多模式集成分级降水概率及落区预报试验
|
摘要
利用欧洲中心模式(EC-thin),日本模式(JAP),德国模式(GER),T639模式(T639)以及天津WRF模式(TJ-WRF)的格点降水预报资料和历史实况资料,分别基于Rank方法、平均法、相关法设计了三种集成降水概率预报方案,并对2014年4-10月进行集成降水概率预报试验及RPS、BS评分、ROC曲线检验,探讨了集成降水概率预报产品在实际降水预报业务中的应用方案。研究表明:对于各个量级的降水概率预报,Rank方法的预报准确性好于平均法和相关法。基于Rank方法的集成降水概率预报具有可预报性,通过晴雨概率阈值的试验分析,发现当集成降水概率预报达到40%时,降水预报准确率最高,晴雨预报准确率可达到85.1%,通过对各量级概率阈值的试验分析,发现基于集成概率预报阈值得到的分级降水预报产品的ETS评分明显高于单一模式成员;通过个例检验,多模式集成降水概率预报大于40%的区域对降水落区具有较好的指示意义,降水落区预报击中的准确率达到62.3%,典型个例分析显示,对天津小概率极端性的全区性暴雨落区和概率较大的一般性降水落区均有较好的预报效果。
By using observational data and gridded precipitation forecast data that came from weather models of EC-thin, Japan, Germany, T639 and TJ-WRF, the experiment for integrated probability forecast of precipitation(IPFP) and score tests of RPS and BS from April to October in 2014 were conducted, which was based on 3 IPFP schemes designed by methods of Rank, averaging and correlation, respectively. The application scheme of IPFP products used in daily rainfall forecast operation was discussed as well. Results show that with regard to IPFP, the comprehensive prediction accuracy of Rank method is better than averaging and correlation methods. IPFP based on Rank method owns well predictability. Tests of clear-rain probability threshold values reveal that when IPFP reaches 40%, the rainfall forecast accuracy is higher, the clear-rain prediction accuracy can reach 85.1%, By the test of the probability threshold of each precipitation level, the ETS score of the graded precipitation forecast products based on the integrated probability forecast threshold was significantly higher than that of a single model; By means of case-examining, areas which the multi-model IPFP values are greater than 40% have better indicative significance of rainfall falling area, and Accuracy rate for the rainfall area forecasting reaches 62.3%, which had a good indication for the prediction of regional torrential rain and precipitation in general. At the same time, the typical case analysis shows that it has good forecasting effect on the region-wide rainstorm area with small probability extreme and the general rainfall area with large probability in Tianjin.
By using observational data and gridded precipitation forecast data that came from weather models of EC-thin, Japan, Germany, T639 and TJ-WRF, the experiment for integrated probability forecast of precipitation(IPFP) and score tests of RPS and BS from April to October in 2014 were conducted, which was based on 3 IPFP schemes designed by methods of Rank, averaging and correlation, respectively. The application scheme of IPFP products used in daily rainfall forecast operation was discussed as well. Results show that with regard to IPFP, the comprehensive prediction accuracy of Rank method is better than averaging and correlation methods. IPFP based on Rank method owns well predictability. Tests of clear-rain probability threshold values reveal that when IPFP reaches 40%, the rainfall forecast accuracy is higher, the clear-rain prediction accuracy can reach 85.1%, By the test of the probability threshold of each precipitation level, the ETS score of the graded precipitation forecast products based on the integrated probability forecast threshold was significantly higher than that of a single model; By means of case-examining, areas which the multi-model IPFP values are greater than 40% have better indicative significance of rainfall falling area, and Accuracy rate for the rainfall area forecasting reaches 62.3%, which had a good indication for the prediction of regional torrential rain and precipitation in general. At the same time, the typical case analysis shows that it has good forecasting effect on the region-wide rainstorm area with small probability extreme and the general rainfall area with large probability in Tianjin.
引文
[1] 耿焕同,吴正雪,计浩军,等.基于GIS 的上海市嘉定区暴雨积涝灾害风险区划研究[J].灾害学,2015,30(1) :96-101.
[2] 翁莉,马林,徐双凤.城市暴雨灾害风险评估及防御对策研究——以江苏省南京市为例[J].灾害学,2015,30(1):130-134.
[3] 李向红,赵洁妮,伍静,等.桂林“5.9”山体滑坡的暴雨成因分析[J].灾害学,2013,28(3):95-99.
[4] 莫建飞,陆甲,李艳兰,等.基于GIS的广西农业暴雨洪涝灾害风险评估[J].灾害学,2012,27(1):38-43.
[5] 刘莹,赵琳娜,段青云,等.一种由单值预报生成定量降水概率预报的方法及初步应用[J].气象,2013,39(3) :313-323.
[6] Lorenz E N.Deterministicn on periodic flow[J].J Atmos Sci,1963,20(3):130-141.
[7] Lorenz E N.A study of the predictability of a 28-variable atmospheric models[J].Tellus,1965,17(3):321-333.
[8] 邓国,陈怀亮,周玉淑.集合预报技术在暴雨灾害风险分析中的应用[J].自然灾害学报,2006,15(1):116.
[9] 王东海,杜钧,柳崇健.正确认识和对待天气气候预报的不确定性[J].气象,37(4):385-391.
[10] Leslie L M,Helland G L.Predicting regional forecast skill using single and ensemble forecast technique[J].Mon Wea Rev,1991,119(2):425-435.
