Processes Influencing Rain-Field Growth and Decay after Tropical Cyclone Landfall in the United States

PAGE 1

ProcessesInuencingRain-FieldGrowthandDecayafterTropicalCyclone LandfallintheUnitedStatesCORENEJ.MATYASDepartmentofGeography,UniversityofFlorida,Gainesville,Florida (Manuscriptreceived21June2012,innalform13December2012) ABSTRACT Thisstudymeasuredrain-eldsizesfortropicalcyclones(TCs)afterU.S.landfallandrelatedchangesin sizetothediurnalcycleandextratropicaltransition(ET).For45TClandfalls,thespatialpropertiesoftherain eldswerecalculatedthroughananalysisofradarreectivityreturnswithinageographicinformationsystem. Variablesrepresentingtheconditionsoftheatmosphereandstormattributeswereexaminedatthreetimes andaschangesovertwotimeperiodstoaccountforlagsbetweenconditiononsetandchangeinraining-area sizes.MannWhitney U testsillustratedwhichofthesevariableshadsignicantlydifferentmedianvalues whenthetotalrainingareaorhigh-reectivityregionsincreasedordecreasedinarealextentovertwo12-h periodsafterlandfall.Resultsindicatethatthediurnalcycleinuencedchangesinrain-eldsize.Rain-eld growthoccurredduringthelatemorningandearlyafternoon,whichisbetweenthetimesforpeakarealextent ofoceanic-andland-basedprecipitationinthetropics.TheraineldsofTCscompletinganETwithin74hof landfallincreasedinarealextentduringtherst12hafterlandfallanddecayedduringthesecond12-hperiod astheynearedthecompletionofET.Theavailabilityofmoisture,whichwasnotrelatedtoeitherthediurnal cycleorprocessesassociatedwithET,wasalsoimportanttorain-eldgrowthordecay.Inaddition,itwas discoveredthat,fortheUnitedStates,landfalltimeshaveshiftedfromapeakbeforemidnightduring195096 toaftermidnightduring19952008.1.IntroductionTherainfallthattropicalcyclones(TCs)produceas theymoveoverlandcanbebenecialtoalleviatedrought conditions(Maxwelletal.2012).Extensivefreshwater oodingcanalsooccur,however,especiallywhenthe stormisslowmoving(Konradetal.2002),theunderlying terrainissloped(Haggardetal.1973),and/orprevious rainshavesaturatedtheground(Sturdevant-Reesetal. 2001).Morethanone-halfofthedeathsintheUnited StatesthatarerelatedtoTCsarearesultoffreshwater ooding(Rappaport2000;Czajkowskietal.2011).When TCraineldsincreaseinsizeafterlandfall,morelocationsreceiverainfallandtheoveralldurationofrainfallis longerforagivenlocation.Bothconvectiveprecipitation andstratiformprecipitationoccurwithinTCs(Jorgensen 1984;YokoyamaandTakayabu2008),and,whetherrainfallismoderateorheavy,alongerdurationofrainfall increasesthechancesthat oodingandassociateddamage anddeathscanoccur.Abetterunderstandingofthe physicalprocessesthatareassociatedwithTCrain-eld growthanddecayasthesestormsmoveoverlandis neededtoimproverainfallforecasts. Inthetropicsandmidlatitudesduringthewarmseason, thegrowthanddecayofconvectivecloudsisstronglyassociatedwiththediurnalcycl e.Oceanicconvectionpeaks inthemorningbetween0600and1000LST,andresearchershavepostulatedthatdaynightdifferencesin radiativecoolingthatalterenvironmentallapserates withinandoutsidethestormenvironmentmayexplain thetimingofthispeak(GrayandJacobson1977;Yang andSmith2006;KikuchiandWang2008).AsTCsform overtheocean,theirdiurnalcyclesofconvectionexhibit amorningpeakcoincidentwithotheroceanicconvection asobservedbyresearchersexaminingcloud-toptemperaturesfromsatelliteobservations(Browneretal. 1977;Muramatsu1983;LajoieandButterworth1984; Sterankaetal.1984;Kossin2002).Itfollowsthatthearea coveredbyhighrainfallrateswouldbelargeraftermidnightthanafternoon.Rainfallobservationsfromislands inthetropicaloceansaswellassatellite-basedrainfall Correspondingauthoraddress: CoreneJ.Matyas,3141Turlington Hall,Dept.ofGeography,UniversityofFlorida,Gainesville, FL32611. E-mail:matyas@u.edu MAY2013MATYAS1085DOI:10.1175/JAMC-D-12-0153.1 2013AmericanMeteorologicalSociety

PAGE 2

estimationssupportthisexpectation(Frank1977;Lajoie andButterworth1984;Jiangetal.2011). Thetimingofpeakrainfallactivityoverlandmasses differsfromthatovertheocean.Diurnalchangesinair temperaturearegreateroverthelandsurfacethanover theoceanassensibleheatuxplaysalargerrolerelative tolatentheatuxintheenergybudget.Land-based convectionpeaksintheafternoonandearlyevening between1600and1900LSTwhenupwardsensibleheat uxismaximizedsothatthelowertroposphereis destabilized(Wallace1975;Dai2001;YangandSmith 2006;Daietal.2007).Oncefullyoverthelandsurface, TCraineldsmayalsobelargerintheafternoonrelativetothemorning.Yet,duringlandfallTCsmayspend severalhoursnearthecoastwherethetimingofpeak rainfallactivitycandifferfromthatoverlandorocean (YangandSmith2006).Muramatsu(1983)showedthat, fortwotyphoonsmovingwithin250kmofOkinawa,an afternoonpeakinconvectivecloudsdevelopedaround 1500LST,whichisslightlyearlierthanthepeakidentiedforland-basedconvectionbutmuchlaterthanthat foroceanicconvection.Alargersampleoflandfalling TCsmustbeanalyzedtobetterdeterminehowthediurnalcycleinuencesrain-eldgrowthanddecayfor TCsoverland. Processesrelatedtoextratropicaltransition(ET)also affectrain-eldgrowthordecay.Duringrecurvature, interactionwithmidlatitudetroughscanrestructurea TCintoanextratropicalcyclone,causingboththewind andraineldstoexpand(Rodgersetal.1991;Jones etal.2003).HartandEvans(2001)foundthatnearly one-halfofAtlanticOceanbasinTCscompleteanET andthatmostmajorcitiesnorthof35 8 Nintheeastern UnitedStateshavereceivedrainfallfromoneormoreof thesesystems.Thechangesinenvironmentalconditions duringETincludeincreasedbaroclinicity,enhanced horizontalmoisturegradients,highverticalwindshear, increasedpositiveverticalvorticityandrelativeeddy momentumuxconvergence,andfastersteeringows (AtallahandBosart2003;Jonesetal.2003).Isentropic ascentofwarmandmoisttropicalairaheadofthestorm oftencausesraineldstoexpandinadelta-shaped areaaheadofthestormcenter(HarrandElsberry2000; Kleinetal.2000;RitchieandElsberry2001).Meanwhile, theadvectionofrelativelydrierandcoolerairaround thesouthsideofthestormreducesrainfallinthatarea. AccordingtoKleinetal.(2000),thetimelineforET spans74h:transformationaverages46h,withanother 28hrequiredforreintensication. Alimitationofmanyprevioussatellite-basedstudies ofdiurnalpeaksinrainfallistheselectionofatemperaturethresholdforanalysisofcloud-toptemperaturesas thetimingofpeakarealcoverageisdependentuponthis threshold(Muramatsu1983;LajoieandButterworth 1984).Ground-basedradarreectivitydataprovide aspatiallyaccuraterepresentationofthearealcoverage ofTCraineldsoverlandareaswithoutthislimitation. Thisstudyemployedageographicinformationsystem (GIS)tomeasurethearealextentofradarreectivity returnsassociatedwith45TClandfallsintheUnited States.Themainhypothesiswasthatthediurnalcycle willcausegrowthtooccurlaterinthemorningrelative tooceanicconvectionbutearlierintheafternoonthanis seenforland-basedconvection.Thus,makinglandfall aftermidnightshouldresultinrainfallregionsofTCs thatincreaseinarealcoverageoverthenext12h,followedbyadecreaseinarealcoveragefor1224hafter landfall,whereastheoppositeshouldoccurforlandfall afternoon.Thesecondhypothesiswasthatprocesses associatedwithETalsoinuencechangesinrain-eld size.TCsclassiedasETwithin74hoflandfallshould exhibitthispatternasthisisthetimelineforETcompletionidentiedbyKleinetal.(2000).Totestthese hypotheses,casesweregroupedaccordingtowhether thetotalrainingareaorhigh-reectivityregionsincreasedordecreasedinarealextentover12-hperiods. MannWhitney U tests(MannandWhitney1947)were employedtodiscoverwhethervariablesassociatedwith thediurnalcyclesuchastimeofdayandrelativehumidity,and/orvariableslinkedtoconditionsassociated withETsuchasfastmotiontothenorthandeasthave statisticallysignicantdifferencesinmedianvaluesfor casesinwhichraineldsgrowversusthoseinwhichthey decay.2.Dataandmethods a.RadaranalysisThenumberofstormsexaminedwasdeterminedby theavailabilityoftheradarreectivitydatautilized todelineatethespatialextentoftherainelds.Data fromtheWeatherSurveillanceRadar1988Doppler (WSR-88D)networkareavailableineachdegreeofthe 360 8 sweepevery1kmoutwardfromtheradarsitewith anouterdistancelimitof230km(OFCM2006).During 19952008,therewere45casesforwhich1)radarreectivitydatawereavailableand2)theraineldsof theTCsremainedwithinrangeoftheWSR-88Dnetworkforaminimumof24hafterlandfall.Level-IIIbase reectivityproductdatawereutilizedtoretainalarge samplesize,andthesedatawereobtainedonlinefrom theNationalClimaticDataCenter'sarchive(http:// www.ncdc.noaa.gov/nexradinv/).Thisproductisgeneratedaftertheremovalofgroundclutterandconsistsof reectivityvaluesfromthelowestscan(0.5 8 tilt)thatare1086JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52

