ECMWF和华东WARMS模式对山东半岛汛期暴雨的预报能力检验
|
摘要
利用欧洲中期天气预报中心细网格模式(以下简称ECMWF-Thin)产品和模式水平分辨率为9 km的华东区域气象中心中尺度数值预报模式V1.0(以下简称SMS-WARMS)产品,对山东半岛2016—2017年汛期35个暴雨日(26次过程)的暴雨预报能力进行检验。结果表明:1)对于降水强度,ECMWF-Thin预报偏弱导致暴雨和大暴雨漏报率偏高,大暴雨几乎全部漏报,当其预报有50mm以上降水时出现暴雨的概率达90%以上,SMS-WARMS则预报降水量偏强、空报率较高,SMSWARMS降水强度量级预报总体优于ECMWF-Thin,24 h预报能力最佳; 2)对于强降水开始时间的预报,两家模式均表现为偏晚为主,且偏晚3 h以内的概率较大,在参考其预报结论的基础上可适当提前3 h; 3)对于强降水落区,ECMWF-Thin略优于SMS-WARMS,SMS-WARMS对台风暴雨的落区预报较为精准,而其他类型暴雨的落区ECMWF-Thin预报多偏南或偏向西南1°以内,因此预报员需向偏东或东北1°范围内的区域调整; 4)对于强降水范围大小的预报,ECMWF-Thin预报暴雨范围偏小的概率较大,而SMS-WARMS预报范围偏大的概率较大,因此需综合考虑两种数值预报结论进行折中预报。
The verification of the rainstorm forecast by ECMWF thin grid model products(ECMWF-Thin for short) and SMS-WARMS V1. 0 with the resolution of 9 km(SMS-WARMS for short) from East China Regional Meteorological Center for the 35 rainstorm days(26 processes) in the flood season of Shandong Peninsula from 2016 to 2017 is conducted. The results are listed as below. 1) For the precipitation intensity,the weak prediction by ECMWF-Thin leads to high omission rate of rainstorms and torrential rain processes and almost all the torrential rain processes are omitted. When the precipitation is predicted to be more than 50 mm,the probability of rainstorms is over 90%. SMS-WARMS predicts relatively strong precipitation and is of high vacancy rate. SMS-WARMS is generally superior to ECMWF-Thin in predicting the precipitation intensity and possesses the best 24-hour forecast ability. 2) For the beginningtime of heavy precipitation,both models are generally predicting later,and the probability of three hours late is greater. Therefore,the precipitation can be suitably predicted to be three hours in advance referring to the conclusion. 3) For the heavy precipitation areas,ECMWF-Thin is slightly better than SMS-WARMS. The latter model is more accurate in forecasting typhoon rainstorm areas,but the areas of other types of rainstorms by the former model are generally to the south or southwest within 1°. Thus forecasters should adjust eastward or northeastward within 1°. 4) For the heavy precipitation range,it is more probable for ECMWF-Thin to forecast a smaller range and SMS-WARMS a larger range. Therefore,it is necessary to integrate two kinds of numerical predictions.
The verification of the rainstorm forecast by ECMWF thin grid model products(ECMWF-Thin for short) and SMS-WARMS V1. 0 with the resolution of 9 km(SMS-WARMS for short) from East China Regional Meteorological Center for the 35 rainstorm days(26 processes) in the flood season of Shandong Peninsula from 2016 to 2017 is conducted. The results are listed as below. 1) For the precipitation intensity,the weak prediction by ECMWF-Thin leads to high omission rate of rainstorms and torrential rain processes and almost all the torrential rain processes are omitted. When the precipitation is predicted to be more than 50 mm,the probability of rainstorms is over 90%. SMS-WARMS predicts relatively strong precipitation and is of high vacancy rate. SMS-WARMS is generally superior to ECMWF-Thin in predicting the precipitation intensity and possesses the best 24-hour forecast ability. 2) For the beginningtime of heavy precipitation,both models are generally predicting later,and the probability of three hours late is greater. Therefore,the precipitation can be suitably predicted to be three hours in advance referring to the conclusion. 3) For the heavy precipitation areas,ECMWF-Thin is slightly better than SMS-WARMS. The latter model is more accurate in forecasting typhoon rainstorm areas,but the areas of other types of rainstorms by the former model are generally to the south or southwest within 1°. Thus forecasters should adjust eastward or northeastward within 1°. 4) For the heavy precipitation range,it is more probable for ECMWF-Thin to forecast a smaller range and SMS-WARMS a larger range. Therefore,it is necessary to integrate two kinds of numerical predictions.
