简介:Inthispaper,akindofexplicitdifferenceschemetosolvenonlinearevolutionequations,perfectlykeepingthesquareconservationbyadjustingthetimestepinterval,isconstructed,fromthecomprehensivemaintenanceofthead-vantagesoftheimplicitcompletesquareconservativeschemeandtheexplicitinstantaneoussquareconservativescheme.Thenewschemesarebasedonthethoughtofaddingasmalldissipation,butitisdifferentfromthesmalldissipationmethod.Thedissipativetermusedinthenewschemesisnotasimpleartificialdissipativeterm,butaso-called(time)harmoniousdissipativetermthatcancompensateforthetruncationerrorsfromthedissociationofthetimedifferentialterm.Therefore,thenewschemesmayhaveahightimeprecisionandmayacquireasatisfactoryeffectinnumericaltests.
简介:
简介:Inthispaper,thenumericalexperimentsontheissueofspin-uptimeforseasonal-scaleregionalclimatemodelingwereconductedwiththenewlyRegionalClimateModel(RegCM3),inthecaseoftheabnormalclimateeventduringthesummerof1998inChina.Totesttheeffectofspin-uptimeontheregionalclimatesimulationresultsforsuchabnormalclimateevent,atotalof11experimentswereperformedwithdifferentspin-uptimefrom10daysto6months,respectively.Thesimulationresultsshowthat,forthemeteorologicalvariablesintheatmosphere,themodelwouldberunningin'climatemode'after4-8-dayspin-uptime,then,itisindependentofthespin-uptimebasically,andthesimulationerrorsaremainlycausedbythemodel'sfailureindescribingtheatmosphericprocessesoverthemodeldomain.Thisverifiesagainthattheregionalclimatemodelingisindeedalateralboundaryconditionproblemasdemonstratedbyearlierresearchwork.Thesimulatedmeanprecipitationrateovereachsubregionisnotsensitivetothespin-uptime,buttheprecipitationscenarioissomewhatdifferentfortheexperimentwithdifferentspin-uptime,whichshowsthatthereexiststheuncertaintyinthesimulationtoprecipitationscenario,andsuchauncertaintyexhibitsmoreovertheareaswhereheavyrainfallhappened.Generally,formonthly-scaleprecipitationsimulation,aspin-uptimeof1monthisenough,whereasaspin-uptimeof2monthsisbetterforseasonal-scaleone.Furthermore,therelationshipbetweentheprecipitationsimulationerrorandtheadvancement/withdrawalofEastAsiansummermonsoonwasanalyzed.Itisfoundthatthevariabilityofcorrelationcoefficientforprecipitationismoresignificantovertheareaswherethesummermonsoonispredominant.Therefore,themodel'scapabilityinreproducingprecipitationfeaturesisrelatedtotheheavyrainfallprocessesassociatedwiththeadvancement/withdrawalofEastAsiansummermonsoon,whichsuggeststhatitisnecessarytodevelopamorereliableparameterizationsch
简介:Inaccordancewithanewcompensationprincipleofdiscretecomputations,thetraditionalmeteo-rologicalglobal(pseudo-)spectralschemesofbarotropicprimitiveequation(s)aretransformedintoperfectenergyconservativefidelityschemes,thusresolvingtheproblemsofbothnonlinearcomputa-tionalinstabilityandincompleteenergyconservation,andraisingthecomputationalefficiencyofthetraditionalschemes.Asthenumericaltestsofthenewschemesdemonstrate,insolvingtheproblemofenergyconser-vationinoperationalcomputations,thenewschemescaneliminatethe(nonlinear)computationalin-stabilityand,tosomeextenteventhe(nonlinear)computationaldivergingasfoundinthetraditionalschemes,Furthercontrastsbetweennewandtraditionalschemesalsoindicatethat,indiscreteopera-tionalcomputations,thenewschemeinthecaseofnondivergenceiscapableofprolongingthevalidin-tegraltimeofthecorrespondingtraditionalscheme,andeliminatingcertainkindofsystematicalcom-putational“climatedrift”,meanwhileincreasingitscomputationalaccuracyandreducingitsamountofcomputation.Theworkingprincipleofthispaperisalsoapplicabletotheproblemconcerningbaroclin-icprimitiveequations.