PAGE 3

roundedtothenearest5dB Z (OFCM2006).Level-II datathatincludevaluesfromallscanelevationsarenot availableforsixradarsafter2001;therefore,useofthese datawouldhavesignicantlydecreasedthenumberof casesanalyzed. Basereectivitydatacollectedduringthescannearest thetimeoflandfall t0,12hafterlandfall t1 12,and24h afterlandfall t1 24ateachradarwereimportedintoa GIS.Afterthedataweretransformedintoanequal-area projection,amosaicwascreatedforeachtimestep.The highestvaluewasretainedincasesinwhichdatafrom adjoiningradarsoverlapped.Thereectivityvalues werethencontouredin5-dB Z increments,anda10-km smoothinglterwasapplied.Aftertheconversionof contoursintopolygons(Fig.1),theirarealextentwas calculated.Intheeventthatresidualgroundclutterremained,polygonswitharealextentsoflessthan25km2wereremovedfromtheanalysis.Althoughtheauthor thenvisuallyinspectedeachcaseasanadditionalmeasureofqualitycontrol,itispossiblethatgroundclutter fromanomalouspropagationmaystillbeembedded withinprecipitation,whichcouldextendtheechoareal coverage.FromtheworkofpreviousTCresearchers, thetotalrainingareawasdenedastheareaoccupied byreectivityvaluesofgreaterthanorequalto20dB Z (Jorgensen1984;Barnesetal.1991;Matyas2007).This areaiscomposedoflow,moderate,andhighrainrates formedthroughstratiformandconvectiveprocesses (Jorgensen1984;Houze1993;Steineretal.1995).To examinethehigh-rain-rateregionsmostlikelytohave formedthroughconvectiveprocesses,theareasoccupied byreectivityvaluesgreaterthanorequalto40dB Z werealsoextractedforanalysis.Thisthresholdwasselectedbasedonpreviousresearchthatclassiedconvectiveregionsofclouds(Parrishetal.1982;Jorgensen1984; TokayandShort1996;BiggerstaffandListemaa2000). Theareaoccupiedbyreectivityvaluesatthetimeof landfallwassubtractedfromthatat t1 12tocalculatethe changeinareaovertherst12-hperiodfollowinglandfall(Per1).Areasat t1 12weresubtractedfromthoseat t1 24tocalculatethechangeinareaoverthesecond12-h periodfollowinglandfall(Per2)(Fig.2).b.Characterizationofstormattributesand environmentalconditionsAlldatapertainingtostormlocationandattributes aswellasenvironmentalconditionswereinterpolated linearlytothetimeoflandfall.Thetimeoflandfall, coordinatesofthecirculationcenter,andintensitywere obtainedfromtheHurricaneSeasonTropicalCyclone ReportsavailableonlinefromtheNationalHurricane Center(NHC;http://www.nhc.noaa.gov;Rappaportetal. 2009).Thesereportswerealsoutilizedtodetermine whetheraTCbecameextratropicalandthetimethatit didsoifapplicable.TheHurricaneDatabase(HURDAT; Jarvinenetal.1984)providedthecoordinatesofthe circulationcenter,stormheading,forwardvelocity,and intensityat0000,0600,1200,and1800UTC.Storm motionwassubdividedint onorthwardandeastward components. Atmosphericconditionswerecharacterizedthrough dataobtainedfromtheStatisticalHurricaneIntensity Scheme(SHIPS)database(DeMariaandKaplan1994; DeMariaetal.2005).TheSHIPSvariablesarederived fromtheNationalCentersforEnvironmentalPrediction (NCEP)GlobalForecastSystemmodelanalyses,and mostvariablesarecalculatedforanannularregionthatis FIG.1.Areascomposedofradarreectivityvaluesof20dB Z andhigherand40dB Z andhigherforHurricaneGaston(2004) atthethreeanalysistimesconsideredinthecurrentstudy( t0, t1 12, and t1 24). FIG.2.Timelineforanalysisofatmosphericconditionsand measurementsofrain-eldarea. MAY2013MATYAS1087

