<%BANNER%>

UFIR



Associations between the size of hurricane rain fields at landfall and their surrounding environments
CITATION PDF VIEWER
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/IR00001203/00001
 Material Information
Title: Associations between the size of hurricane rain fields at landfall and their surrounding environments
Series Title: Matyas, C. J., 2010: Associations between the size of hurricane rain fields at landfall and their surrounding environments. Meteorology and Atmospheric Physics, 106, 135-148.
Physical Description: Journal Article
Creator: Matyas, Corene
Publisher: Meteorology and Atmospheric Physics
Publication Date: 2010
 Notes
Abstract: This study examines relationships between the extent of hurricane rain fields, storm size, and the environment surrounding the storm. A Geographic Information System is employed to measure the extent of the rain fields in each quadrant of 31 hurricanes at landfall-time. After correlating the extents with measures of storm size, multiple linear regression models are developed to determine which atmospheric forcing(s) at 0, 12, and 24 h prior to landfall are most highly related to rain field size in each quadrant. Results show that the radius of the outermost closed isobar encompasses the rain fields in 90% of the observations. Strong vertical wind shear from the southwest correlates with a larger (smaller) rain field extent toward the northeast (southwest), while higher relative humidity values correlate with a larger extent toward the northwest, southwest, and southeast. Storm intensity and location also exhibit statistically significant correlations with rain field size.
Acquisition: Collected for University of Florida's Institutional Repository by the UFIR Self-Submittal tool. Submitted by Corene Matyas.
 Record Information
Source Institution: University of Florida Institutional Repository
Holding Location: University of Florida
Rights Management: All rights reserved by the submitter.
System ID: IR00001203:00001

Downloads

This item is only available as the following downloads:

7Matyas2010MAP ( PDF )


Full Text

PAGE 1

ORIGINALPAPERAssociationsbetweenthesizeofhurricanerainelds atlandfallandtheirsurroundingenvironmentsCoreneJ.MatyasReceived:21July2009/Accepted:22December2009/Publishedonline:20January2010 Springer-Verlag2010Abstract Thisstudyexaminesrelationshipsbetweenthe extentofhurricanerainelds,stormsize,andtheenvironmentsurroundingthestorm.AGeographicInformation Systemisemployedtomeasuretheextentoftherainelds ineachquadrantof31hurricanesatlandfall-time.After correlatingtheextentswithmeasuresofstormsize,multiplelinearregressionmodelsaredevelopedtodetermine whichatmosphericforcing(s)at0,12,and24hpriorto