APPLICATION OF ROBUST ESTIMATION TO CHAMP GRAVITY FIELD RECOVERY
Xu Tianhe 1, 2) ; and He Kaifei 3)
1）Xi’an Research Institute of Surveying and Mapping, Xi’an 7100542）School of Geodesy and Geomatics, Wuhan University, Wuhan 4300793）Chang’an University , Xi’an 710054
Abstract A numerical simulation was made to examine the influence of outliers on CHAMP gravity field recovery. It shows that the model accuracy from least square ( LS) estimation is wholly decreased even if with only one large outlier occurs, and there is a large bias in geoid heights in the large area around the location where the outlier exists. In order to overcome this problem, robust estimation is intruduced into the CHAMP gravity field recovery and its influence function of errors is also given.For one month CHAMP rapid science orbit(RSO) and accelerometer data from GFZ with randomly added to simulated outliers from 5 to 100
m 2s -2 two 60×60 degree gravity field models were solved by LS and robust estimation individually and compared with other models such as EIGENCG03C,EIGEN2. The results show that the robust solution can efficiently control the influence of the outliers. ROCHAMP model is obviously more accurate than LSCHAMP one, and more accurate than EGM96, EIGEN1S, and EIGEN2 at the same degree and order.