PAGE 4

200800kmfromthecirculationcenter.Dataareavailableatthestandardsynoptictimes.WithintheSHIPS dataset,deep-layerverticalwindsheariscalculatedover 850200hPa.Forthecurrentstudy,thisvectorwassubdividedintosouth-to-northandwest-to-eastcomponents. Vorticityat850hPaand200-hPadivergencearederived foraradiusof01000kmaboutthecirculationcenter, and200-hParelativeeddymomentumuxconvergence (REFC)isaveragedover100600km.Previousresearch (e.g.,Jonesetal.2003)suggeststhatmanyofthe10variablesderivedfromtheSHIPSdataset(Table1)should exhibitdifferencesforTCsthatbecomeextratropical within74hoflandfallascomparedwiththosethatdonot. Solarheatingduringthedayandlossoflongwave radiationduringthenightdrivethediurnalcycleofair temperaturenearEarth'ssurface,andchangesinrelativehumidity(RH)areaffectedbyairtemperatureas wellassoilmoistureandprecipitation(Dai2001).Thus, thestrongestevidencethatthediurnalcyclemayinuencechangesinthesizesofraineldswhileTCsmove overlandshouldbefoundthroughananalysisofnearsurfacetemperatureandRHvalues.Toexaminethese variablesevery3h,valueswereobtainedfromNCEP's NorthAmericanRegionalReanalysis(NARR)database(Mesingeretal.2006).AirtemperatureandRH dataat2mabovegroundlevelwereconvertedinto shapele''formatandenteredintoaGIS.Buffersof 200and800kmfromthecirculationcenterateachanalysistimewereusedtoselectonlythedatapointsinside thisdistancerange.Theaver agevalueofdatapointsover thisregionwasthenutilizedintheanalysisforcompatibilitywiththeotherSHIPSvariables. Thevaluesforallvariablesmentionedabovewere analyzedatthreedifferenttimesrelativetolandfall. Previousresearchhasshownthatatimelagof1224h existsbetweentheonsetofenvironmentalconditions andresultingchangesinTCstructure(e.g.,Frankand Ritchie1999;Kimball2008;Matyas2010).Toaccountfor thislageffect,environmentconditionswereanalyzedat t2 12, t0,and t1 12,whichcoincidewith24,12,and0hprior totheendofPer1and36,24,and12hpriortotheendof Per2(Fig.2).Inaddition,thechangesintheenvironmentalconditionsovereach12-hperiodwerecalculated asthesevaluesmaybemorerepresentativeofthediurnal cycleorprogressionofconditionsassociatedwithET thantheactualvalueofthevariableatanyonetime. Forexample,a12-hincreaseinRHanddecreaseinair temperatureindicatethatthemidpointoftheperiod occursneardawn,whereasincreasingstormforwardvelocitycoincideswiththeprogressionofanET.Thus,each environmentalconditionderivedfromHURDAT,SHIPS, andNARRwasexploredatthreeinstantaneoustimes ( t2 12, t0,and t1 12)andovertwoperiods( t2 12: t0and t0: t1 12),yielding90totalvariables(Table1). AsaTCmovesinland,itscenterofcirculationbecomesfartherremovedfromthewarmoceanwatersthat supplytheprimarysourceoflatentheattosustainthe storm(e.g.,Tuleya1994).Itfollowsthat,ingeneral,the TABLE1.Name,abbreviation,unitofmeasure,andsourceforvariablesthatwereusedinthestudy.VariablesfromHURDAT,SHIPS, andNARRhavevevariations(threeinstantaneoustimesandtwochangesovertime)thatweresubjectedtoMannWhitney U tests. VariableAbbreviationUnitsSource LatitudeLat 8 NHURDAT LongitudeLon 8 WHURDAT MotionnorthMotNms2 1HURDAT MotioneastMotEms2 1HURDAT Velocityofmaxsustainedwinds Vmaxms2 1HURDAT MincentralpressureMCPhPaHURDAT SoutherlyverticalwindshearShrSms2 1SHIPS WesterlyverticalwindshearShrWms2 1SHIPS Airtemperatureat200hPaT200 8 CSHIPS Zonalwindsat200hPaU200ms2 1SHIPS Divergence200hPaD200s2 13 107SHIPS Relativehumidityhigh(500300hPa)RhHi%SHIPS Relativehumiditymiddle(700500hPa)RhMd%SHIPS Relativehumiditylow(850700hPa)RhLo%SHIPS Vorticityat850hPaZ850s2 13 107SHIPS RelativeeddymomentumuxconvergenceREFCms2 1day2 1SHIPS Airtemperatureat2mT2m 8 CNARR Relativehumidityat2mRh2m%NARR Distanceinland12hafterlandfallDin12kmGISanalysis Distanceinland24hafterlandfallDin24kmGISanalysis Sineoflandfalltimeminusvalue*e.g.,SinLT-90DimensionlessNHCreport *Eightvalueswereutilized:0,45,90,135,180,225,270,and315.1088JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52

PAGE 5

greaterthedistancebetweenthenearestpointalongthe coastlineandthecirculationcenterofaTCis,thelessis thepotentialforrainfallenhancementastheavailability oflow-levelmoistureisreduced.Remainingnearthe coastlinemayallowraineldstogrowregardlessofthe timeofdayorwhetherthestormisundergoingET. WithinaGIS,theU.S.coastlinefromBrownsville, Texas,toEastport,Maine,wasconvertedintoaline featureandtheNear''functioncalculatedthedistance betweenthepositionofthecirculationcenterat12and 24hafterlandfallandthenearestpointalongtheline representingthecoast. TheinclusionofthelocaltimeofeachTC'slandfallin thestatisticalanalysisisimportantbecauseitreveals whichlandfalltimesaremostassociatedwithpostlandfallgrowthordecay.Forexample,iftheraineldsof landfallingTCsexhibitadiurnalcyclesimilartothatof ocean-basedconvectioninwhichpeakactivityoccursat 06001000LST,thenarain-eldsizethatislargerat t1 12thanatlandfallshouldoccurforlandfalltimesranging from1800to2200LST.Aftertheconversionoflandfall timefromUTCtoLST,eachhourwasmultipliedby 15toextendtherangeofvaluesto360.Sothatlandfall timesoccurringevery3hwererepresentedinthe analysis,avalueof45wassubtractedfromeachconvertedtime.Afterconvertingthesevaluesintoradians, thesineofalleightvalueswascalculated.Asaresult,the sineofthelandfalltimeminus180(SinLT-180)producesavalueof1.0forthelandfalltimeof1800LST and 2 1.0forthetimeof0600LST.Ifthediurnalcycle forlandfallingTCswascoincidentwiththatofoceanicbasedconvection,thenrain-eldgrowthwouldoccur duringthe12-hperiodsof18000600and21000900 LST,andtheseperiodswouldbeindicatedbyvalues near1.0forvariablesSinLT-180andSinLT-225.c.StatisticalanalysesThisstudyemploysnonparametricMannWhitney U tests(MWU;MannandWhitney1947)torelatedifferencesinenvironmentalconditionstorain-eldgrowth ordecay.Thistestcomparesthemediansofvariables whencasesaredividedintotwoindependentgroups,with anullhypothesisthatdatafromthetwogroupsoriginate fromthesamepopulation.Thistestwasutilizedbecause itismorerobustforskeweddatathanisanindependent samples t testandislesslikelytoyieldfalseresultswhen groupsizesaresmalland/ornotequal(Wilks1995).For eachvariable,valuesarerankedregardlessofgroup membership.Theteststatisticthencomparesthesumof theranksfromtherstgroupwiththatofthesecond.If thesamplesaredrawnfromthesamepopulation,the summedtotalofranksforeachgroupshouldbesimilar. Inthecurrentstudy,thenullhypothesiswasrejected when p valueswerelessthanasignicancelevelof 0.05,whichisacommonlyusedlevelaccordingtoWilks (1995).Althoughthevariablestestedarenotindependent,theperformanceof100individualMWUtests withineachsetasdescribedbelowdoesincreasethe possibilityofincorrectlyrejectingthenullhypothesis whentheexperiment-wiseerrorrateisconsidered.However,thetechniqueisvalidgiventhattheacceptanceof thetwomainhypothesesofthestudydoesnotrelysolely ontheresultsofanyonetest;anumberofvariablesare representativeofthediurnalcycleandET.All p values meetingthe0.05thresholdarereportedsothatresultscan alsobescrutinizedmorestringentlyifdesired. ForeachMWUtest,casesweregroupedaccordingto whetherrainfallareasincreasedordecreasedduring a12-hperiod.Twodifferent12-hperiodsandtworadar reectivitythresholdstocalculatetherain-eldarea wereanalyzed,yieldingfoursetsofMWUtestsinwhich all100variables(discussedinsection2b)weregrouped accordingtorain-eldgrowthordecay.Set1examined rain-eldgrowthordecayfortotalrainingareaduring Per1,andset2comparedgrowthanddecayforhighreectivityregionsduringPer1.Insets3and4,cases weregroupedaccordingtorain-eldgrowthordecay duringPer2,withset3(4)focusingontotalrainingarea (high-reectivityregions).Variablesforwhichthenull hypothesisisrejectedidentifytheconditionsassociated withrain-eldgrowthordecayduringeachperiod.The inuenceofthediurnalcyclemaybeindicatedbyvariablessuchas12-hchangesinnear-surfaceairtemperatureandRHalongwiththevariationsinsineofthe landfalltime.Inotherwords,ifthemeanrankofthe 12-hchangeinairtemperatureorRHforrain-eld growthcasesdiffersfromthemeanrankofthe12-h changeforrain-elddecaycases,thenadiurnaleffect maybeoccurring.In20cases,TCsmadethetransition toextratropicalcycloneswithin74hoflandfall.Ifthe nullhypothesisisrejectedforvariablescharacterizing stormmotion,vorticity,REFC,andverticalwindshear, itindicatesthatprocessesassociatedwithETinuence rain-eldgrowthanddecay.Notethatlandfalloccurred atvarioustimesofdayforTCsthatbecameextratropicalwithin74hoflandfall.3.Results a.Rain-eldsizesandstatisticsAtthetimeoflandfall,themedianraineldhadan areaof109263km2andthemedianvalueforhighreectivityregionswas10473km2(Fig.3).Hurricane Floyd(1999)wasanoutlier(notvisibleinFig.3),having thelargestextentofbothmeasuresofrainfallarea.TCsMAY2013MATYAS1089

