GRAPES全球四维变分同化分析场质量的初步诊断分析  

Quality diagnostic and assessment of GRAPES 4DVAR analysis

作　　者：刘昕 陈静[2,3] 朱跃建[4] 陈法敬 马雅楠 王静[6] LIU Xin;CHEN Jing;ZHU Yuejian;CHEN Fajing;MA Yanan;WANG Jing(Chengdu University of Information Technology,Chengdu 610225,China;Numerical Weather Prediction Center,China Meteorological Administration,Beijing 100081,China;National Meteorological Center,Beijing 100081,China;NCEP/Environmental Modeling Center,Maryland 20740,USA;Chinese Academy of Meteorological Sciences,Beijing 100081,China;Tianjin Meteorological Service,Tianjin 300074,China)

机构地区：[1]成都信息工程大学,成都610225 [2]中国气象局数值预报中心,北京100081 [3]国家气象中心,北京100081 [4]美国国家环境预报中心/环境模拟中心,马里兰20740 [5]中国气象科学研究院,北京100081 [6]天津市气象台,天津300074

出　　处：《气象科学》2021年第6期804-817,共14页Journal of the Meteorological Sciences

基　　金：国家重点研发计划项目(2021YFC3000902)。

摘　　要：为了深入认识全球四维变分同化系统(Global Regional Assimilation and Prediction System Four-Dimensional Variational,GRAPES 4DVAR)分析场质量,利用2019年3月1日至2020年2月29日GRAPES 4DVAR 7层等压面的高度、温度、纬向和经向风等4个要素的分析场及同期ERA5再分析场资料和探空观测资料,基于Pearson相关系数、偏差统计和EOF分解等多种统计分析方法,对GRAPES 4DVAR分析场质量进行了评估。结果表明:(1)GRAPES 4DVAR各要素分析场与ERA5再分析场具有显著的正相关(相关系数均通过α=0.01显著性水平检验),其中高度场和温度场的相关系数大于0.99,风场的相关系数大于0.85,GRAPES 4DVAR分析场和ERA5再分析场与探空观测的均方根偏差值的大小及分布相似,表明GRAPES 4DVAR同化分析场具有较高的质量,且北半球和南半球的分析质量优于热带地区;(2)GRAPES 4DVAR温度和风的分析场与ERA5再分析场在北半球和南半球的偏差存在较明显的季节变化和空间变化,表现为夏半年偏差小于冬半年,中层偏差小于低层和高层,且在对流层低层(高层),南半球(热带地区)的偏差大于北半球和热带地区(南半球);(3)不同要素分析偏差的特征显示,温度分析偏差主要出现在对流层低层,极大值区位于北极和南极地区,而风场分析偏差主要位于对流层中高层,极大值区主要分布于青藏高原、非洲大陆西部及北部、东太平洋及大西洋赤道附近区域。To insight into the analysis quality of GRAPES 4 DVAR, by using Pearson’s correlation coefficient, deviation statistics and EOF(Empirical Orthogonal Function) analysis, the GRAPES 4 DVAR analysis quality for a detailed and thorough assessment was evaluated, based on GRAPES 4 DVAR geopotential height, temperature, zonal wind and meridian wind analysis, ERA5 reanalysis and radiosonde from March 1, 2019 to February 29, 2020. Results show that(1) GRAPES 4 DVAR analysis has a significant positive correlation with ERA5 reanalysis(the correlation coefficients all pass the 0.01 significance level test), among which, the correlation coefficients of geopotential height and temperature analysis are more than 0.99, meanwhile, the root mean square error differences between GRAPES 4 DVAR analysis, ERA5 reanalysis and radiosonde are similar, indicating that GRAPES 4 DVAR analysis is comparable to ERA5 reanalysis with high quality, especially in Northern Hemisphere(NH) and Southern Hemisphere(SH), these analysis quality is better than tropics.(2) Some differences between GRAPES 4 DVAR temperature, wind analysis and ERA5 reanalysis changed in different seasons, in general, the analysis difference of summer months in NH and SH is smaller than in winter months. Moreover, analysis difference in the middle troposphere between GRAPES 4 DVAR analysis and ERA5 reanalysis is smaller than in the low and high troposphere, to be specific, the analysis difference in SH is greater than NH and tropics in the low troposphere, on the other hand, the analysis difference in tropics is greater than NH and SH in the high troposphere.(3) In addition, temperature analysis differences mainly occur in the NH and SH Pole in the low troposphere, additionally, wind analysis differences mainly spread in the Qinghai-Tibet Plateau, west and northern part of Africa, East Pacific and Atlantic East near the equator in the high troposphere.

关 键 词：同化分析 质量评估 GRAPES 四维变分同化 ERA5 

分 类 号：P435[天文地球—大气科学及气象学]