資源描述：
《基于集合kalman濾波數(shù)據(jù)同化的熱帶氣旋路徑集合預(yù)報(bào)研究》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫(kù)。
1、第31卷第3期大氣科學(xué)Vol131No132007年5月ChineseJournalofAtmosphericSciencesMay2007基于集合Kalman濾波數(shù)據(jù)同化的熱帶氣旋路徑集合預(yù)報(bào)研究黃小剛費(fèi)建芳陸漢城解放軍理工大學(xué)氣象學(xué)院軍事氣象系,南京211101摘要構(gòu)建了一個(gè)基于集合Kalman濾波數(shù)據(jù)同化的熱帶氣旋集合預(yù)報(bào)系統(tǒng),通過(guò)積云參數(shù)化方案和邊界層參數(shù)化方案的9個(gè)不同組合,采用MM5模式進(jìn)行了不同時(shí)間的短時(shí)預(yù)報(bào)。對(duì)預(yù)報(bào)結(jié)果使用“鏡像法”得到18個(gè)初始成員,為同化提供初始背景集合。將人造臺(tái)風(fēng)作為觀測(cè)場(chǎng),同化后的結(jié)果作為集合
2、預(yù)報(bào)的初值,通過(guò)不同參數(shù)組合的MM5模式進(jìn)行集合預(yù)報(bào)。對(duì)2003～2004年16個(gè)臺(tái)風(fēng)個(gè)例的分析表明,初始成員產(chǎn)生方法能夠?qū)釒庑囊貓?chǎng)、中心強(qiáng)度和位置進(jìn)行合理擾動(dòng)。同化結(jié)果使臺(tái)風(fēng)強(qiáng)度得到加強(qiáng),結(jié)構(gòu)更接近實(shí)際?；谕募下窂筋A(yù)報(bào)結(jié)果要優(yōu)于未同化的集合預(yù)報(bào)。使用“鏡像法”增加集合成員提高了預(yù)報(bào)準(zhǔn)確度,路徑預(yù)報(bào)誤差在48小時(shí)和72小時(shí)分別低于200km和250km。關(guān)鍵詞熱帶氣旋集合預(yù)報(bào)集合kalman濾波數(shù)據(jù)同化路徑預(yù)報(bào)文章編號(hào)10069895(2007)03046811中圖分類號(hào)P444文獻(xiàn)標(biāo)識(shí)碼ATheEnsembleFo
3、recastingofTropicalCycloneTrackBasedonEnsembleKalmanFilterDataAssimilationHUANGXiao2Gang,FEIJian2Fang,andLUHan2ChengInstituteofMeteorologicalCollege,PLAUniversityofScienceandTechnology,Nanjing211101AbstractThetechniqueofensembleforecastingbasedonEnsembleKalmanFilter(En
4、KF)dataassimilationisap2pliedtotheproblemoftropicalcyclonetrackpredictionusingMM5model.AdoptingtheAnthes2kuo,GrellandBetts2Millercumulusparameterizationschemes,High2resolutionBlackadar,Burk2ThompsonandMRFPBLprocessparameterizationschemestodesign9groupsmodelconfiguratio
5、n,452,602and752minuteforecastsareconductedforeachsituation.Withthe“mirrorimagingmethod”,18differentinitialconditionsareobtained.Takingthe“Rankinevortex”asobservationdataandthe18differentinitialconditionsasthebackgroundensemble,theEnKFdataassim2ilationwithEnSRFarithmeti
6、carethencarriedout.Utilizingthe18dataassimilationresultsastheensembleforecas2tinginitialfields,andwith9differentmodelconfiguration,722hourforecastissimulated.Twoexperimentsaredesigned.Oneisnon2assimilationensembleforecasting,inwhichbogustyphoonisdirect2lyjoinedand6typh
7、ooncasesin2004areselected.TheotheristheensembleforecastingbasedonEnKFdataassimi2lation,inwhich16typhooncasesin2003and2004areselected.Therearethreemethods,fullensembleaverage,clusteraverageandselectaverage,inensembleaverage.Theresultsshowthatbecauseofnoadjointprocessing
8、,EnKFdataassimilationmethodismoreefficientthanthatof4D2VAR,andwiththeassimilation,theintensityoftyphoonbecomesstronge