landfallaremosthighlyrelatedtoraineldsizeineach quadrant.Resultsshowthattheradiusoftheoutermost closedisobarencompassestheraineldsin90%ofthe observations.Strongverticalwindshearfromthesouthwestcorrelateswithalarger(smaller)raineldextent towardthenortheast(southwest),whilehigherrelative humidityvaluescorrelatewithalargerextenttowardthe northwest,southwest,andsoutheast.Stormintensityand locationalsoexhibitstatisticallysignicantcorrelations withraineldsize. 1Introduction Manystudieshaveshownthatconsiderablespatialvariabilityexistsintherainfallproducedbytropicalcyclones (TCs).Highrainfalltotalscanoccurbothclosetoand hundredsofkilometersfromthecenterofcirculationand onbothsidesofthestormtrack.Forexample,highrainfall totalsoccurrednearthecirculationcenterontheleftsideof HurricaneDanny(1997)duetoaslowforwardvelocity,a contractionoftheeyewall,andanomalouslywarmwaters inMobileBay(Blackwell 2000 ;Medlinetal. 2007 ).As HurricaneFloyd(1999)transitionedintoanextratropical cyclone,itproducedheavyrainfallfartherawayfromthe circulationcenterbutalsoontheleftsideofthetrack (Atallahetal. 2007 ;AtallahandBosart 2003 ).Several recentobservationalstudieshaveinvestigatedthelocations withinTCswhereconvectiveprecipitationdevelopsin responsetochangingenvironmentalconditions(Cecil 2007 ;Chanetal. 2004 ;Chenetal. 2006 ;Corbosieroand Molinari 2002 ;CorbosieroandMolinari 2003 ;Liuetal. 2007 ).However,tobetterpredictwhichlocationsareat riskofreceivingheavyrainfallasaTCapproaches,itis importanttounderstandhowtheenvironmentsurrounding aTCinuencestheradialextentoftheraineldsas landfalloccurs. ThesizesofTCsaremeasuredwithrespecttotheradius ofaparticularwindspeed(i.e.,radiusofgale-forcewinds) (Demuthetal. 2006 ;KimballandMulekar 2004 ;Moyer etal. 2007 )ortheradiusoftheoutermostclosedisobar (KimballandMulekar 2004 ;Merrill 1984 ).AsaTCmakes landfall,interactionwiththelandsurfacecausesthewind eldstodecreaseinsize(e.g.,KaplanandDeMaria 1995 ) whiletheextentofrainfallmayincrease(e.g.,Kimball 2008 )sothattheedgesofthewindandraineldsdonot align.Thus,thedistancefromthecirculationcenterover whichtheraineldsextendcouldalsobeconsidereda measureofstormsizeasitrelatestopost-landfallrainfall potential.Asmoredeathshaveresultedfromfreshwater oodingtriggeredbyheavyrainfallfromTCsintheUSin recentyearsascomparedtowind-relateddamage(Rappaport 2000 ),itcanbearguedthatthesizeofaTC'srain eldsisanimportantattributeofthestorm.Knowledgeof theraineldextentfacilitatesboththeidenticationof areasthatcouldpotentiallyreceiveheavyrainfallandthe calculationofthedurationoftherainfallevent.Boththese C.J.Matyas( & ) DepartmentofGeography,UniversityofFlorida, 3141TurlingtonHall,Gainesville,FL32611,USA e-mail:matyas@u.edu123MeteorolAtmosPhys(2010)106:135148 DOI10.1007/s00703-009-0056-1