简介:IntheEnsembleKalmanFilter(EnKF)dataassimilation-predictionsystem,mostofthecomputationtimeisspentonthepredictionrunsofensemblemembers.Alimitedorsmallensemblesizedoesreducethecomputationalcost,butanexcessivelysmallensemblesizeusuallyleadstofilterdivergence,especiallywhentherearemodelerrors.InordertoimprovetheefficiencyoftheEnKFdataassimilation-predictionsystemandpreventitagainstfilterdivergence,atime-expandedsamplingapproachforEnKFbasedontheWRF(WeatherResearchandForecasting)modelisusedtoassimilatesimulatedsoundingdata.TheapproachsamplesaseriesofperturbedstatevectorsfromNbmemberpredictionrunsnotonlyattheanalysistime(astheconventionalapproachdoes)butalsoatequallyseparatedtimelevels(timeintervalis△t)beforeandaftertheanalysistimewithMtimes.Alltheabovesampledstatevectorsareusedtoconstructtheensembleandcomputethebackgroundcovariancefortheanalysis,sotheensemblesizeisincreasedfromNbtoNb+2M×Nb=(1+2M)×Nb)withoutincreasingthenumberofpredictionruns(itisstillNb).Thisreducesthecomputationalcost.Aseriesofexperimentsareconductedtoinvestigatetheimpactof△t(thetimeintervaloftime-expandedsampling)andM(themaximumsamplingtimes)ontheanalysis.TheresultsshowthatiftandMareproperlyselected,thetime-expandedsamplingapproachachievesthesimilareffecttothatfromtheconventionalapproachwithanensemblesizeof(1+2M)×Nb,butthenumberofpredictionrunsisgreatlyreduced.
简介:BasedontheTaylorseriesmethodandLi’sspatialdifferentialmethod,ahigh-orderhybridTaylor–Lischemeisproposed.Theresultsofalinearadvectionequationindicatethat,usingtheinitialvaluesofthesquare-wavetype,aresultwiththirdorderaccuracyoccurs.However,usinginitialvaluesassociatedwiththeGaussianfunctiontype,aresultwithveryhighprecisionappears.Thestudydemonstratesthat,whentheorderofthetimeintegralismorethanthree,thecorrespondingoptimalspatialdifferenceordercouldbehigherthansix.Theresultsindicatethatthereasonforwhythereisnoimprovementrelatedtoanorderofspatialdifferenceabovesixistheuseofatimeintegralschemethatisnothighenough.TheauthoralsoproposesarecursivedifferentialmethodtoimprovetheTaylor–Lischeme’scomputationspeed.Amorerapidandhighprecisionprogramthandirectcomputationofthehigh-orderspacedifferentialitemisemployed,andthecomputationspeedisdramaticallyboosted.Basedonamultiple-precisionlibrary,theultrahigh-orderTaylor–LischemecanbeusedtosolvetheadvectionequationandBurgers’equation.
简介:WhenlinearregressivemodelssuchasARorARMAmodelareusedforfittingandpredictingclimatictimeseries,resultsareoftennotsufficientlygoodbecausenonlinearvariationsinthetimeseries.Inthispaper,anonlinearself-excitingthresholdautoregressive(SETAR)modelisappliedtomodelingandpredictingthetimeseriesofflood/droughtrunsinBeijing,whichwerederivedfromthegradedhistoricalflood/droughtrecordsinthelast511years(1470—1980).TheresultsshowthatthemodelingandpredictingwiththeSETARmodelaremuchbetterthanthatoftheARmodel.Thelattercanpredicttheflood/droughtrunswithalengthonlylessthantwoyears,whiletheformalcanpredictmorethanthree-yearlengthruns.ThismaybeduetothefactthattheSETARmodelcanrenewthemodelaccordingtotherun-turningpointsintheprocessofpredic-tion,thoughthetimeseriesisnonstationary.
简介:AssimilatingsatelliteradiancesintoNumericalWeatherPrediction(NWP)modelshasbecomeanimportantapproachtoincreasetheaccuracyofnumericalweatherforecasting.Inthisstudy,theassimilationtechniqueschemewasemployedinNOAA’sSTMAS(Space-TimeMultiscaleAnalysisSystem)toassimilateAMSU-Aradiancesdata.Channelselectionsensitivityexperimentswereconductedonassimilatedsatellitedatainthefirstplace.Then,realcaseanalysisofAMSU-Adataassimilationwasperformed.Theanalysisresultsshowedthat,followingassimilatingofAMSU-Achannels5-11inSTMAS,theobjectivefunctionquicklyconverged,andthechannelverticalresponsewasconsistentwiththeAMSU-Aweightingfunctiondistribution,whichsuggeststhatthechannelscanbeusedintheassimilationofsatellitedatainSTMAS.WiththecaseoftheTyphoonMorakotinTaiwanIslandinAugust2009asanexample,experimentsonassimilatedandunassimilatedAMSU-AradiancesdataweredesignedtoanalyzetheimpactoftheassimilationofsatellitedataonSTMAS.TheresultsdemonstratedthatassimilationofAMSU-Adataprovidedmoreaccuratepredictionoftheprecipitationregionandintensity,andespecially,itimprovedthe0-6hprecipitationforecastsignificantly.
简介:ThenormalmodemethodisadoptedtodecomposethedifferencesbetweensimulationswithSST(seasurfacetemperature)anomahesovercentra-easternPacificandnormalSSTbyuseofanine-layerglobalspec-tralmodelinordertoinvestigateshort-rangeclimaticoscillationwithvarioustimescalesforcedbyElNinoduringthenorthernsummer.InvestigationshowsthatElNinomayhavethefollowinginfluenceonatmosphereonvariousspace-timescales.Extra-longwavecomponentsofRossbymodeforcedbyconvectiveanomalyoverequatorialwesternPacificresultingfromElNinoproduceclimaticoscillationonmonthly(sea-sonal)timescaleinmiddle-highlatitudesofSouthernandNorthernHemispheres;extra-longwavecomponentsofKelvinmodeforcedbySSTanomaliespropagatealongtheequator,resultingin30—60dayoscillationoftropicalandsubtropicalatmosphere;anditslongwavesmoveeastwardwithwesterly,resultinginquasi-biweekoscillation.