PAGE 6

suchasFloydthatbecameextratropicalwithin74hhad thelargestrainelds(Fig.3).Nineof13TCswithtotal rainingareasthatwerelargerthan150000km2and11 of16TCswithhigh-reectivityregionsthatspanned morethan15000km2completedanETwithin60hof landfall.Spearman'srankcorrelationcoefcients(not shown)demonstratethatthetimeoflandfallwasnot signicantlycorrelatedwithrain-eldsize.Yet,thediurnalcyclemaybeinuencingthetimingofrain-eld growthanddecaygiventhatoverPer1largerrainelds decreasedwhilesmallerraineldsincreasedinareal coverage.Asaresult,themeanandmedianareaswere similartothevaluesatlandfall.DuringPer1,23(21)of 45casesexperiencedincreasesintotalrainingarea (high-reectivityregions).Rain-eld(highreectivity) growthoccurredfor14(16)of45casesduringPer2. Althoughby t1 24medianrainfallareashaddecreased fromthatatlandfall,19(12)TCsincreasedinareal coveragefortotalrainingarea(high-reectivityregions) duringthisperiod,withameanincreaseof27734km2(4228km2).b.ResultsoftheMannWhitney U testsTheoutcomesofthefoursetsofMWUtestssupport thehypothesesthatthediurnalcycleandETaffectraineldgrowthanddecay.Therewere51variableswith p valuesoflessthanorequalto0.05(Tables25).Of these,20wereuniquetoonesetandtwovariationsof thesineoflandfalltimeweresignicantinallfoursetsof tests,indicatingthatrain-eldgrowthoccurredwhen eitherPer1orPer2begannear0000and0300LST.The statisticallysignicantvariablesarebroadlycategorized asrelatingto1)thediurnalcycle,2)ET,and3)the availabilityofmoisture.Atleastonevariablefromeach categoryappearedineachofthefoursets,suggesting thatallthreeprocessesinuencerain-eldgrowthand decayafterlandfall.Theresultsalsodemonstratethe importanceofincludingbothinstantaneousandchangeinvariablesaswellas12-and24-htimelagsbetween conditiononsetandrain-eldresponse.Amorestringent p levelof0.025(0.01)yields21(9)signicantvariablesacrossthefoursetsoftests.1)RAIN-FIELDGROWTHANDDECAYANDTHE DIURNALCYCLETCsmakinglandfallnearmidnightand0300LST tendedtogrowduringPer1anddecayduringPer2, whereastheoppositewastrueforlandfallsoccurring nearnoonand1500LST(Fig.4).Statisticalevidence supportingthispatternwasfoundthroughthetimerelatedvariablesandnear-surfacetemperatureandRH valuesashypothesized.Casesinwhichlandfalloccurred closertomidnight(valueofSinLT-270closeto1.0)or 0300LST(valueofSinLT-315closeto1.0)wereassociatedwithgrowthduringPer1(Tables2and3)and decayduringPer2(Tables4and5).Landfallsoccurring FIG.3.ArealcoverageofraineldsforTCsthatbecameextratropical(labeledET)within74hoflandfallascomparedwiththosethat didnot(nonET'')at t0, t1 12,and t1 24andthechangeinareaforTCsthatdecayed(decreasing)andgrew(increasing)during t0: t1 12and t1 12: t1 24for(a)totalrainingareaand(b)high-reectivityregions.HurricaneFloyd(1999)wasanoutlier,withvaluesthatwouldappear offthetopofeachpanel.1090JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52

PAGE 7

approximately12hoffsetfromthesetimes(valuesof SinLT-270andSinLT-315closeto 2 1.0)experienced decayatrstfollowedbygrowthduringPer2.Forboth analysisperiods,rain-eldgrowthwasassociatedwith decreasingairtemperaturesandincreasingRHvalues nearthesurfaceduringtheprevious12h,indicatingthat thisprecedingperiodwasduringthenight.DuringPer2, growthwasassociatedwithcaseshavinglowerRH valuesatthetimeoflandfall(whichoccurredduringthe day)andvaluesdecreasingpriortoandincreasingafter landfall.Correspondingly,near-surfacetemperatures increasedpriortolandfallanddecreasedafterlandfall forTCswhoseraineldsgrewduringPer2.Overall, growthtendedtooccurduringthelate-morningandearlyafternoonhoursasarealcoveragewasgreaterat1200and 1500LSTthanitwasat0000and0300LST. Thesetimesforrain-eldgrowthoccurbetweenthe timesreportedbypreviousresearchersforoceanicand land-basedconvection.Becausemostoceanicconvection peaksbetween0600and1000LST(GrayandJacobson 1977;YangandSmith2006;KikuchiandWang2008), landfallingTCsfollowingthispatternwouldhavehad valuesnear1.0forvariablesSinLT-180andSinLT-225, indicatinglandfallat1800or2100LSTandrain-eld growthoverthenext12h.Thisscenariodidnotoccur. Afternoonpeaksinrainfalloverlandcorrespondtotimes of16001900LST(Wallace1975;Dai2001;Daietal. 2007).TCsmakinglandfallnear0600LSTdidnottend tohavelargerrainingareas12hlaterastheanalysis ofvariableSinLT-0didnotproducestatisticallysignicantresultsintheMWUtests.Oneexplanationforthese ndingsisthatTCsaremakingthetransitionfromthe diurnalcycleofanocean-basedstormtothatofalandbasedstormduringlandfall.AsTCsadvectmoistureladentropicalairmassesoverthelandsurfaceduringthe morninghours,itispossiblethatinstabilityisenhanced sothatconvectiondevelopsearlierinthedaythanwhen thenormalcontinentalairmassisinplace.Knowledge TABLE2.ResultsofMannWhitney U testsforrain-eldgrowthvsdecayfortotalrainingareaduringPer1.FormeaningoftheabbreviatedvariablenamesseeTable1. VariableTimeMedianvaluegrowMedianvaluedecay U SignicanceAssociation Din12 t1 1266.22144.101340.007Moisture SinLT-270* t00.71 2 0.26367.50.009Diurnal RhMd t2 12: t02.00 2 1.253640.011Moisture SinLT-315** t00.71 2 0.263590.016Diurnal MotE t1 120.28 2 2.063500.026ET Z850 t2 12: t010.25 2 5.213490.027ET RhLo t2 12: t02.30 2 0.10346.50.031Moisture Lon t0: t1 120.50 2 0.70345.50.033ET RhHi t2 12: t01.00 2 1.653400.045Moisture MotE t2 12: t00.780.143380.050ET REFC t2 12: t05.00 2 0.383380.050ET *Highpositivevaluesindicatelandfalltimecloseto0000LST. **Highpositivevaluesindicatelandfalltimecloseto0300LST. TABLE3.AsinTable2,butforhigh-reectivityregionsduringPer1. VariableTimeMedianvaluegrowMedianvaluedecay U SignicanceAssociation Din12 t1 1245.64145.751090.001Moisture ShrS t1 122 1.341.921100.002ET REFC t2 125.338.581360.008ET SinLT-270* t00.71 2 0.26358.50.015Diurnal Rh2m t2 12: t07.04 2 2.313510.024Diurnal SinLT-315** t00.71 2 0.26349.50.026Diurnal RhMd t2 12: t02.00 2 0.65345.50.033Moisture T2m t2 12: t02 2.740.061600.036Diurnal RhHi t2 12: t00.80 2 2.15341.50.042Moisture Lon t1 122 83.50 2 88.743400.045ET Rh2m t082.7778.023400.045Diurnal Lon t2 122 83.30 2 88.58339.50.047ET Lon t02 83.65 2 88.733380.050ET *Highpositivevaluesindicatelandfalltimecloseto0000LST. **Highpositivevaluesindicatelandfalltimecloseto0300LST. MAY2013MATYAS1091