PAGE 2

parametersmustbeconsideredwhendeterminingspecic areasthatmayexperienceoodingfromtherainfallproducedbyaTC(Elsberry 2002 ;Kidderetal. 2005 ;Liuetal. 2008 ;VieuxandBedient 1998 ). Tofullyexplorethespatialextentofconvectiverainfall associatedwithlandfallTCs,thisstudyutilizesaGeographicInformationSystem(GIS)tomeasuretheextentof theraineldsasindicatedbyradarreectivityreturns.The distancefromthecirculationcentertotheedgeoftherain eldineachquadrantofthestormiscalculatedasmany previousstudies(e.g.,Blackwell 2000 ;Chanetal. 2004 ; Marks 1985 )havenotedtheasymmetricaldistributionof convectionwithinTCsatthetimeoflandfall.Correlation coefcientsarecalculatedbetweenthesedistancesanddata representingthesizesofthewindeldstodeterminehow closelyrelatedtheextentoftheraineldsaretothewind elds.Themeasureofstormsizethatmostoftenencompassestherainelds,andthatwhichismostoftencolocatedwiththeedgeoftheraineldineachquadrant,is determined.Finally,multiplelinearregressionmodelsare constructedtoidentifythevariablesrepresentingtheconditionsoftheenvironmentsurroundingthehurricanethat arethemosthighlycorrelatedtotheextentoftheraineld ineachquadrant.Separateanalysesareperformedutilizing predictorsfromthetimeoflandfall(t0),and12(t12)and 24(t24)hourspriortolandfall.Theresultsoftheregressionanalysessuggestthatverticalwindshearandatmosphericmoistureexertthemostinuenceonthesizeofthe raineldsatthetimeoflandfall. 2Data AsthisstudyexaminesTCsatlandfall,thersttaskisto establishthetimeandlocationofthelandfallofthe circulationcenterforeachstorm.Thisisaccomplished usingtheHurricaneSeasonTropicalCycloneReports availableforeachstormontheNationalHurricaneCenter (NHC)'swebsite( http://www.nhc.noaa.gov/pastall.shtml ). ThecoordinatesofthecirculationcenterofeachTCatthe timeoflandfallprovidetheinformationnecessarytoobtain radarreectivitydatafromtheradarsitesnearesttothe pointlandfall,andalsoserveastheoriginformeasurementsoftheradiusoftherainelds. Initially,53landfallsofhurricanes,tropicalstorms,and tropicaldepressionswereanalyzed.However,thepredictors ofraineldsize,suchasverticalwindshearandmaximum sustainedwindspeed,didnotfollowanormalorGaussian distributionwhenall53landfallswereexaminedcollectively.Additionally,manyofthetropicalstormsand depressionshadoneormorequadrantsthatwerevoidof convectiverainfallsothattheraineldsizesdidnotfollowa normaldistribution.Removalofthetropicalstormsand depressionsallowedtheindependentanddependentvariablestoassumenormaldistributionssothattheycouldbe includedinthestatisticalanalyses.Thus,thisstudyonly includesresultsfromtheanalysisofTCsthatwerehurricanes atthetimeoflandfall.Fiveofthe26hurricanesexamined makemultiplelandfallswithintheUSathurricaneintensity. Whenasecondlandfalloccursmorethan18haftertherst, bothlandfallsareutilized.Thisyields31hurricanelandfalls thatareexaminedinthecurrentstudy(Table 1 ). Radarreectivitydataareutilizedtodenetheedgeof theraineldforeachhurricane.AsdataformostWeather SurveillanceRadar1988Doppler(WSR-88D)sitesare availablebeginningin1995,thisyearmarksthebeginning ofthestudyperiod.HurricanesOpal(1995)andLili(2002) Table1 Hurricanesanalyzedinthecurrentstudylistedinorderof decreasingmaximumsustainedwindspeed( Vmax)atthetimeof landfall HurricaneYearLandfall timeanddate Landfall state Vmax(ms1) Charley2004-12045UTC13AugFL64 Katrina2005-21110UTC29AugLA57 Ivan20040650UTC16SepAL54 Jeanne20040400UTC26SepFL54 Dennis20051930UTC10JulFL54 Wilma20051030UTC24OctFL54 Rita20050740UTC24SepLA51 Bret19990000UTC23AugTX51 Fran19960030UTC6SepNC51 Bonnie19980400UTC27AugNC49 Ike20080700UTC13SepTX49 Bertha19962000UTC12JulNC46 Georges1998-11530UTC25SepMS46 Georges1998-21130UTC28SepFL46 Floyd19990630UTC16SepNC46 Isabel20031700UTC18SepNC46 Frances20040430UTC5SepFL46 Gustav20081500UTC1SepLA46 Claudette20031530UTC15JulTX41 Humberto20070700UTC13SepTX41 Erin1995-10615UTC2AugFL39 Erin1995-21600UTC3AugFL39 Dolly20081830UTC23JulTX39 Danny1997-21000UTC19JulAL36 Earl19980600UTC3SepFL36 Irene19992000UTC15OctFL36 Charley2004-21400UTC14AugSC36 Katrina2005-12230UTC25AugFL36 Danny1997-10900UTC18JulLA33 Gaston20041400UTC29AugSC33 Cindy20050300UTC6JulLA33 136 C.J.Matyas123

