中文摘要：

利用无线电全息方法, 如正则变换方法或全谱反演方法, 可以有效地解决大气多路径条件下GPS掩星信号的反演问题. 本文采用正则变换方法反演掩星资料, 模拟仿真反演结果显示正则变换方法可以准确地反演包含大气多路径效应的信号. 在模拟信号的相位中加入不同程度的高斯相位噪声后, 正则变换方法的反演结果会受到不同程度的影响. 用正则变换方法对2007年第71天至73天共约4500个COSMIC数据进行处理. 将其折射率反演结果和atmPrf资料 （利用全谱反演方法计算得到） 一起, 与对应的ECMWF 分析场资料进行统计比较, 结果表明： 在5 km以下, 正则变换方法计算的折射率的相对误差的平均值普遍大于atmPrf资料. 其原因可能为： 正则变换方法将信号从LEO轨迹后传播至辅助屏, 造成孔径缩小, 精度下降. 同时也讨论了信号截断对低对流层中反演精度和掩星个数的影响.

英文摘要：

Under the atmospheric multipath conditions, both canonical transform （CT） and full spectrum inversion （FSI） method can solve the problem of calculating bending angle profiles within the multipath area. The atmospheric propagation of GPS signals under atmospheric multipath conditions is simulated using multiple phase screens （MPS） technique. Bending angle profiles computed by CT method are compared with corresponding solutions to Abel integral （taken as the true value）. The results show that CT method is close to the true value in the multipath area. The retrieval accuracy of CT method is degraded to some extent when Gaussian noises are added to the phase of simulated signal. About 4500 COSMIC （Constellation Observing System for Meteorology, Ionosphere and Climate） atmPhs profiles from DOY （day of year） 71 to DOY 73 in 2007 are retrieved by CT method. Statistical comparisons of the retrieved refractivity profiles, together with atmPrf data （retrieved by FSI method）, with those from corresponding ECMWF （European Centre for Medium-Range Weather Forecasts） analysis show that CT method contains greater systematic negative bias than atmPrf data below 5 km. A possible reason is that the signal aperture is decreased for back-propagating the signal from LEO position to the back-propagation plane in CT method. The small aperture means low accuracy in the refractivity. The influence of signal truncation on both retrieval accuracy and occultation number is also discussed.