PAGE 8

thatrainfallmaycommenceearlierinthedaythanexpectedforalocationallowsweatherforecasterstoalert thepublicaspreparationsaremadeforthearrivalofTC conditions. Thecurrentstudyisbelievedtobethersttoanalyze radarreectivityreturnsforalargesampleoflandfalling TCsintheUnitedStatesandtoassociaterain-eldgrowth anddecaywiththediurnalcycle.Jiangetal.(2011)providedtherstdocumentationo fdiurnalvariationsofglobal TCrainfallusingdatafromtheTropicalRainfallMeasuring Mission(TRMM)satellite.TheTRMManalysisshowed thatvolumetricrainfallfromTCsoverlandpeakedat0100 0730LSTand16301930LST.However,Jiangetal.(2011) examinedglobalTCrainfallvariations.TheirFig.12shows littlediurnalvariationforAtlantic-basinstormsascomparedwiththeotherTCbasins.Atmosphericconditionsdo varyclimatologicallyamongthedifferentTCbasins.For example,VincentandFink(2001)discussdifferencesin theprecipitablewatercontentforthewesternandeastern PacicTCbasins.Thus,itis possiblethatthetimingof thepeakoverlandrainfallfromAtlantic-basinTCsdiffers fromthatobservedinotherbasinswhereTCrainratescan behigher.Also,studieshaveshownthattheshapeofthe coastlineaswellastopographyofthelandsurfacecanalter stormstructure.TheuniquefeaturesoftheU.S.coastline maycontributetoapeakinrainfallthatdiffersintiming fromotherlandfalllocations(e.g.,RogersandDavis1993; Cubukcuetal.2000;Liuetal.2007;Kimball2008;AuYeungandChan2010).2)CONDITIONSRELATEDTOEXTRATROPICAL TRANSITIONOfthe45casesexamined,20becameextratropical within74hoflandfall(Fig.4).ThisstudyndsthatET contributestooverallrain-eldgrowthearlyinthe processanddecayastheprocessnearscompletionwhile TABLE4.AsinTable2,butfortotalrainingareaduringPer2. VariableTimeMedianvaluegrowMedianvaluedecay U SignicanceAssociation Rh2m t077.1982.161070.007Diurnal SinLT-315* t00.71 2 0.441190.016Diurnal RhLo t0: t1 123.45 2 0.30315.50.017Moisture T2m t2 12: t00.19 2 2.713130.019Diurnal Rh2m t2 12: t02 2.867.041290.031Diurnal RhMd t0: t1 121.90 2 1.303070.033Moisture SinLT-270** t00.61 2 0.791320.037Diurnal Rh2m t0: t1 126.77 2 5.893020.037Diurnal Lat t0: t1 121.391.731350.044ET T2m t0: t1 122 2.410.811380.050Diurnal *Highpositivevaluesindicatelandfalltimecloseto0300LST. **Highpositivevaluesindicatelandfalltimecloseto0000LST. TABLE5.AsinTable2,butforhigh-reectivityregionsduringPer2. VariableTimeMedianvaluegrowMedianvaluedecay U SignicanceAssociation Lon t0: t1 122 1.120.62103.50.002ET REFC t2 12: t02 2.463.501090.004ET Lon t2 12: t02 1.160.23113.50.005ET MotE t1 122 2.630.94128.50.014ET T2m t2 12: t00.36 2 2.713360.014Diurnal Rh2m t076.4882.031290.015Diurnal Lat t0: t1 121.222.13131.50.017ET T2m t027.8126.153300.020Diurnal MotN t2 12: t02 0.081.631410.031ET U200 t2 12: t02 1.254.481410.031ET MotE t02 1.930.681450.039ET T2m t0: t1 122 1.730.411450.039Diurnal SinLT-315* t00.71 2 0.44146.50.042Diurnal Rh2m t2 12: t02 1.366.971480.046Diurnal RhMd t0: t1 120.68 2 1.65318.50.047Moisture Rh2m t0: t1 126.41 2 4.983150.049Diurnal SinLT-270** t00.71 2 0.13150.50.050Diurnal *Highpositivevaluesindicatelandfalltimecloseto0300LST. **Highpositivevaluesindicatelandfalltimecloseto0000LST.1092JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52