PAGE 3

couldnotbeincludedastherequiredradardatawerenot available.LevelIIIbasereectivitydatafromtheWSR88Dsiteslocatednearesttothepointoflandfallandfrom thescanoccurringatthetimenearesttothatoflandfallare obtainedfromtheNationalClimaticDataCenter's (NCDC)archive( http://www.ncdc.noaa.gov/nexradinv/ ). Thesedataarederivedfromthelowestscanoftheradar (0.5 tilt)andareroundedtothenearest5dB Z (OFCM 2006).AtsomeWSR-88DsitessuchasKEVXandKEOX, LevelIIIdataarenotavailablepriorto2001.Inthese cases,LevelIIdataareutilized.LevelIIdataincludeall tiltsoftheradar,andreectivityvaluesarerecordedin0.5 dB Z increments(OFCM2006).Forthisstudy,reectivity valuesfromthe0.5 tiltareextractedfromtheLevelIIdata andareroundedtothenearest5dBZsothattheyfollow thesamenumericalscaleastheLevelIIIdata. VariablesthatcharacterizetheconditionsoftheatmospherearoundthestormsareobtainedfromtheStatistical HurricaneIntensityScheme(SHIPS)database(DeMaria andKaplan 1994 1999 ;DeMariaetal. 2005 ).Theatmosphericenvironmentalfactorsinthisdatabaseareall derivedfromtheNCEPGFSmodelanalyses.Dataare availableevery6h(00,06,12,and18UTC),andmost variablesarecalculatedforanannularregionlocated200 800kmfromthecirculationcenter.Forthecurrentstudy, thesedataareinterpolatedtot0,t12,andt24byalinear function.Thetemperatureandrelativehumidityofthe environmentsurroundingeachstormareavailablenearthe surfaceat1,000hPaaswellasforthreeadditionallevelsor layersabovetheboundarylayer(Table 2 ).Themagnitude anddirectionofverticalwindshearoveradeeplayeris calculatedover850200hPa;thedirectionofshearis subdividedintosouth-to-northandwest-to-eastcomponentsandmultipliedbythevelocityofthedeeplayerwind shearforthecurrentstudy.Thevelocityofthe 850200hPashearisalsoincludedintheanalysis,asare divergenceandthemagnitudeofthezonalwindat200hPa (Table 2 ).OthervariablescontainedwithintheSHIPS database,suchasvorticityat850hPa,couldnot beemployedastheirvaluesdidnotfollowanormal distribution. Stormmotion,intensity,andlocationarealsolikelyto inuencetheradiusofTCrainelds.Stormmotionis calculatedusingthecoordinatesofthecirculationcenters obtainedfromthehurricanebesttrackdatabase(HURDAT),whichcontainsobservationsevery6h(00,06,12, and18UTC)(Jarvinenetal. 1984 ).Thesedataarelinearly interpolatedtot0,t12,andt24,andstormmotioniscalculatedusingthechangeinthepositionofthecirculation center.Stormheadingissubdividedintosouth-to-northand west-to-eastcomponentsandmultipliedbytheforward velocity.Thelatitudeofthestormatthetimeoflandfall andmaximumsustainedwindspeedareobtainedfromthe HurricaneSeasonTropicalCycloneReports.Previous researchhasshownthathurricanesgrowinsizeafter reachingtheirpeakintensity(KimballandMulekar 2004 ; Maclayetal. 2008 ;Merrill 1984 1988 ),meaningthat hurricanesmakinglandfallseveralhoursafterattaining peakintensitymayhavelargerraineldsthanstorms attainingpeakintensitynearthetimeoflandfall.Thus,the latitudeofthehurricaneatthetimeofpeakintensity,which isalsoavailablefromtheseasonreports,isincludedinthe analysis(Table 2 ).Thelongitudes,bothofthestormsatt0 andatthetimeofpeakintensity,couldnotbeincludedas theydonotfollowanormaldistribution.However,longitudevaluesatt12andt24aredistributednormallyandare includedaspredictorsintheregressionmodelsthatemploy datafromthesetimes. Datapertainingtothesizeofeachhurricaneare obtainedfromtheextendedbesttrackdataset(Demuth etal. 2006 ).Thecurrentstudyutilizesfourmeasuresof stormsizefromthisdataset,includingtheradiusofthe outermostclosedisobar(ROCI),andforeachquadrantof thestorm,theextentof17ms1(gale-force),26ms1 Table2 Variablesthatarenormallydistributedandutilizedat0,12, and24hpriortolandfalltopredicttheradiusoftheraineldat landfall-time VariableAbbreviationUnits CirculationcenterlatitudeLat CirculationcenterlongitudeaLon Maximumsustainedwindspeed Vmaxms1South-to-northstormmotionDirNms1West-to-eaststormmotionDirEms1Deep-layer(850200hPa)shear south-to-north ShDNms1Deep-layer(850200hPa)shear west-to-east ShDEms1Velocityofthedeep-layershearShDVms1Zonalwindsat200hPaU200ms1Relativehumiditylow-level (850700hPa) RhLo% Relativehumiditymid-level (700500hPa) RhMd% Relativehumidityhigh-level (500300hPa) RhHi% Near-surfacerelativehumidity (1,000hPa) R000% Temperatureat150hPaT150 C Temperatureat200hPaT200 C Temperatureat250hPaT250 C Near-surfacetemperature (1,000hPa) T000 C Divergenceat200hPaD20010^5s1LatitudeattimeofpeakintensityLatM aOnlydistributednormallyandutilizedasapredictoratt12andt24 Associationsbetweenthesizeofhurricanerainelds 137123

