Doppler Radar Data Assimilation in KMA's Operational Forecasting
Doppler Radar Data Assimilation in KMA's Operational Forecasting
复制标题
DOI：
10.1175/bams-89-1-39
复制
发表时间：
2008
影响因子：
8
通讯作者：
Q. Xiao;Juanzhen Sun;Wen-Chau Lee;Y. Kuo;D. Barker;E. Lim;D. Won;Mi-Seon Lee;Woo-Jin Lee;Joo-Young Cho;Dong‐Kyou Lee;Hee-Sang Lee
中科院分区：
文献类型：
--
作者：
Q. Xiao;Juanzhen Sun;Wen-Chau Lee;Y. Kuo;D. Barker;E. Lim;D. Won;Mi-Seon Lee;Woo-Jin Lee;Joo-Young Cho;Dong‐Kyou Lee;Hee-Sang Lee
多普勒雷达资料同化简介。同化高分辨率多普勒雷达观测资料一直被认为是提高短期定量降水预报（QPF）的有效途径。自美国天气监视雷达88多普勒（WSR-88 D）网络建立以来，人们对同化多普勒雷达数据的方法进行了广泛的探索。尽管还有很多问题，但过去十年的研究（包括实时实验）已经取得了进展。与90年代初期的研究阶段相比，多普勒雷达资料同化正显示出巨大的潜力。对于多普勒雷达数据同化，4D-Var和集合卡尔曼滤波器近年来一直在积极研究，但仍然计算昂贵，因此不切实际。然而，3D-Var是一种可行的和先进的技术，多普勒雷达资料同化业务应用。在目前的十年中，3D-Var已成为业务数据同化中使用最广泛的技术之一。使用3D-Var是将多普勒雷达数据加入业务预报的一种快速方法。除了计算效率外，3-D Var还提供了一种通过使用观测算子直接同化非模型变量的观测（径向速度和反射率）的方法。径向速度和反射率的直接同化是牛顿松弛轻推和最佳插值方法，其中检索需要产生同化的模型变量的变分方法的优势。在2001- 2003年期间，国家大气研究中心（NCAR）的中尺度和微尺度气象学部（MMM）与韩国气象局（KMA）和首尔国立大学（SNU）合作，利用MM 5中的多普勒雷达数据和天气研究与预报（WRF）的三维变分（3D-Var）数据同化系统。在2004年进行了案例研究和有无雷达数据同化的一个月比较试验之后，该系统于2005年进行了半运行测试，并于2006年投入全面运行。案例研究显示了雷达数据同化的好处，进一步的测试表明，多普勒雷达数据同化在WRF 3D-Var表现强劲，改善了降雨预报。KMA多普勒雷达资料同化的过程是相当复杂的：它包括数据预处理（质量控制，误差统计和格式化），同化与WRF 3D-Var，分析更新循环。径向速度和反射率的直接同化算法是先进的和创新的。在KMA业务预报中，从NCAR的研究模式到业务多普勒雷达资料同化的发展系统的转换
INTRODUCTION TO DOPPLER RADAR DATA ASSIMILATION. Assimilation of highresolution Doppler radar observations has long been recognized as an effi cient way to improve short-range quantitative precipitation forecasting (QPF). Since the Weather Surveillance Radar 88 Doppler (WSR-88D) network in the United States was established, methods to assimilate Doppler radar data have been extensively explored. Although a lot of questions remain, research (including real-time experiments) through the past decade has yielded progress. Doppler radar data assimilation is showing signifi cant promise now compared to its initial research stage in the early 1990s. For Doppler radar data assimilation, 4D-Var and an ensemble Kalman filter have been actively researched in recent years but remain computationally expensive and thus impractical. However, 3D-Var is a feasible and advanced technique for Doppler radar data assimilation in operational applications. During the present decade, 3D-Var has been one of the most widely used techniques for operational data assimilation. Use of 3D-Var is a quick approach to adding Doppler radar data to operational forecasts. In addition to computational efficiency, 3-D Var offers a way to directly assimilate observations (radial velocity and reflectivity) that are not model variables through the use of the observation operators. Direct assimilation of radial velocity and reflectivity is an advantage of the variational method over the Newtonian relaxation nudging and optimal interpolation approaches, in which retrievals are required to produce model variables for assimilation. D uring 2001–03, the Mesoscale and Microscale Meteorology Division (MMM) at the National Center for Atmospheric Research (NCAR) partnered with the Korean Meteorological Administration (KMA) and Seoul National University (SNU) to use Doppler radar data in the MM5 with the Weather Research and Forecasting (WRF) 3-dimensional variational (3D-Var) data-assimilation system. After case studies and one-month comparison experiments with and without radar data assimilation in 2004, the system proceeded to semioperational testing in 2005 and was implemented for full operational production in 2006. The case studies showed benefits of radar data assimilation, and further tests indicated that Doppler radar data assimilation in WRF 3D-Var performed robustly and improved rainfall forecasting. The procedure for KMA Doppler radar data assimilation is rather sophisticated: it includes data preprocessing (quality control, error statistics, and formatting), assimilation with WRF 3D-Var, and analysis update cycling. The algorithms for direct assimilations of radial velocity and reflectivity are advanced and innovative. The transfer of the developed system from research mode at NCAR to operational Doppler Radar Data Assimilation in KMA’s Operational Forecasting