中文摘要：

高精度星间基线确定是分布式InSAR干涉测量中的关键技术。针对卫星在轨运行期间的质心与地面标校结果不一致问题，仿真分析了星体系x、y、z不同方向1 cm质心误差对InSAR星间基线确定的影响。结果表明：x、y方向的质心误差对基线解算影响很小；z方向的质心误差对基线解算影响显著，使得InSAR基线产品中含有明显的系统误差。提出了两种消除卫星质心误差影响的方法：一是在星载GNSS定轨过程中增加轨道径向的经验加速度予以补偿；二是增加星体系z方向的质心偏差估计。实验结果表明，经两种方法处理后得到的InSAR基线3维误差由8.8 mm分别降至0.13 mm和0.09 mm，98%以上的卫星质心误差影响被消除。

英文摘要：

High-precision inter-satellite baseline determination is essential for the distributed InSAR system. The reduced dynamic orbit de- termination method, which employs onboard GNSS measurements and the dynamical constraints, is most extensively used to generate a baseline solution between two formation-fly satellites in low-Earth orbits. However, the reference point of the dynamical model is the center- of-mass （CoM） of the satellite rather than the phase center of the SAR antenna. Owing to the effect of structural deformation and the inevit- able formation-keeping maneuvers, the CoM of a satellite in orbit is usually different from the point calibrated on ground. Thus, the effect of CoM errors must be carefully considered in GNSS-based precision baseline determination; otherwise, CoM errors are likely to induce sys- tematic errors in the spatial baseline solution of the InSAR formation. Simulations are carried out in this study to analyze and mitigate the effect of CoM errors on the InSAR baseline determination. First, a constant CoM error of 1 cm is added to the x-, y-, and z-directions of the satellite-fixed frame. Simulation results show that the effect of such 1 cm error in the x- and y-directions are extremely tiny and can be safely neglected. By contrast, the effect of CoM error in the z-direction is significant, and the results in a systematic variation in the baseline product. In view of this, we further propose two independ- ent methods in GNSS-based InSAR baseline determination to mitigate the z-component CoM errors by adding the following： （1） constant empirical acceleration in the radial direction; （2） an offset parameter in the z-direction of the satellite-fixed frame. The first method is based on dynamical modeling and does not need any extra correction for CoM variations. In addition, the first method is suitable for the mitigation of the CoM errors caused by formation-keeping maneuvers. The second method is based on geometric modeling, and an online calibration for the CoM errors is