ECMWF模式初值场的确定

1、ECMWF模式初值场的确定,确定性数值预报概要（一）,Lets hope that nobody will fall asleep!,AH,3,ECMWF 预报技巧演变,4,5,提高水平与垂直分辨率 减少近似与简化来提高动力方程的精度 采用更准确的物理过程来表达小尺度过程，如辐射、云、降水与热量、水汽与动量的湍流交换 采用更精确的分析与四维资料同化形成模式的初值 增加观测资料,科学家们的努力,引自陈静，业务数值天气预报，2016,6,提纲,资料同化的概念,ECMWF同化系统介绍,我国资料同化现状和展望,7,8,9,主流业务数值预报系统,输入资料,资料同化,数值模式,模式后处理,产品释用,产品发

2、布,3DVar/4DVar/EnKF,Gridded Data,Terrestrial Data,Observation Data,动力框架(全方程组，静力/非静力),物理过程（云物理，边界层，陆面）,(引自孙继松，数值预报产品应用),数值模式,新的初值,Operational configurations of the ECMWF Integrated Forecasting System (IFS),/,11,12,Model Forecast (with errors),Observations (with errors),Computer (wit

3、h a lot of CPUs),People (with a lot of good ideas),Analysis (with - smaller errors),13,Observed temperature (To): 8C Background forecast temperature (Tb): 10 degrees. Analysis (Ta): xC,Observations and model background have errors It is important to specify them accurately,背景场,观测,分析场,背景场和观测的权重与它们的“精

4、确度”（误差方差的倒数）成正比,14,15,16,分析场是通过求解目标函数极小值得到的,：用模式从分析时刻 t0 预报到观测时刻 ti,四维变分（4DVar）,：分析场（新的模式初值）,：背景场（短期预报结果）,：观测（如探空，地面，辐射率，雷达径向风/反射率等）,：背景误差协方差,：观测误差协方差,：观测算子（将模式变量转换为观测变量，“模拟观测”）,17,12-h同化窗,18,Meteosat imagery water vapour channel,Model T,u,v,q,o3,卫星资料的观测算子,Another model versus observation example,CL

5、OUDSAT is a NASA Satellite with a cloud profiling radar on-board,Model,Observations,Cloud Radar Reflectivity,20,雷达径向风的观测算子,把模式空间风矢量 (u,v,w) 场投影到雷达坐标平面， 模拟雷达径向风,：降水粒子的下落末速度,：观测算子（模式空间到观测空间的投影）,(Sun and Crook 1997, 1998),(WRF 3DVAR),The observation operator provides the link between the model variable

6、s and the observations (Lorenc 1986; Pailleux 1990). The observation operator is typically implemented as a sequence of operators transforming the analysis control variable x into the equivalents of each observed quantity y, at observation locations. This sequence of operators can be multi-variate (

7、can depend on many variables) and may include: “Interpolation” from forecast time to observation time (in 4D-Var this is actually running the forecast model over the assimilation window) Horizontal and vertical interpolations Vertical integration If limb-geometry, also horizontal integration If radi

8、ances, radiative transfer computation Any other transformation to go from model space to observation space.,Actual implementation of observation operators used in the ECMWF 4D Variational Data Assimilation System,22,求解目标函数极小值是迭代极小化过程,cost function J,model variable x2,model variable x1,J(xb),Jmini,23

9、,24,25,26,27,28,29,30,：背景场扰动集合预报,用集合预报成员统计背景误差协方差,31,集合预报统计出来的背景误差协方差表征了随时间演变的大气流型,32,EDA（Ensemble of Data Assimilations）,4DVAR,Forecast,X25b(tk),Y+25o,Boundary pert. 25,X25a(tk),X25b(tk+1),25m,34,35,36,欧洲中期天气预报中心（ECMWF）的数值预报产品是我国气象业务部门的重要参考,资料同化技术发展的前瞻性,卫星资料同化-使用更多卫星仪器类型和卫星资料,3DVar (1996),4DVar (19

10、97),EDA (2011),“全天候”微波同化技术,长时间窗-弱约束4DVar,增加使用受云降水和气溶胶影响的红外、微波卫星观测,优化资料挑选、质量控制和观测误差的定义,优化先进垂直探测仪资料的使用,37,38,Conventional observations used by ECMWFs analysis,Note: Data assimilation only use a limited number of the observed variables - especially over land.,39,Satellite data sources used by ECMWFs an

11、alysis,Imagers: SSMI, SSMIS, AMSR-E, TMI,Ozone,GPS radio occultations,Sounders: NOAA AMSU-A/B, HIRS, AIRS, IASI, MHS,Geostationary+MODIS: IR and AMV,Scatterometer ocean low-level winds: ASCAT,40,What types of satellites are used in NWP?,Advantages Disadvantages GEO - Regional coverage No global cove

12、rage by single satellite - Temporal coverage LEO - Global coverage with single satellite,41,Metop,42,Role of observations,RMS error (m),Time (hours),SEVIRI 6.2 m,Every 12 hours we assimilate 20,000,000 observations to correct the models variables. The model has many more variables than we have obser

13、vations.,Observations limit error growth and make forecasting possible.,43,FSO(Forecast Sensitivity to Observations) ECMWF System (June 2011),AMSU-A,GNSS-RO,IASI,AIRS,Remark: GNSS-RO contributes 2-3 % of obs. assimilated,(引自顾建峰，气象业务现代化与信息化),TOVS,ERA-Interim skill shows little change 1980-2000 when t

14、here was little change in the GOS. The ATOVS era shows a gain of around 6-18 hours in predictable skill compared to the TOVS era. ERA-Interim excludes more recent data e.g. IASI and ASCAT so the impact of newer data is hard to judge. This compares to overall ECMWF gain in skill of about 2-3 hours pe

15、r year.,ATOVS,ATOVS + Hyperspectral IR,Full operational system. GOS + forecast changes with time,Re-analysis system. Only GOS changes with time,45,卫星资料同化_GRAPES模式系统中同化的观测资料,GRAPES-GFS中同化的卫星资料量占总资料量的比例：60%左右,引自陈静，业务数值天气预报，2016,46,GRAPES与UKMO业务模式资料使用情况对比,过去5年、特别是近三年，GRAPES-GFS同化的资料量有很大提高 与英国模式资料应用量的差距

16、显著减小,引自陈静，业务数值天气预报，2016,47,气象现代化发展纲要 （2015-2030年）,一、指导思想和发展目标 二、大力提升气象防灾减灾和公共气象服务水平 （一）加强气象防灾减灾能力建设 （二）提升公共气象服务均等化水平 （三）推进气象服务社会化 三、积极应对气候变化 （一）加强气候变化科学研究 （二）提升气候变化适应能力 （三）强化生态文明气象保障 四、加快发展现代气象业务 （一）建立现代气象服务业务 （二）提高气象预报预测水平 （三）强化综合气象观测能力 五、着力推进气象信息化 （一）建立集约化气象信息业务体系 （二）提升气象数据质量与开放共享水平 （三）增强气象政务管理信息化能力 （四）完善气象信息化运行保障 六、加强科技创新和人才发展 （一）强化科技引领和创新驱动 （二）坚持气象人才优先发展 七、强化气象现代化的法治保障 （一）提高气象依法行政能力 （二）强化气象标准化管理 （三）完善与气象现代化相适应的体制机制,(引自顾建峰，气象业务现代化与信息化),48,重点任务与目标,(引自顾建峰，气象业务现代