基于GRAPES_3 km模式输出的风雹概率预报技术研究  

Severe Convective Wind and Hail Probabilistic Forecasting Method Based on Outputs of GRAPES_3 km Model

作　　者：曹艳察 郑永光 盛杰 唐文苑 CAO Yancha;ZHENG Yongguang;SHENG Jie;TANG Wenyuan(National Meteorological Centre,Beijing 100081)

机构地区：[1]国家气象中心,北京100081

出　　处：《气象》2021年第9期1047-1061,共15页Meteorological Monthly

基　　金：国家重点研发计划(2018YFC1506104、2018YFC1507504);中国气象局数值预报(GRAPES)发展专项共同资助。

摘　　要：基于我国自主研发的GRAPES_3 km高分辨率模式预报数据,通过计算逐时最大上升螺旋度(updraft helicity,UH)产品,发展了不同预报时效时段内超过指定阈值的UH格点概率预报技术。由于UH能够表征对流风暴的上升运动和中低层旋转强度,因此通过UH格点概率预报产品得到了综合表征对流性大风或冰雹(简称风雹)的概率预报指导产品。对2019年6月14日至7月31日我国华北东北区域和华南区域两个风雹高发区的逐日试验和典型个例预报结果进行了细致评估,结果表明该产品具有良好的预报能力。该产品对华北东北区域和华南区域的风雹TS评分均明显高于主观预报,特别是对于华南区域可预报性较低的弱天气尺度强迫过程,能够显著降低漏报率从而明显提高预报准确率。预报产品还能够很好地指示对流风暴的形态分布和移动传播特征,概率落区与实况风雹的落区位置具有较好的匹配效果。对选取不同的UH阈值和空间高斯平滑系数的概率产品对比检验结果表明,基于较低UH阈值计算的概率预报产品由于漏报率更低,其TS评分要高于较高阈值的预报结果,高斯平滑系数取值20 km得到的概率产品在各等级概率评分中预报效果总体表现最佳。Based on the GRAPES_3 km model developed independently by China and calculating the hourly maximum updraft helicity(UH),we have developed the probabilistic forecast technology which exceeds a certain threshold of UH in different forecast time periods.Since the value of UH can represent the ascending motion and rotation intensity of convective storm,the UH probabilistic forecast product can be used as the probabilistic forecast guidance which characterizes severe convective wind and hail.We verified the probabilistic forecast products of daily tests and typical cases in North China,Northeast China and South China from 14 June to 31 July 2019,the results show that it has a good forecast performance.Compared to the subjective forecast,the threat score(TS)of the objective product for severe convective wind and hail is greatly increased in North China,Northeast China and South China,respectively.Especially for the weak large-scale forcing process in South China,the product can significantly reduce the missing alarm rate(MAR)and significantly increase the TS.In addition,the product can predict the shape distribution and moving propagation characteristics of convective storm,and the region of probabilistic forecast is similar to the observation of severe convective wind and hail.We verified the probabilistic forecast products with different UH thresholds and spatial Gaussian smoothing parameters,the results show that the TS of the products calculated by lower UH threshold is higher than that of higher threshold due to lower MAR.The products with Gaussian smoothing parameter of 20 km have the best forecast performance.

关 键 词：GRAPES_3 km模式 上升螺旋度 概率预报 预报评估 

分 类 号：P456[天文地球—大气科学及气象学]