PAGE 9

high-reectivityregionsmainlydecayed.DuringPer1, ETstormsexhibitinggrowthofthetotalrainingarea movedfastertotheeastwhilepositivevorticityand REFCincreased(Table2).Thisnetgrowthcorresponds todescriptionsofthedevelopmentofthedelta-shaped rainshieldthatoccursaheadofTCsintheearlystagesof ET(HarrandElsberry2000;Kleinetal.2000;Ritchie andElsberry2001).Thetropicalairmassthatisadvectedoverthecooleranddrierairmassaheadofthe stormcentercreatesastableenvironmentthatfavors stratiformratherthanconvectiveprocesses.Thisexplainswhyhigh-reectivityregionsdecayedforET cases.HighervaluesofverticalwindshearandREFC aswellaslongitudeslocatedfarthertotheeastascomparedwithcasesinwhichhigh-reectivityregionsgrew supportthendingthatETcontributestothisdecrease inhigh-reectivityregions(Table3).FortheveET casesfeaturinggrowthinhigh-reectivityregionsduring Per1,landfalloccurrednearmidnightLST,indicating thatthediurnalcyclemayhavecontributedtothe growthoftheseregionsinthelate-morninghours.Thus, itislikelythatthediurnalcycleaffectsconvectiverainfallgrowthanddecayevenwhenETisoccurring. Theadvectionofrelativelycoolanddryairaroundthe westernandthensouthernsideoftheTCdecreases rainfallproductionduringstep3ofthetransformation stageofET(Kleinetal.2000).Thecurrentstudysupportsthispreviouswork.Saveforonecaseinwhichthe diurnalcyclemayhavecontributedtorain-eldgrowth, allTCscompletingETwithin30hoflandfallexperienceddecayinbothrainfallareasduringPer2.TCs movingfarthertothenorthdecayedintotalrainingarea ascomparedwiththosethatremainedsouth(Table4). Forhigh-reectivityregions,decayduringPer2wasassociatedwithETasfastnorthwardandeastwardmotion andfastwesterlywindsat200hPawithincreasedREFC beforelandfall(Table5).3)IMPORTANCEOFMOISTUREFORGROWTH ANDDECAYDuringPer1,oneofthemostsignicantdifferences betweencasesexperiencingrain-eldgrowthordecay wasthedistanceinlandtraveledbystormsineachgroup. TCsremainingclosertothecoastlinegrewwhilethose locatedmorethantwiceasfarinlanddecayed(Tables2 and3).PreviousresearchhasshownthatTCsremaining closertothecoastlinecanhavelargerraineldsthat producemorerainfallbecauseremainingnearthecoastlineallowsTCstodrawinlow-levelmoisturefromthe oceantoincreaserainfallproduction(Lonfatetal.2007; Matyas2007;Medlinetal.2007).Yet,thedistancetraveled inlandby t1 24wasnotsignicantforPer2,andSpearman's rankcorrelationcoefcientscalculatedbetweenvariable Din24andtheactualarealcoveragerainfall(notshown) didnotproducestatistically signicantresults.Previous researchhasshownthatrainf allproductioncanstillincreaseforTCslocatedmorethan500kminlandifalowleveljetadvectsmoistureintothestorm,soilmoistureis highenoughtosupplymorelatentheatuxthanisexperiencedoverdryland,ororographicenhancementof rainfalloccurs(Haggardetal.1973;BluesteinandHazen 1989;Tuleya1994;Emanueletal.2008;Arndtetal.2009; Kellneretal.2012).Thus,trackingfarinlanddoesnot necessarilymeanthatraineldswilldecreaseinareal coverage. FIG.4.Tracksofthe45TCs,indicatingwhetherlandfalloccurred closertonoonormidnight,whetherEToccurredwithin74hof landfall,andwhetherraineldsgrewordecayedduringPer1and Per2for(a)totalrainingareaand(b)high-reectivityregions. MAY2013MATYAS1093

PAGE 10

Moistureabovetheboundarylayerisalsoimportant asrain-eldgrowthwasassociatedwithincreasingRH valuesinthelow,middle,orhightroposphereduringthe previous12h(Tables25).ArelativelymoistsurroundingenvironmentiskeytoincreasedTCrainfall production(Jiangetal.2008;HillandLackmann2009; Matyas2010).Theadvectionofdryairintothesystem enhancesevaporation,whichleadstoadecreasein rainfallproduction(Chanetal.2004;Kimball2008). AdditionalMWUtestsperformedwiththeseRHvariablesandcasesgroupedaccordingtowhetherornotET occurredwithin74hoflandfallandwhetherlandfall occurredduringthedayornightindicatedthatchanges inRHabovetheboundarylayerwerenotstronglyassociatedwitheitherthediurnalcycleorET.c.DiurnalvariationsinlandfalltimesAnexaminationofthedistributionoflandfalltimes forTCsinthecurrentstudyyieldsaninterestingresult whencomparedwithapreviousstudy.One-thirdofthe landfallsinthecurrentstudyoccurredbetween0000and 0300LST(Fig.5).Incontrast,Konrad(2001)foundthat for195096nearlyone-halfofTCslandfallingalongthe GulfofMexicoandAtlanticcoastlinesoftheUnited Statesdidsobetween1700and0000LSTwhereasonly fourlandfallsoccurredbetween0100and0300LST. Afterbinningthelandfalltimesinthecurrentstudy accordingtothetimegroupslistedinTable1ofKonrad (2001),achi-squaretestshowedstatisticallysignicant differencesbetweentheobservedandexpecteddistributionsofcasesamongthelandfall-timecategoriesfor thetwodatasets( x25 10.912; p 5 0.012).Toruleoutthe possibilitythatthecurrentstudy'ssamplewasnotrepresentativeofalllandfallsduring19952008,anadditional chi-squaretestrevealedthatthedistributionofthecurrent studyverystronglyresemblesthedistributionoftheentire population( x25 0.836; p 5 0.858).Thus,itappearsthat ashiftinlandfalltimeshastakenplacefrom195096, whenthemajorityoflandfallsoccurredintheevening hourspriortomidnight,to19952008,whenmanyoccurredatorjustaftermidnight. AshiftinthemajorityofTClandfallsfrombeforeto aftermidnighthasimportantimplicationsforthereceipt ofhurricanewarningmessagesbythepublic.Priorto 2010,theNHCissuedhurricanewarningswhenhurricaneconditionswereexpectedfortheareawithin24h (Sheets1990;Rappaportetal.2009).Withpeaklandfalls occurringbetween1700and0000LST,thesewarnings wouldlikelyhavereachedpeopleduringthelateafternoonandeveninghours,givingthemadequatetimeto makepreparations.Duringthemorerecentperiodwhen landfallspeakedatandaftermidnight,thesewarnings weremorelikelytohavebeenissuedafterpeoplehad retiredfortheevening.Uponawakeningthenextmorning, theymaythenhavehadlessthan18htocompletepreparationsforthearrivalofh urricane-forcewinds.In2010, theNHCchangeditsleadtimesothathurricanewarnings arenowissued36hinadvanceoftheanticipatedonsetof tropicalstormforcewinds(NHC2012).Thenewlonger leadtimesreducetheproblemswithTCsmakinglandfall aftermidnight,whichhasbecomemorecommoninrecent years(Fig.5).4.ConclusionsandfutureresearchInthisstudy,thearealcoverageofTCraineldswas examinedafterU.S.landfallthroughaGIS-basedspatialanalysisofradarreectivityreturns.Thegrowthand decayofthetotalrainingareaandhigh-reectivityregionswereexaminedovertwo12-hperiods.Variables characterizingthetimeofday,conditionsoftheatmosphere,anddistancefromthecoastlinewereexamined infoursetsofMannWhitney U testsinwhichcases weregroupedaccordingtowhetherrainingareasincreasedordecreasedovera12-hperiod.Changesin rain-eldsizewereassociatedwiththediurnalcycleas growthoccurredduringthelatemorningandearlyafternoonaftermaximumRHandminimumairtemperatureshadbeenreached.Raineldsdecreasedinareal extentinthelate-eveninghours.TCscompletinganET within74hoflandfallhadthelargestraineldsatthe timeoflandfall,andtotalrainingareaincreasedduring therst12hafterlandfallbutdecreasedduringPer2. RegionsofhighreectivitydecreasedinarealcoverageforETcasessavewhenthediurnalcyclelikely contributedtogrowthinthelatemorningandearlyafternoon.Moisturewasalsoimportantforrain-eld FIG.5.Frequencyoflandfalltimesforthe45casesinthestudy.1094JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52