引文
[1]赵宁坤,万石云,杞明辉.3种数值模式对云南地区雨季降水预报的检验[J].气象与环境学报,2015,31(5):39-44.
[2]杨晓霞,夏凡,杨璐瑛.山东WRF集合概率预报对台风“麦德姆”逐6 h精细化预报检验[J].山东气象,2016,36(3):20-27.
[3]崔粉娥,王勇,李慧君.多家数值产品沿海大暴雨预报性能检验[J].气象科技,2013,41(4):696-702.
[4]肖红茹,王灿伟,周秋雪,等.T639、ECMWF细网格模式对2012年5~8月四川盆地降水预报的天气学检验[J].高原山地气象研究,2013,33(1):80-85.
[5]曲巧娜,盛春岩,车军辉,等.山东省多模式强降水落区预报检验[J].气象科技,2016,44(3):392-399.
[6]孙素琴,郑婧,金米娜,等.基于多模式2015年江西省汛期区域性暴雨的检验[J].气象与环境学报,2017,33(2):1-7.
[7]翟振芳,魏春璇,邓斌,等.安徽省ECMWF数值模式降水预报性能的检验[J].气象与环境学报,2017,33(5):1-9.
[8]程立渤,崔宜少,仇彦辉.T639降水预报产品的应用和检验[J].山东气象,2011,31(1):24-27.
[9]徐姝,魏琳,邓岩.4种数值降水预报产品在海河流域的应用检验[J].海河水利,2017(2):45-51.
[10]张永婧,荣艳敏,尹承美,等.WRF集合预报对济南暴雨过程的检验与分析[J].山东气象,2016,36(2):10-14.
[11]肖明静,盛春岩,石春玲,等.2010年汛期多模式对山东降水预报的检验[J].气象与环境学报,2013,29(2):27-33.
[12]梁寒,陈传雷,聂安祺,等.辽宁地区“8.16”特大暴雨过程数值预报产品检验[J].气象与环境学报,2016,32(6):1-9.
[13]张国华,张江涛,张南,等.基于河北区域天气分型的多模式降水检验评估[J].气象,2011,37(8):968-976.
[14]杨学斌,谌芸,代玉田.山东一次区域性暴雨中尺度特征分析[J].气象科技,2012,40(4):627-634.
[15]周黎明,王庆,龚佃利,等.山东一次暴雨过程的云降水微物理特征分析[J].气象,2015,41(2):192-199.
[2]杨晓霞,夏凡,杨璐瑛.山东WRF集合概率预报对台风“麦德姆”逐6 h精细化预报检验[J].山东气象,2016,36(3):20-27.
[3]崔粉娥,王勇,李慧君.多家数值产品沿海大暴雨预报性能检验[J].气象科技,2013,41(4):696-702.
[4]肖红茹,王灿伟,周秋雪,等.T639、ECMWF细网格模式对2012年5~8月四川盆地降水预报的天气学检验[J].高原山地气象研究,2013,33(1):80-85.
[5]曲巧娜,盛春岩,车军辉,等.山东省多模式强降水落区预报检验[J].气象科技,2016,44(3):392-399.
[6]孙素琴,郑婧,金米娜,等.基于多模式2015年江西省汛期区域性暴雨的检验[J].气象与环境学报,2017,33(2):1-7.
[7]翟振芳,魏春璇,邓斌,等.安徽省ECMWF数值模式降水预报性能的检验[J].气象与环境学报,2017,33(5):1-9.
[8]程立渤,崔宜少,仇彦辉.T639降水预报产品的应用和检验[J].山东气象,2011,31(1):24-27.
[9]徐姝,魏琳,邓岩.4种数值降水预报产品在海河流域的应用检验[J].海河水利,2017(2):45-51.
[10]张永婧,荣艳敏,尹承美,等.WRF集合预报对济南暴雨过程的检验与分析[J].山东气象,2016,36(2):10-14.
[11]肖明静,盛春岩,石春玲,等.2010年汛期多模式对山东降水预报的检验[J].气象与环境学报,2013,29(2):27-33.
[12]梁寒,陈传雷,聂安祺,等.辽宁地区“8.16”特大暴雨过程数值预报产品检验[J].气象与环境学报,2016,32(6):1-9.
[13]张国华,张江涛,张南,等.基于河北区域天气分型的多模式降水检验评估[J].气象,2011,37(8):968-976.
[14]杨学斌,谌芸,代玉田.山东一次区域性暴雨中尺度特征分析[J].气象科技,2012,40(4):627-634.
[15]周黎明,王庆,龚佃利,等.山东一次暴雨过程的云降水微物理特征分析[J].气象,2015,41(2):192-199.