PAGE 4

(damaging-force),and33ms1(hurricane-force)winds. Thequadrantsutilizedtoprovidesizemeasuresarelocated northeast,northwest,southwest,andsoutheastofthecirculationcenterofthestorm.Thesizedataareestimated fromshipreports,aircraftreconnaissancedata,andsatellite imageryandarealsolinearlyinterpolatedtothetime oflandfallastheoriginaldataareavailableevery6h (Muelleretal. 2006 ). 3Radaranalysisandquadrantplacement Theradiusoftheraineldineachquadrantofthehurricanesisdeterminedthroughaspatialanalysisofradar reectivitydatawithinaGIS.Radarreectivitydatain theirnativeformatcannotbedirectlyimportedintoaGIS. However,capabilitiesexisttoconvertthedataintoaGIScompatibleformatthroughtheJavaNext-Generation WeatherRadar(NEXRAD)ToolsdevelopedbyAnsariand DelGreco( 2005 ),whichareavailablefromtheNCDC website.Theconvertedradardataarethenimportedinto ArcGIS9.2(ESRI2006)asageoreferencedshapele, alongwiththecoordinatesofthecirculationcentersofthe hurricanesatthetimeoflandfall.WithintheGIS,data fromadjoiningradarsitesaremergedintoasinglelayerso thattheentireraineldcanbeanalyzed.Inlocationswhere reectivityvaluesareavailablefrommultipleradarstations,thehighestvaluesareretained. Thenexttaskistodenetheedgeoftheraineldfor eachhurricane.Spatialattributessuchasthelengthofthe perimeterarecalculatedforclustersofreectivityreturns havingthesamevalue.Toqualifyastheedgeoftherain eld,aclusterofreectivityvaluesmusthaveaperimeter lengthgreaterthan100kmandcontainreectivityvalues equaltoorgreaterthan35dB Z .TheGISisutilizedto measurethedistancefromthecirculationcentertotheedge oftheclusters,andthedistancetothefarthestclusterin eachquadrantofthestormservesastheradiusoftherain eldforthatquadrant(Fig. 1 ).Themeanradiusoftherain eldiscalculatedbyaveragingthemeasurementsfromthe fourquadrants.Subtractingtheradiusofthequadrantwith thesmallestspanfromthequadrantwiththelargestspan yieldsasimplemeasureofstormasymmetry. AsTCscanbedividedintoquadrantsbyusingcardinal directions,stormmotion,andthedirectionofverticalwind shear,itisimportanttounderstandhowthesequadrants tendtooverlapinspacetocorrectlyinterprettheresultsof thestatisticalanalyses.Datapertainingtothewindeldsof TCsarecollectedduringaircraftreconnaissanceightsin thenortheast,northwest,southwest,andsoutheastquadrantsofthestorm(Muelleretal. 2006 ).Thesedataare utilizedtodeterminetheextentofdamaging-forcewinds andthesizeofthestorm.Toallowtheextentsoftherain eldsandwindeldstobedirectlycompared,thehurricanesinthisstudyaredividedintoquadrantsbasedupon cardinaldirectionssothatthenortheastquadrantspans0 90 ,thesoutheastquadrantspans90 180 ,thesouthwest quadrantspans180 270 andthenorthwestquadrant spans270 360 .Theradiusoftheraineldisthen measuredineachquadrant. Previousresearchhasshownthatstormmotionand verticalwindshearcanintroduceasymmetriesintostorm structurethataffectwhereconvectiverainfalldevelopsin boththeinner100kmofthestorm,andatlargerradii(e.g., Chenetal. 2006 ;CorbosieroandMolinari 2003 ;Rogers etal. 2003 ).Stormmotioncausesheavierrainfalleither aheadofthestorm(rightandleftfrontquadrants)(Miller 1958 ;Shapiro 1983 )ortotherightofthestorm(rightfront andrearquadrants)(Frank 1977 ).Theleftrearquadrant typicallycontainstheleastconvectionatthetimeof landfallasoffshorewindsatthesurfacedivergeand enhancedownwardmotionthatsuppressesconvection (Frank 1977 ;Powell 1987 ).Aspreviouslydescribed,storm motionatthetimeoflandfalliscalculatedusingthe coordinatesofthecirculationcentercontainedwithin HURDAT,andtheradiusoftheraineldismeasuredfor eachmotion-relativequadrant. Whenverticalwindshearispresent,thevortexofaTC becomestilted.Thistiltingenhancesupwardsmotionand increasesrainfallinthedirectionoftilt,anddownward motionisenhancedinthedirectionoppositetothetilt (CorbosieroandMolinari2002 ;FrankandRitchie 2001 ; Zhengetal. 2007 ).Althoughverticalwindsheartendsto Fig.1 Measuringtheradiusoftheraineldineachquadrantof HurricaneBret(1999)atthetimeoflandfallwithquadrantsplaced relativetocardinaldirections 138 C.J.Matyas123

