海量卫星重力梯度观测数据确定地球重力位模型的数值方法

基于空域最小二乘法，对卫星重力梯度数据确定地球重力场中的Cholesky分解法、预条件共轭梯度法以及OpenMP并行算法3种数值方法进行比较与分析。研究表明，在计算机硬件资源有限的情况下，传统的Cholesky分解法已经无法满足求解要求；预条件共轭梯度法的求解效率较之Cholesky分解法有改进，但其以损失小量精度为代价；OpenMP并行算法在不损失求解精度的条件下，可提高求解的效率。

ON NUMERICAL METHODS FOR DETERMINATION OF EARTH GRAVITY FIELD MODEL USING MASS SATELLITE GRAVITY GRADIOMETRY DATA

On the basis of Space Wise Least Square method, three numerical methods including Cholesky decomposition, Pre conditioned conjugate gradient and Open Multi Processing parallel algorithm are applied into the determination of gravity field with satellite gravity gradiometry data. The results show that, Cholesky decomposition method has been unable to meet the requirements of computation efficiency when the computer hardware is limited. Pre conditioned conjugate gradient method can improve the computation efficiency of huge matrix inversion, but it also brings a certain loss of accuracy. The application of Open Multi Processing parallel algorithm could achieve a good compromise between accuracy and computation efficiency.