段旭, 王曼, 段玮, 徐开. 孟加拉湾风暴过程TRMM日降水数据变分订正及精度检验[J]. 云南大学学报(自然科学版), 2014, 36(1): 51-58. doi: 10.7540/j.ynu.20130140
引用本文: 段旭, 王曼, 段玮, 徐开. 孟加拉湾风暴过程TRMM日降水数据变分订正及精度检验[J]. 云南大学学报(自然科学版), 2014, 36(1): 51-58. doi: 10.7540/j.ynu.20130140
DUAN Xu, WANG Man, DUAN Wei, XU Kai. On TRMM daily precipitation data variational correction experiment and the accuracy verification for the bay of Bengal storm process[J]. Journal of Yunnan University: Natural Sciences Edition, 2014, 36(1): 51-58. DOI: 10.7540/j.ynu.20130140
Citation: DUAN Xu, WANG Man, DUAN Wei, XU Kai. On TRMM daily precipitation data variational correction experiment and the accuracy verification for the bay of Bengal storm process[J]. Journal of Yunnan University: Natural Sciences Edition, 2014, 36(1): 51-58. DOI: 10.7540/j.ynu.20130140

孟加拉湾风暴过程TRMM日降水数据变分订正及精度检验

On TRMM daily precipitation data variational correction experiment and the accuracy verification for the bay of Bengal storm process

  • 摘要: 利用逐日TRMM(3B42)和地面降水观测资料,对1998—2010年25个孟加拉湾风暴降水过程共184个降水日进行变分订正试验及精度检验分析,结果表明:① 通过变分订正方法迭代建立的TRMM广义的误差场具有一致收敛性,收敛时无论逼近值大小如何,当取精度≤0.001 mm时,迭代次数均不超过800次;② 订正后的TRMM场能较好地反映了地面观测网天气尺度信息,对降水量的离散特征有较好的描述,在无观测的区域,TRMM信息得到了充分体现;③ TRMM值和TRMM订正值经过与未参与订正的100站观测值的平均均方根误差比较可知,所有订正后的样本的误差均小于订正前的误差,184个样本的平均误差由 3.73 mm降低到了1.51 mm,订正效果对减小TRMM场的误差是显著的.

     

    Abstract: This is an experimental research on the TRMM(3B42) data variational correction for a total of 184 days of precipitation in the 25 Bengal storm rainfall process during 1998—2010.The results show thatfirstly,the generalized error field of TRMM data which were based on variational correction method is uniform convergence.The number of iterations will be no more than 800 times when precision is equal or less than 0.001 mm,regardless of the approximate value being large or small.Secondly,the corrected TRMM field which embodied the synoptic scale information and was obtained by the ground observation network well described the discrete features of precipitation,especially for the regions lacking precipitation observation.In this case,TRMM data is of great help.Thirdly,considering the average RMSE of the original TRMM data and the corrected TRMM data against one hundred observations which are not involved in variation correction ,the errors of corrected sample are less than those in the un-corrected,and the average error of 184 samples has been reduced from 3.73 mm to 1.51 mm.Evidently,variational correction helps to reduce the errors of TRMM field.

     

/

返回文章
返回