简介:基于三个全球年度吝啬的表面温度时间系列和三个中国年度平均数表面空气温度时间系列,多重timescales上的气候变化趋势被使用多滑动的时间窗户的趋势评价方法分析。结果被用来在1998-2012期间讨论所谓的全球温暖的中断。不同开始和结束时间在趋势评价的结果上有明显的效果,这被表明,并且当使用一扇短窗户时,含意特别地大。全球温暖的中断在1998-2012期间是在短timescales上看温度系列的结果;并且类似于它的事件,或有甚至冷的趋势的事件,实际上历史上多次发生了。因此,全球温暖的中断是可能的是长期的温度变化的一个期刊特征。它主要在短学期反映温度,和如此的现象的十的可变性不从长远的观点看改变全面温暖趋势。
简介:Thesecond-generationGlobalOceanDataAssimilationSystemoftheBeijingClimateCenter(BCC_GODAS2.0)hasbeenrundailyinapre-operationalmode.Itspanstheperiod1990tothepresentday.ThegoalofthispaperistointroducethemaincomponentsandtoevaluateBCC_GODAS2.0fortheusercommunity.BCC_GODAS2.0consistsofanobservationaldatapreprocess,oceandataqualitycontrolsystem,athree-dimensionalvariational(3DVAR)dataassimilation,andglobaloceancirculationmodel[ModularOceanModel4(MOM4)].MOM4isdrivenbysix-hourlyfluxesfromtheNationalCentersforEnvironmentalPrediction.Satellitealtimetrydata,SST,andin-situtemperatureandsalinitydataareassimilatedinrealtime.ThemonthlyresultsfromtheBCC_GODAS2.0reanalysisarecomparedandassessedwithobservationsfor1990-2011.TheclimatologyofthemixedlayerdepthofBCC-GODAS2.0isgenerallyinagreementwiththatofWorldOceanAtlas2001.ThemodeledsealevelvariationsinthetropicalPacificareconsistentwithobservationsfromsatellitealtimetryoninterannualtodecadaltimescales.PerformancesinpredictingvariationsintheSSTusingBCC_GODAS2.0areevaluated.ThestandarddeviationoftheSSTinBCC-GODAS2.0agreeswellwithobservationsinthetropicalPacific.BCC-GODAS2.0isabletocapturethemainfeaturesofE1NinoModokiIandModokiⅡ,whichhavedifferentimpactsonrainfallinsouthernChina.Inaddition,therelationshipsbetweentheIndianOceanandthetwotypesofE1NinoModokiarealsoreproduced.
简介:Inthispaper,theforecastingequationsofa2nd-orderspace-timedifferentialremainderarededucedfromtheNavier-StokesprimitiveequationsandEulerianoperatorbyTaylor-seriesexpansion.Hereweintroduceacubicsplinenumericalmodel(SplineModelforshort),whichiswithaquasi-Lagrangiantime-splitintegrationschemeoffittingcubicspline/bicubicsurfacetoallphysicalvariablefieldsintheatmosphericequationsonsphericaldiscretelatitude-longitudemesh.Anewalgorithmof'fittingcubicspline—timestepintegration—fittingcubicspline—……'isdevelopedtodeterminetheirfirst-and2nd-orderderivativesandtheirupstreampointsfortimediscreteintegraltothegoverningequationsinSplineModel.AndthecubicsplinefunctionanditsmathematicalpolaritiesarealsodiscussedtounderstandtheSplineModel’smathematicalfoundationofnumericalanalysis.ItispointedoutthattheSplineModelhasmathematicallawsof'convergence'ofthecubicsplinefunctionscontractingtotheoriginalfunctionsaswellasits1st-orderand2nd-orderderivatives.The'optimality'ofthe2nd-orderderivativeofthecubicsplinefunctionsisoptimalapproximationtothatoftheoriginalfunctions.Inaddition,aHermitebicubicpatchisequivalenttooperateonagridfora2nd-orderderivativevariablefield.Besides,theslopesandcurvaturesofacentraldifferenceareidentifiedrespectively,withasmoothingcoefficientof1/3,three-pointsmoothingofthatofacubicspline.Thentheslopesandcurvaturesofacentraldifferencearecalculatedfromthesmoothingcoefficient1/3andthree-pointsmoothingofthatofacubicspline,respectively.Furthermore,aglobalsimulationcaseofadiabatic,non-frictionaland'incompressible'modelatmosphereisshownwiththequasi-LagrangiantimeintegrationbyusingaglobalSplineModel,whoseinitialconditioncomesfromtheNCEPreanalysisdata,alongwithquasi-uniformlatitude-longitudegridsandtheso-called'shallowatmosphere'Navier-Stokesprimitiveequationsinthes