PAGE 5

enhancerainfallinthedownshear-leftquadrantnearthe coreofahurricane,itcanenhancerainfallinboththe downshear-rightanddownshear-leftquadrantsintheouter regionsofthestorm(CorbosieroandMolinari 2003 ). Becauseverticalwindshearmaybeacrossoralongthe stormtrack(CorbosieroandMolinari 2003 ;Rogersetal. 2003 ),itispossiblefortheright-frontquadranttooverlap withanyofthefourshear-relativequadrants(downshearleft,downshear-right,upshear-right,upshear-left).Thus, quadrantsarealsoplacedaccordingtothedirectionofthe deeplayer(850200hPa)verticalwindshearobtained fromtheSHIPSdataset,andtheradiusoftheraineldis measuredineachoftheseshear-relativequadrants.It shouldbenotedthatuseofquadrantsbaseduponcardinal directionsalsofacilitatesthecomparisonoftheraineld extentwiththedirectionoftheverticalwindshear.Many TCsintheAtlanticbasinexperienceverticalwindshear fromthewestorsouthwest(CorbosieroandMolinari 2003 ),whichplacesthedownshearleftorrightquadrants tothenortheastofthecirculationcenter. Themajorityofthehurricanesexaminedinthecurrent studyaremovingnorthward(Fig. 2 )astheymakelandfall alongtheUSGulfCoastorthecoastofNorthCarolina. Thismotionresultsinanoverlapoftherightfrontand northeastquadrantsin58%ofthestorms(Fig. 3 ).The verticalwindshearexperiencedbymostofthehurricanes hasstrongwesterlyandsoutherlycomponentstoits direction(Fig. 4 ).Thenortheastquadrantismostoften inthesamelocationasthedownshear-leftquadrant,andis inthedownsheardirectionin74%ofthecases(Fig. 5 ).In 68%ofcases,theright-frontquadrantislocatedwithinone ofthetwodownshearquadrants(Fig. 6 ).Whenconsidering allthreesetsofquadrants,themajorityofhurricanesfeatureoverlapamongthenortheast,right-front,anddownshear-leftquadrants(Table 3 ).Theresultsofpaired-sample t testsconrmthattheradiioftheraineldswithinthe northeastandright-frontquadrantsandnortheastand downshear-leftquadrantsaresimilaramongthe31casesin thecurrentstudyastheyarepositivelycorrelatedandhave similarmeans.Thequadrantplacementsarediscussedin termsoftheircardinaldirectionsintherestofthepaper, whichfacilitatescomparisonofthewindelds,aswellas thedirectionoftheverticalwindshear,withtheextentof therainelds. 4Rainelddistances Theaverageraineldsizeis223kmatthetimeoflandfall. However,theasymmetrymetricrevealstheextenttowhich theraineldsofhurricanesareasymmetricallydistributed Fig.2 Headingandforwardvelocityatthetimeoflandfallforthe31 hurricanesinthecurrentstudy;radialsareplacedevery2ms1out to12ms1 Fig.3 Distributionofthedirectionofstormheadingandthenumber ofcaseswhentherightfrontquadrant(RF)fallswithineach direction-relativequadrant(NE,SE,SW,NW) Fig.4 Directionandvelocityof850200hPaverticalwindshearat thetimeoflandfallforthe31hurricanesinthecurrentstudy;radials areplacedevery5ms1outto25ms1 Associationsbetweenthesizeofhurricanerainelds 139123

