Abstract: The assimilation of GPS（Global Positioning System）/LEO（Low Earth Orbit） occultation data may effectively improve the vertical profiles and hence improve the current Numerical Weather Prediction（NWP） model.At present the greatest obstacle for variational assimilation retrieval technique to enter the current numerical weather prediction model is its enormous calculation.Numeric methods is one of the direct reasons for the computing speed.Through theoretical analysis and calculation of observation example，this paper makes analysis and comparison on the three commonly used methods and provides a theoretical basis for the improvements of the technology of variational assimilation.

Key words: GPS occultation, variational assimilation, algorithm efficiency, calculation method

摘要： 掩星数据资料的一维变分同化反演技术可以有效地改进气象参数廓线，但是把该技术用于气象业务流程的最大困难是它的计算量较大，而计算方法是影响计算速度的直接因素之一。通过理论分析和观测实例的计算，对变分同化常用的三种计算方法进行分析和比较，为变分同化技术的改进提供理论依据。

关键词: GPS掩星, 变分同化, 算法效率, 计算方法