观测资料的质量控制是数值预报系统中不可缺少的重要组成部分。质量控制方案的好坏影响到数值预报分析系统中观测资料的使用， 影响数值模式初始分析场的效果， 还直接影响模式预报的结果。本文将变分质量控制方案应用于我国的区域GRAPES三维变分（3DVAR）同化系统中， 改善了该系统分析与预报效果， 并对2009年7月2日00:00（北京时）-3日00:00发生在我国南方的一次暴雨过程进行了数值模拟试验及2009年夏季（6-8月）3个月的连续试验。个例试验和连续试验结果均表明， 应用变分质量控制后的分析结果更加合理。尽管对高度场和风场的分析结果改变不显著， 但湿度场的分析结果得到改善， 水汽输送强度得到调整， 更加合理且接近实况； 对预报模式初值的改善具有积极意义。同时也提高了模式对降水落区、 强度的预报能力。
Quality control of observational data is an important and indispensable componentsin the numerical prediction system. Quality control impactson the observational data using the numerical weather prediction analysis system, impacts the effect of the initial analysis field, also directly affect the model prediction results. In recent yearsthe variationalassimilation has developed,the variational quality control began the widespread use of applied research. The basic idea is based on the quality of the error distribution principles so that the different observations of the variational assimilation in the more reasonable contribution. This article attempts to the variational quality control program applied to Chineseregional GRAPES-3DVAR assimilation system to improve the numerical prediction results. A simulation test is done. It was a rainstorm in southern China from 2 to 3 July. The numerical simulation tests are also done for three months of the summer of 2009. Acase testing and continuous results show that: After quality control of the variational analysis is reasonable. Despite the geopotential and wind field changes are not significant, but the humidity field results improved water vapor transmission intensity is adjusted, close to live a more rational analysis of forecast models to improve the initial value of positive significance. Meanwhilet the pattern of precipitation areaandforecasting intensity are improved.
[1]矫梅燕, 主编. 现代数值预报业务[M].北京: 气象出版社,2010.
[2]陶士伟,仲跻芹,徐枝芳,等.地面自动站资料质量控制方案及应用[J].高原气象, 2009, 28(5): 1202-1209.
[3]Purser R J. A new approach to the optimal assimilation of meteorological data by iterative Bayesian analysis[M]. Amer Meteor Soc, 1984: 102-105.
[4]Lorenc A C. Analysis methods for numerical weather prediction[J]. Quart J Roy Meteor Soc, 1986, 112: 1177-1194.
[5]Dharssi I, Lorence A C, Ingleby N B. Treatment of Gross errors using Maximum Probability Theory[J]. Quart J Roy Meteor Soc, 1992, 118: 1017-1036.
[6]Lorence A C, Hammon P. Objective quality control of observations using Bayesian methods, theory, and a practical implementation[J]. Quart J Roy Meteor Soc, 1998, 114: 515-543.
[7]Andersson E, Jarvinen H. Variational quality control[J]. Quart J Roy Meteor Soc, 1999, 125: 697-722.
[8]Gelman A, Carlin J B, Stern H S, et al. Bayesian data analysis[M]. Statistical Science. London: Chapman and Hall, 1995.
[9]薛纪善, 陈德辉,等著.数值预报系统GRAPES的科学设计与应用[M].北京: 科学出版社, 2008.
[10]马旭林,庄照容, 薛纪善,等. GRAPES非静力数值预报模式的三维变分资料同化系统的发展[J].气象学报, 2009, 67(1): 50-60.
[11]郝民,徐枝芳,陶士伟,等. GRAPES RUC系统模拟研究及应用试验[J].高原气象, 2011, 30(6): 1573-1583.
[12]郝民, 郭英华, 马再忠. 一次降水天气过程的GPS掩星资料在GSI同化系统中的应用[J]. 高原气象, 2010, 29(1): 164-174.
[13]杨毅, 邱崇践, 龚建东,等. 同化多普勒雷达风资料的两种方法比较[J]. 高原气象, 2007, 26(3): 547-554.
[14]邵爱梅,邱晓滨,邱崇践.使用混合样本的集合四维变分同化试验研究[J].高原气象, 2011, 30(3): 583-593.
[15]朱立娟,龚建东,李泽椿.资料同化中雷达VAD风质量控制研究[J].高原气象, 2012, 31(6): 1731-1738.
[16]范水勇, 郭永润, 陈敏, 等.高分辨率WRF三维变分同化在北京地区降水预报中的应用[J].高原气象, 2008, 27(6): 1181-1188.
[17]陈敏,范水勇,仲跻芹,等.全球定位系统的可降水量资料在北京地区快速更新循环系统中的同化[J].气象学报, 2010, 68（4）: 450-463.
