方差分量估计方法在ERS—2卫星精密定轨中的应用

将方差分量估计（VCE）方法应用于ERS－2卫星的精密定轨,用SLR和PRARE资料计算了1998年前3个月的23个长度为5天的弧段（除了调整轨道的时段外,相邻弧段有两天的重叠）,从观测值残差分析和重叠弧段比较两个方面,考察VCE方法对定轨计算的影响,并给出了各组观测值的平均验后均方差,对观测值残差的分析表明,使用VCE方法明显地改善了观测值的拟合程度,但从阿卑（Abbey)标准对观测值残差的检验结果来看,使用VCE方法不能消除轨道中由力学模型和几何模型误差引入的系统差,重叠弧段比较的结果表明：（1）使用VCE方法缩小了重叠弧段的平均距离差,并改善了一部分权段明显不合理的偏离,使最后得到的轨道具有更均匀的精度,（2）相比较而言,VCE方法使相邻弧段靠拢的趋势在轨道切向体现得较为明显,由各组观测值的平均验后方差可见,说单个标准点观测值而言,部分SLR台站的观测资料在定轨计算中占有比其他观测资料更重的地位,纵观全文,使用VCE后得到的观测值的平衡验后均方差来确定资料的双重将比使用均方差更为合理。

APPLICATION OF VARIANCE COMPONENT ESTIMATION TO PRECISION ORBIT DETERMINATION FOR ERS-2

The variance component estimation (VCE) method was applied to the precision orbit determination (POD) for ERS 2 satellite. Totally, 23 5 day long maneuver free arcs with 2 days overlap (when without orbit maneuver separation between the successive arcs) in the early three months of 1998 were computed using SLR and PRARE data. To inspect the effect of the VCE method on the POD computation, the measurement residuals were analyzed, and the overlapping arcs were compared. At last, the mean posteriori standard deviations of each group of measurement were given. The analyses of residuals show that using the VCE method improved the fitness of measurements. Using Abbey criterion to examine systematic error checked the residuals. The checking results show that the VCE method could not eliminate the systematic errors in the orbit caused by the errors in dynamic and geometric models. The comparison results of overlapping arcs show that (1) using the VCE method reduced the mean deviations of overlapping arcs and the obviously unreasonable deviations in some arcs, thus made the orbit have more uniform precision; (2) relatively,the reduction of the mean deviations is more apparent in the transverse direction. From the mean posteriori standard deviations of each group of measurements, we could see that, for single normal point, the data of several SLR stations are more important than other measurements in POD computation. In general, the yielded mean posteriori standard deviations are more reasonable than the initial standard deviations to weight the observations.