首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于快速更新同化数值预报的小时降水量时间滞后集合订正技术
引用本文:唐文苑,郑永光.基于快速更新同化数值预报的小时降水量时间滞后集合订正技术[J].气象,2019,45(3):305-317.
作者姓名:唐文苑  郑永光
作者单位:国家气象中心, 北京100081,国家气象中心, 北京100081
基金项目:国家重点研发计划(2017YFC1502003、2018YFC1507504)、中国气象局气象预报业务关键技术发展专项(YBGJXM(2017)02 01)共同资助
摘    要:由中小尺度对流系统造成短时强降水天气的发生发展十分迅速,对其落区和时效的预报预警一直都是预报业务中的难点。本文基于快速更新同化的中尺度数值预报系统GRAPES-RAFS 0. 1°×0. 1°分辨率逐小时降水预报,首先通过时间滞后集合预报方法构建了多个集合成员,使用平均TS评分值计算相应预报成员权重系数建立预报方程,然后采用频率匹配订正法进行降水量级订正,从而得到集合订正的逐小时降水量预报。2017年8-9月的逐日试验和典型个例预报结果评估表明效果良好。(1)GRAPES-RAFS最新时次的预报场并不完全代表最好的预报效果,通过时间滞后集合订正方法自动识别优选预报成员,显著提高了预报能力;(2)GRAPES-RAFS预报存在降水量级偏弱的系统性误差,经过频率匹配方法订正后,在量级预报上更接近实况;(3)时间滞后集合预报对我国中东部(包括黄淮、江淮、江南地区)的小时降水量订正效果最好;(4)进一步使用的频率匹配订正方法显著提升了逐时降水量的预报效果,其在降水频率更高、强度更大的江南南部、华南、西南地区效果更为显著;(5)对于中小尺度的强降水过程,经过上述方法订正后,显著提高了模式对强降水系统位置、形态及降水量级的预报水平。

关 键 词:小时降水预报,快速更新同化预报,时间滞后,集合预报,频率匹配订正
收稿时间:2018/2/4 0:00:00
修稿时间:2018/8/29 0:00:00

Improvement of Hourly Precipitation Forecast Using a Time Lagged Ensemble Based on Rapid Refresh Assimilation and Forecast
TANG Wenyuan and ZHENG Yongguang.Improvement of Hourly Precipitation Forecast Using a Time Lagged Ensemble Based on Rapid Refresh Assimilation and Forecast[J].Meteorological Monthly,2019,45(3):305-317.
Authors:TANG Wenyuan and ZHENG Yongguang
Institution:National Meteorological Centre, Beijing 100081 and National Meteorological Centre, Beijing 100081
Abstract:The formation and development of short term heavy precipitation caused by small and mesoscale convective systems are very rapid, and the prediction and warning of the location and period of precipitation are always difficult in forecasting operation. In recent years, the accuracy and resolution of mesoscale model have been improved, and it plays more and more important role in forecasting and warning of severe convective weather. In this paper, based on hourly precipitation forecast from Rapid Analysis and Forecasting System GRAPES RAFS (0.1°×0.1°) during August-September 2017, ensemble members are formed by the time lagged ensemble forecast method. The average TS score is used to calculate weight coefficients of corresponding ensemble members, and then frequency matching method is adopted to correct precipitation forecast bias. The conclusions drawn from this study are as follows. (1) For GRAPES RAFS, the most accurate precipitation forecast does not always come from the most recent ensemble member. Time lagged ensemble method can significantly improve the prediction ability of the model by automatically identifying the optimal forecast members. (2) The GRAPES RAFS hourly precipitation forecast presents systematic weak biases. After corrected by the frequency matching method, the hourly precipitation forecast gets more close to the actual situation in magnitude. (3) The time lagged ensemble method works better for central and eastern parts of China. (4) The frequency matching method works better for south of the Yangtze River, South China and Southwest China, where precipitation occurs more frequently with greater intensities. (5) The method can significantly improve the prediction capacity of the model for the location, amount and patterns of rainfalls in severe precipitation process of small and medium scales.
Keywords:hourly precipitation forecast  rapid refresh assimilation and forecast  time lagged  ensemble forecast  frequency matching method
本文献已被 CNKI 等数据库收录!
点击此处可从《气象》浏览原始摘要信息
点击此处可从《气象》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号