[11] Hamill T M,Hagedorn R,Wjotaker J S.Probabilistic forecast calibration using ECMWF and GFS ensemble reforecasts.Part 2:Precipitation[J].Mon Wea Rev,2008,136(7):2620-2632.
[12] Hamill T M,Whitaker J S.Probabilistic quantitative precipitation forecasts based on reforecast analogs:Theory and application[J].Mon Wea Rev,2006,134(11):3209-3229.
[13] 吴振玲,潘璇,董昊,等.天津市多模式气温集成预报方法[J].应用气象学报,2014,25(3):293-301.
[14] 金龙,陈宁,林振山.基于人工神经网络的集成预报方法研究比较[J].气象学报,1999,57(2):198-207.
[15] 魏凤英.全国夏季降水区域动态权重集成预报试验[J].应用气象学报,1999,10(4):402-409.
[16] 陈桂英.权重分布法集成预报试验[J].应用气象学报,2000,11(S1):51-57.
[17] Stensrud D J,Bao J W,Warner T T.Using initial condition and model physics perturbations in short-range ensemble simulations of mesoscale convective systems[J].MonWea Rev,2000,128:2077-2107.
[18] DUJ,Mullen S L,Sanders F.Short-range ensemble forecasting(SREF) of quantitative precipitation[J].MonWea Rev,1997,125:2427-2459.
[19] Ebert E E.Ability of a poor man’s ensemble to predict the probability and distribution of precipitation[J].MonWea Rev,2001,129:2461-2480.
[20] Hamill T M,Colucci S J.Verification of Eta-RSM short-range ensemble forecasts[J].MonWea Rev,1997,125:1312-1327.
[21] Hamill T M,Colucci S J.Evaluation of Eta-RSM ensemble probabilistic precipitation forecasts[J].MonWea Rev,1998,126:711-724.
[22] 王晨稀.短期集合降水概率预报试验[J].应用气象学报.2005,16(1):78-88.
[23] 陈超辉,王铁,谭言科,等.多模式短期集合降水概率预报试验[J].南京气象学院学报,2009,32 (2) :206-214.
[2] 翁莉,马林,徐双凤.城市暴雨灾害风险评估及防御对策研究——以江苏省南京市为例[J].灾害学,2015,30(1):130-134.
[3] 李向红,赵洁妮,伍静,等.桂林“5.9”山体滑坡的暴雨成因分析[J].灾害学,2013,28(3):95-99.
[4] 莫建飞,陆甲,李艳兰,等.基于GIS的广西农业暴雨洪涝灾害风险评估[J].灾害学,2012,27(1):38-43.
[5] 刘莹,赵琳娜,段青云,等.一种由单值预报生成定量降水概率预报的方法及初步应用[J].气象,2013,39(3) :313-323.
[6] Lorenz E N.Deterministicn on periodic flow[J].J Atmos Sci,1963,20(3):130-141.
[7] Lorenz E N.A study of the predictability of a 28-variable atmospheric models[J].Tellus,1965,17(3):321-333.
[8] 邓国,陈怀亮,周玉淑.集合预报技术在暴雨灾害风险分析中的应用[J].自然灾害学报,2006,15(1):116.
[9] 王东海,杜钧,柳崇健.正确认识和对待天气气候预报的不确定性[J].气象,37(4):385-391.
[10] Leslie L M,Helland G L.Predicting regional forecast skill using single and ensemble forecast technique[J].Mon Wea Rev,1991,119(2):425-435.
[11] Hamill T M,Hagedorn R,Wjotaker J S.Probabilistic forecast calibration using ECMWF and GFS ensemble reforecasts.Part 2:Precipitation[J].Mon Wea Rev,2008,136(7):2620-2632.
[12] Hamill T M,Whitaker J S.Probabilistic quantitative precipitation forecasts based on reforecast analogs:Theory and application[J].Mon Wea Rev,2006,134(11):3209-3229.
[13] 吴振玲,潘璇,董昊,等.天津市多模式气温集成预报方法[J].应用气象学报,2014,25(3):293-301.
[14] 金龙,陈宁,林振山.基于人工神经网络的集成预报方法研究比较[J].气象学报,1999,57(2):198-207.
[15] 魏凤英.全国夏季降水区域动态权重集成预报试验[J].应用气象学报,1999,10(4):402-409.
[16] 陈桂英.权重分布法集成预报试验[J].应用气象学报,2000,11(S1):51-57.
[17] Stensrud D J,Bao J W,Warner T T.Using initial condition and model physics perturbations in short-range ensemble simulations of mesoscale convective systems[J].MonWea Rev,2000,128:2077-2107.
[18] DUJ,Mullen S L,Sanders F.Short-range ensemble forecasting(SREF) of quantitative precipitation[J].MonWea Rev,1997,125:2427-2459.
[19] Ebert E E.Ability of a poor man’s ensemble to predict the probability and distribution of precipitation[J].MonWea Rev,2001,129:2461-2480.
[20] Hamill T M,Colucci S J.Verification of Eta-RSM short-range ensemble forecasts[J].MonWea Rev,1997,125:1312-1327.
[21] Hamill T M,Colucci S J.Evaluation of Eta-RSM ensemble probabilistic precipitation forecasts[J].MonWea Rev,1998,126:711-724.
[22] 王晨稀.短期集合降水概率预报试验[J].应用气象学报.2005,16(1):78-88.
[23] 陈超辉,王铁,谭言科,等.多模式短期集合降水概率预报试验[J].南京气象学院学报,2009,32 (2) :206-214.