PAGE 6

aboutthecirculationcenterofthestorm.Onaverage,the raineldextends204kmfartherfromthecirculation centerinthequadrantwiththelargestradiusthanitdoesin thequadrantwiththesmallestradius.Afterdividingthe casesintothreegroupsbasedonwhenandifastorm becomesextratropical(Fig. 7 ),itcanbeseenthatthemost symmetricalstorms,withanaverageasymmetryof 126km,dissipateafterlandfallwithoutbecomingextratropical.Forthesestorms,raineldsspanlargerdistances onthesouthsideandsmallerdistancesonthenorthsideof thestormwhencomparedtostormsthatdobecome extratropical. Themostasymmetrichurricanes,withanaverage asymmetryof277km,arewithin2daysofcompletingan extratropicaltransition(Fig. 7 ).HurricaneFloyd(1999)is themostasymmetricalhurricaneinthestudy.Assuming thatitssizeremainedunchangedpriortolandfall,therain eldofFloyd(1999)couldhavemadelandfallsome30h priortothecirculationcentergivenits8.7ms1forward velocityand595kmradiusoftheraineldinaforward quadrant.Thus,itisimportanttoaccountforasymmetries intheradiusoftheraineld,ratherthanrelyingona quadrant-averagedmeasureofraineldsize,tomore accuratelypredictwhenheavyrainfallwillbegintoaffect thecoastline. Hurricanetroughinteractionsthatresultinanextratropicaltransition(ET)likelyexplainwhyraineldtends Fig.5 Distributionofthedirectionoftheheadingofthewindshear andthenumberofcaseswhenthedownshearleftquadrant(DSL) fallswithineachdirection-relativequadrant(NE,NW,SE,SW) Fig.6 Distributionoftheanglebetweentheheadingofthevertical windshearandstormmotionmeasuredcounterclockwisefromthe verticalsheartothemotionvector,andthenumberofcaseswhenthe downshearleftquadrant(DSL)fallswithineachmotion-relative quadrant(RF,RR,LR,LF) Table3 Numberofcaseswheneachmotion-andshear-relative quadrantoverlapswiththenortheastquadrant QuadrantoverlapNo.ofcases NE-RF-DSL10 NE-RF-DSR5 NE-RF-USR1 NE-RF-USL2 NE-RR-DSL4 NE-RR-DSR1 NE-RR-USL4 NE-LF-DSL1 NE-LF-DSR2 NE-LF-USL1 Fig.7 Extentofraineldineachquadrantaveragedforhurricanes thatcompleteanextratropicaltransition( ET )within2daysof landfall,completean ET morethan2daysafterlandfall,anddonot completean ET ;radialsareplacedevery50400km 140 C.J.Matyas123

PAGE 7

tobelargeronthenorthandsmalleronthesouthsidesof thestormsinthecurrentstudy(Fig. 8 ).Over70%ofthe hurricanesinthecurrentstudybecomeextratropicalafter landfall.Thus,thisstudycontainsahigherpercentageof thesecasesthanthatfoundbyHartandEvans( 2001 ),who determinedthat46%ofAtlanticbasinTCsduring1899 1996becameextratropical.Duringthistransition,the environmentsurroundingthehurricanebecomesbaroclinic andisentropicascentoccursnearthesteepthermalgradient betweentropicalandcontinentalairmassestypically locatednorthofthestorm'scenterthatenhancesprecipitationinthisregion(Kleinetal. 2000 ;Pengetal. 2007 ) Cooleranddrierairmassesadvectintothehurricane's circulationonthesouthsideofthestorm,whichreduces convectioninthisdirection(AtallahandBosart 2003 ). Thelargestquadrantonaverageisthenortheastquadrant(Fig. 8 ).Twentyofthe31hurricaneshavetheirlargest rain-eldspaninthenortheastquadrant,andin17ofthese 20hurricanes,thisquadrantisinthedownsheardirection. AreviewofNHCTropicalCycloneSeasonReportsconrmsthatstrongverticalwindshearfromthesouthwest commonlyaffectshurricanesastheymakelandfallinthe US.Verticalwindshearenhancesconvectioninthe downsheardirection(Blacketal. 2002 ;Chenetal. 2006 ; CorbosieroandMolinari 2002 2003 ;Rogersetal. 2003 ; Willoughbyetal. 1984 ),andMaclayetal.( 2008 )found thathighverticalwindshearcausesthedisplacementof convectionintotheouterregionsofTCs.Thismostlikely explainswhytheraineldextendsmorethan400kmin thenortheastquadrantsofHurricanesFloyd(1999), Charley(2004-2),andWilma(2005).Theverticalwind shearexceeds20ms1alongaheadingof24 26 in thesestorms.Thus,theregressionmodelspredictingthe raineldextenttowardthenortheastareexpectedtocontainvariablesrelatedtoverticalwindshear. Thenorthwestquadranthasthesecondlargestraineld extentonaverage(Fig. 8 ).Largerspansinthisquadrant mayariseduringanETthatproducesadelta-shapedrain Fig.8 Distribution,mean,andstandarddeviationoftheraineldextentinthe a northwest, b northeast, c southwest,and d southeastquadrants ofthe31hurricanesatlandfall Associationsbetweenthesizeofhurricanerainelds 141123

