An improved GRACE monthly gravity field solution by modeling the non-conservative acceleration and attitude observation errors
An improved GRACE monthly gravity field solution by modeling the non-conservative acceleration and attitude observation errors
复制标题
通过对非保守加速度和姿态观测误差建模改进的 GRACE 月重力场解
DOI：
10.1007/s00190-016-0889-6
复制
发表时间：
2016
影响因子：
4.4
通讯作者：
Hsu Houze
中科院分区：
文献类型：
--
作者：
Chen Qiujie;Shen Yunzhong;Chen Wu;Zhang Xingfu;Hsu Houze
本研究的主要贡献是在GRACE重力场解的基础上，考虑了非保守加速度和姿态观测的误差，改进了GRACE重力场解。与以往的研究不同，姿态和非保守的加速度数据，重力场参数，以及加速度计的偏差估计的加权最小二乘平差。然后利用GRACE双星资料计算了2003年1月至2012年12月的60阶月重力场模型时间序列。导出的GRACE解决方案（称为同济GRACE 02）进行了比较，在大地水准面的程度方差和时间质量变化与其他GRACE解决方案，即CSR RL 05，GFZ RL 05 a，和JPL RL 05。结果表明：（1）同济GRACE 02的质量信号与CSR RL 05、GFZ RL 05 a和JPL RL 05的质量信号基本一致;（2）与CSR RL 05相比，Tongji-GRACE 02在海洋上仅使用300 km高斯平滑时，噪声降低了约21%，在沙漠上降低了60%以上（Australia，Kalahari，Karakum和Thar），而不使用高斯平滑和去相关滤波;以及（3）对于所有示例，无论是否应用平滑和滤波，噪声降低都比信号降低更显著。与GLDAS资料的对比表明，Tongji-GRACE 02的信号与CSR RL 05、GFZ RL 05 a和JPL RL 05的信号较为接近，而GLDAS的结果与Tongji-GRACE 02的结果吻合最好。
The main contribution of this study is to improve the GRACE gravity field solution by taking errors of non-conservative acceleration and attitude observations into account. Unlike previous studies, the errors of the attitude and non-conservative acceleration data, and gravity field parameters, as well as accelerometer biases are estimated by means of weighted least squares adjustment. Then we compute a new time series of monthly gravity field models complete to degree and order 60 covering the period Jan. 2003 to Dec. 2012 from the twin GRACE satellites’ data. The derived GRACE solution (called Tongji-GRACE02) is compared in terms of geoid degree variances and temporal mass changes with the other GRACE solutions, namely CSR RL05, GFZ RL05a, and JPL RL05. The results show that (1) the global mass signals of Tongji-GRACE02 are generally consistent with those of CSR RL05, GFZ RL05a, and JPL RL05; (2) compared to CSR RL05, the noise of Tongji-GRACE02 is reduced by about 21 % over ocean when only using 300 km Gaussian smoothing, and 60 % or more over deserts (Australia, Kalahari, Karakum and Thar) without using Gaussian smoothing and decorrelation filtering; and (3) for all examples, the noise reductions are more significant than signal reductions, no matter whether smoothing and filtering are applied or not. The comparison with GLDAS data supports that the signals of Tongji-GRACE02 over St. Lawrence River basin are close to those from CSR RL05, GFZ RL05a and JPL RL05, while the GLDAS result shows the best agreement with the Tongji-GRACE02 result.