PAGE 11

growth.Abovetheboundarylayer,increasingRHinthe preceding12hwasrelatedtorain-eldgrowthaswas remainingnearthecoastlineduringPer1.Inaddition, thisstudyfoundthatmoreTCsmadelandfalljustafter midnightascomparedwith195096whenmostlandfalls occurredjustpriortomidnight.Thisndingislikelydue toimprovementsinobservingsystems. Thisstudyhasprovidedobservationalevidencefor diurnalvariationsofrainingareasofTCsevenasthey facerapidlychangingenvironmentalconditionswhile movingovertheUnitedStates.Associationsbetween environmentalconditionsandrain-eldgrowthand decaywereexploredonabroadspatialscaleintermsof theenvironmentalvariablesandonatemporalscalein termsoftheradaranalysis.Thelogicalnextsteptoward conrmingalinkbetweenthediurnalcycleandTCrain eldsoverlandthroughobservationaldataistoexamine rain-eldevolutionatahighertemporalresolution. Furtheranalysesutilizingradarreectivitydatafromthe WSR-88DnetworkarejustiedbecausetheprecipitationradarswathusedtocreatetheTRMMdatasetis only250kmwide.Also,WSR-88Dreectivitydataare availableevery56min,whichisahighertemporal resolutionthanisavailablefromtheTRMMdata.The analysisofWSR-88Ddataoverhourlyperiodswillallow thetimingofpeakrainfallwithinthediurnalcycleto bemorepreciselydenedalongwiththerateatwhich growthanddecayoccur.Inclusionoflevel-IIreectivity datainfutureanalysiswouldallowtheverticaldevelopmentofTCraineldstobeanalyzedtoaidinthe identicationofconvectiveregionsoftherainelds wherethehighestrainratesoccur.Itwillalsobeadvantageoustomodelatmosphericconditionsatahigher spatialresolutionthanthe200800-kmaverageprovidedbytheSHIPSdataset,particularlyinlightofthe associationsbetweenmoistureandrain-eldgrowth identiedbythecurrentstudy.Itwouldbeappropriate tocharacterizeenvironmenta lconditionsutilizingNARR datagiventhattheyareavailableevery3hata32-km spatialresolution(Mesingeretal.2006). Acknowledgments. Thisresearchwassupportedby aNationalScienceFoundationCAREERAwardBCS1053864.Thecommentsoffouranonymousreviewers improvedthismanuscript.REFERENCES Arndt,D.S.,J.B.Basara,R.A.McPherson,B.G.Illston,G.D. McManus,andD.B.Demko,2009:Observationsofthe overlandreintensicationofTropicalStormErin(2007). Bull. Amer.Meteor.Soc., 90, 10791093. Atallah,E.H.,andL.R.Bosart,2003:Theextratropicaltransition andprecipitationdistributionofHurricaneFloyd(1999). Mon.Wea.Rev., 131, 10631081. Au-Yeung,A.Y.M.,andJ.C.L.Chan,2010:Theeffectofariver deltaandcoastalroughnessvariationonalandfallingtropical cyclone. J.Geophys.Res., 115, D19121,doi:10.1029/2009JD013631. Barnes,G.M.,J.F.Gamache,M.A.Lemone,andG.J.Stossmeister, 1991:Aconvectivecellinahurricanerainband. Mon.Wea. Rev., 119, 776794. Biggerstaff,M.I.,andS.A.Listemaa,2000:Animprovedscheme forconvective/stratiformechoclassicationusingradarreectivity. J.Appl.Meteor., 39, 21292150. Bluestein,H.B.,andD.S.Hazen,1989:Doppler-radaranalysisof atropicalcycloneoverland:HurricaneAlicia(1983)in Oklahoma. Mon.Wea.Rev., 117, 25942611. Browner,S.P.,W.L.Woodley,andC.G.Grifth,1977:Diurnal oscillationoftheareaofcloudinessassociatedwithtropical storms. Mon.Wea.Rev., 105, 856864. Chan,J.C.L.,K.S.Liu,S.E.Ching,andE.S.T.Lai,2004: AsymmetricdistributionofconvectionassociatedwithtropicalcyclonesmakinglandfallalongthesouthChinacoast. Mon. Wea.Rev., 132, 24102420. Cubukcu,N.,R.L.Pfeffer,andD.E.Dietrich,2000:Simulationof theeffectsofbathymetryandlandseacontrastsonhurricane developmentusingacoupledoceanatmospheremodel. J.Atmos. Sci., 57, 481492. Czajkowski,J.,K.Simmons,andD.Sutter,2011:Ananalysisof coastalandinlandfatalitiesinlandfallingUShurricanes. Nat. Hazards, 59, 15131531. Dai,A.,2001:Globalprecipitationandthunderstormfrequencies. PartII:Diurnalvariations. J.Climate, 14, 11121128. ,X.Lin,andK.L.Hsu,2007:Thefrequency,intensity,and diurnalcycleofprecipitationinsurfaceandsatelliteobservations overlow-andmid-latitudes. ClimateDyn., 29, 727744. DeMaria,M.,andJ.Kaplan,1994:AStatisticalHurricaneIntensity PredictionScheme(SHIPS)fortheAtlanticbasin. Wea. Forecasting, 9, 209220. ,M.Mainelli,L.K.Shay,J.A.Knaff,andJ.Kaplan,2005: FurtherimprovementstotheStatisticalHurricaneIntensity PredictionScheme(SHIPS). Wea.Forecasting, 20, 531543. Emanuel,K.,J.Callaghan,andP.Otto,2008:Ahypothesisforthe redevelopmentofwarm-corecyclonesovernorthernAustralia. Mon.Wea.Rev., 136, 38633872. Frank,W.M.,1977:Thestructureandenergeticsofthetropical cycloneI.Stormstructure. Mon.Wea.Rev., 105, 11191135. ,andE.A.Ritchie,1999:Effectsofenvironmentalowupon tropicalcyclonestructure. Mon.Wea.Rev., 127, 20442061. Gray,W.M.,andR.W.J.Jacobson,1977:Diurnalvariationof deepcumulusconvection. Mon.Wea.Rev., 105, 11711188. Haggard,W.H.,T.H.Bilton,andH.L.Crutcher,1973:Maximum rainfallfromtropicalcyclonesystemswhichcrosstheAppalachians. J.Appl.Meteor.,12, 5061. Harr,P.A.,andR.L.Elsberry,2000:Extratropicaltransitionof tropicalcyclonesoverthewesternNorthPacic.PartI:Evolutionofstructuralcharacteristicsduringthetransitionprocess. Mon.Wea.Rev., 128, 26132633. Hart,R.E.,andJ.L.Evans,2001:Ac limatologyoftheextratropical transitionofAtlantictropicalcyclones. J.Climate, 14, 546564. Hill,K.,andG.M.Lackmann,2009:Inuenceofenvironmental humidityontropicalcyclonesize. Mon.Wea.Rev., 137, 3294 3315. Houze,R.A.,1993: CloudDynamics. AcademicPress,573pp. Jarvinen,B.R.,C.J.Neumann,andM.A.S.Davis,1984:A tropicalcyclonedatatapefortheNorthAtlanticbasin,1886 1983.NOAATech.Memo.NWSNHC22,21pp.[Available onlineathttp://www.nhc.noaa.go v/pdf/NWS-NHC1988-22.pdf.] MAY2013MATYAS1095

