the same data at the stage of trial data processing, i.e. without an accurate stochastic model (at least, if the number of data substantially exceeds the number of unknown parameters). Finally, the quality of gravity field model significantly depends on the reference gravity field. In particular, the EGM96 model noticeably improves the results of CHAMP data processing both at low and at high spherical harmonic degrees.
- 出版日期2007-3