PAGE 12

Jiang,H.,J.B.Halverson,J.Simpson,andE.J.Zipser,2008:On thedifferencesinstormrainfallfromHurricanesIsidoreand Lili.PartII:Waterbudget. Wea.Forecasting, 23, 4461. ,C.Liu,andE.J.Zipser,2011:ATRMM-basedtropicalcyclonecloudandprecipitationfeaturedatabase. J.Appl.Meteor.Climatol., 50, 12551274. Jones,S.C.,andCoauthors,2003:Theextratropicaltransitionof tropicalcyclones:Forecastchallenges,currentunderstanding, andfuturedirections. Wea.Forecasting, 18, 10521092. Jorgensen,D.P.,1984:Mesoscaleandconvective-scalecharacteristicsofmaturehurricanes.PartI:Generalobservationsby researchaircraft. J.Atmos.Sci., 41, 12681285. Kellner,O.,D.Niyogi,M.Lei,andA.Kumar,2012:Theroleof anomaloussoilmoistureontheinlandreintensicationof TropicalStormErin(2007). Nat.Hazards, 63, 15731600. Kikuchi,K.,andB.Wang,2008:Diurnalprecipitationregimesin theglobaltropics. J.Climate, 21, 26802696. Kimball,S.K.,2008:Structureandevolutionofrainfallinnumericallysimulatedlandfallinghurricanes. Mon.Wea.Rev., 136, 38223847. Klein,P.M.,P.A.Harr,andR.L.Elsberry,2000:Extratropical transitionofwesternNorthPacictropicalcyclones:An overviewandconceptualmodelofthetransformationstage. Wea.Forecasting, 15, 373395. Konrad,C.E.,2001:Diurnalvariationsinthelandfalltimesof tropicalcyclonesovertheeasternUnitedStates. Mon.Wea. Rev., 129, 26272631. ,M.F.Meaux,andD.A.Meaux,2002:Relationshipsbetween tropicalcycloneattributesandprecipitationtotals:Considerationsofscale. Int.J.Climatol., 22, 237247. Kossin,J.P.,2002:DailyhurricanevariabilityinferredfromGOES infraredimagery. Mon.Wea.Rev., 130, 22602270. Lajoie,F.A.,andI.J.Butterworth,1984:Oscillationofhigh-level cirrusandheavyprecipitationaroundAustralianregion tropicalcyclones. Mon.Wea.Rev., 112, 535544. Liu,K.S.,J.C.L.Chan,W.C.Cheng,S.L.Tai,andP.W.Wong, 2007:DistributionofconvectionassociatedwithtropicalcyclonesmakinglandfallalongtheSouthChinacoast. Meteor. Atmos.Phys., 97, 5768. Lonfat,M.,R.Rogers,T.Marchok,andF.D.Marks,2007:A parametricmodelforpredictinghurricanerainfall. Mon.Wea. Rev., 135, 30863097. Mann,H.B.,andD.R.Whitney,1947:Onatestofwhetheroneof tworandomvariablesisstochasticallylargerthantheother. Ann.Math.Stat., 18, 5060. Matyas,C.J.,2007:QuantifyingtheshapesofU.S.landfalling tropicalcyclonerainshields. Prof.Geogr., 59, 158172. ,2010:Associationsbetweenthesizeofhurricaneraineldsat landfallandtheirsurroundingenvironments. Meteor.Atmos. Phys., 106, 135148. Maxwell,J.T.,P.T.Soule,J.T.Ortegren,andP.A.Knapp,2012: Drought-bustingtropicalcyclonesinthesoutheasternAtlanticUnitedStates:19502008. Ann.Assoc.Amer.Geogr., 102, 259275. Medlin,J.M.,S.K.Kimball,andK.G.Blackwell,2007:Radarand raingaugeanalysisoftheextremerainfallduringHurricane Danny's(1997)landfall.Mon.Wea.Rev., 135, 18691888. Mesinger,F.,andCoauthors,2006:NorthAmericanRegional Reanalysis. Bull.Amer.Meteor.Soc., 87, 343360. Muramatsu,T.,1983:Diurnal-variationsofsatellite-measured TBBarealdistributionandeyediameterofmaturetyphoons. J.Meteor.Soc.Japan, 61, 7790. NHC,cited2012:NHCissuancecriteriachangesfortropicalcyclonewatches/warnings.[Availableonlineathttp://www.nhc. noaa.gov/watchwarn_changes.shtml.] OfceoftheFederalCoordinator forMeteorologicalServices andSupportingResearch(OFCM),2006:Federalmeteorologicalhandbook11:Dopplerradarmeteorological observationsPartD,WSR-88DUnitDescriptionand OperationalApplications.FCM-H11D-2006,218pp.[Availableonlineathttp://www.ofcm.gov/fmh11/fmh11partd/pdf/ FMH11D-2006.pdf.] Parrish,J.R.,R.W.Burpee,F.D.Marks,andR.Grebe,1982: Rainfallpatternsobservedbydigitizedradarduringthe landfallofHurricaneFrederic(1979). Mon.Wea.Rev., 110, 19331944. Rappaport,E.N.,2000:LossoflifeintheUnitedStatesassociated withrecentAtlantictropicalcyclones. Bull.Amer.Meteor. Soc., 81, 20652073. ,andCoauthors,2009:AdvancesandchallengesattheNationalHurricaneCenter. Wea.Forecasting, 24, 395419. Ritchie,E.A.,andR.L.Elsberry,2001:Simulationsofthetransformationstageoftheextratropicaltransitionoftropicalcyclones. Mon.Wea.Rev., 129, 14621480. Rodgers,E.B.,S.W.Chang,J.Stout,J.Steranka,andJ.J.Shi, 1991:Satellite-observationsofvariationsintropicalcyclone convectioncausedbyupper-tropospherictroughs. J.Appl. Meteor., 30, 11631184. Rogers,R.F.,andR.E.Davis,1993:TheeffectofcoastlinecurvatureontheweakeningofAtlantictropicalcyclones. Int. J.Climatol., 13, 287299. Sheets,R.C.,1990:TheNationalHurricaneCenterPast,present, andfuture. Wea.Forecasting, 5, 185232. Steiner,M.,R.A.Houze,andS.E.Yuter,1995:Climatological characterizationofthree-dimensionalstormstructurefrom operationalradarandrain-gaugedata. J.Appl.Meteor., 34, 19782007. Steranka,J.,E.B.Rodgers,andR.C.Gentry,1984:Thediurnal variationofAtlanticOceantropicalcycloneclouddistribution inferredfromgeostationarysatelliteinfraredmeasurements. Mon.Wea.Rev., 112, 23382344. Sturdevant-Rees,P.,J.A.Smith,J.Morrison,andM.L.Baeck, 2001:Tropicalstormsandtheoodhydrologyofthecentral Appalachians. WaterResour.Res., 37, 21432168. Tokay,A.,andD.A.Short,1996:Evidencefromtropicalraindrop spectraoftheoriginofrainfromstratiformversusconvective clouds. J.Appl.Meteor., 35, 355371. Tuleya,R.E.,1994:Tropicalstormdevelopmentanddecay:Sensitivitytosurfaceboundaryconditions. Mon.Wea.Rev., 122, 291304. Vincent,D.G.,andA.H.Fink,2001:TropicalcycloneenvironmentsoverthenortheasternandnorthwesternPacicbased onERA-15analyses. Mon.Wea.Rev., 129, 19281948. Wallace,J.M.,1975:DiurnalvariationsinprecipitationandthunderstormfrequencyovertheconterminousUnitedStates. Mon.Wea.Rev., 103, 406419. Wilks,D.S.,1995: StatisticalMethodsintheAtmosphericSciences:AnIntroduction. AcademicPress,467pp. Yang,S.,andE.A.Smith,2006:Mechanismsfordiurnalvariability ofglobaltropicalrainfallobservedfromTRMM. J.Climate, 19, 51905226. Yokoyama,C.,andY.N.Takayabu,2008:Astatisticalstudyon raincharacteristicsoftropicalcyclonesusingTRMMsatellite data. Mon.Wea.Rev., 136, 38483862.1096JOURNALOFAPPLIEDMETEOROLOGYANDCLIMATOLOGYVOLUME52