PAGE 8

eldextendingpolewardfromthecirculationcenter(Harr andElsberry 2000 ;Kleinetal. 2000 ;Milradetal. 2009 ). HurricanesFloyd(1999),Isabel(2003),Katrina(2005-2), andCharley(2004-2)haveraineldspansexceeding 350kmtowardthenorthwestandtheyallcompleteanET within37hoflandfall.Smallerspanscouldresultfromthe advectionofdriercontinentalairmassesintothewestern sideofthestorm(Cubukcuetal. 2000 ;Kimball 2008 ; Powell 1987 )asallhurricanessaveCharley(2004-1)and Wilma(2005)havetheirnorthwestquadrantlocatedover landatthetimeoflandfall.TCsthathavefastertangential windscanadvectmoisttropicalairfromtheireasternto theirwesternside,allowingforalargerraineldspan towardthenorthwest.Variablesassociatedwithmoisture, intensity,andenvironmentalconditionsassociatedwithan ETcouldallbeimportanttopredictingtheextentofthe raineldtowardthenorthwest. Mostofthehurricanes(23of31cases)havetheir smallestraineldextentinthesouthwestquadrant.Seven hurricanes,whosesouthwestquadrantisintheupshear direction,haveradiibelow100km(Fig. 8 ).Similartothe northwestquadrant,advectionofdryairfromcontinental airmassesasthestormsapproachlandfallorinteractwith middlelatitudeweathersystemscanlimittheraineld extenttowardthesouthwest.Variablesassociatedwith verticalwindshearormoisturemaybeimportanttothe predictionoftheraineldspaninthisquadrant. Five(four)hurricaneshavetheirlargest(smallest)span towardthesoutheast.Hurricaneswiththeirwesternhalf overland(e.g.,IkeandDolly(2008)),wherecontinental airmassesmaylimittheextentoftheraineldonthisside ofthestorm,havetheirlargestradiusinthesoutheast quadrant.Thevehurricaneswithradiilessthan150kmin thesoutheastquadrant(Fig. 8 )madelandfallalongthe coastofNorthCarolinaandeventuallybecameextratropical.Dryslotsthatformsouthofthecirculationcentersof TCsastheybecomeextratropical(Kleinetal. 2000 )limit theextentofconvectioneventhoughthesoutheastquadrantmaybelocateddownshear,aconditionthattendsto enhanceconvection(CorbosieroandMolinari 2002 ).Thus, variablesrelatedtostormlocationandmotion,ratherthan verticalwindshear,maybeimportantwhenpredictingthe raineldextenttowardthesoutheast. 5Comparisonstosizedata Theextentoftheraineldisnextcomparedtovariables thatrepresentthesizeofthestorm.Asthesizedatain severalquadrantsdonotfollowanormaldistribution, Spearman'srankcorrelationcoefcientsarecalculated betweentheextentoftheraineldineachquadrantandthe radiusof17,26,and33ms1windsineachquadrantas wellastheROCI.Theextentoftheraineldisalsosubtractedfromeachsizemeasuretodeterminewhichofthe sizemeasuresismostcloselyco-locatedwiththeedgeof theraineld,andwhichmeasureofstormsizemost completelyencompassestherainelds. ThemeasureofstormsizeexhibitingthehighestcorrelationacrossallquadrantsofraineldextentistheROCI (Table 4 ).ThepositiverelationshipbetweenROCIandthe raineldextentinthenortheast,northwest,andsouthwest quadrantsindicatesthattheraineldsspanagreaterdistancewhentheROCIislarge.Thisresultissimilartothat ofKonrad( 2001 ),whofoundastatisticallysignicant correlationbetweenROCIandthescaleofprecipitation eventscausedbyTCsinthe2daysfollowingtheirUS landfall.ClimatologicalstudiesofthestormsizebyMerrill ( 1984 )andKimballandMulekar( 2004 )foundtheaverage ROCItobe333and350km,respectively,fortheAtlantic basinTCsintheirstudy,whichagreeswellwiththe 365kmaverageinthecurrentstudy. Theextentsoftheraineldsonthewestsidesofthe hurricanesaremostcloselyrelatedtothesizeofthewind eld(Table 4 ).Theedgeoftheraineldinthenortheast quadrantisonlycorrelatedtotheextentofwindsinthe southeastquadrant,andtheraineldextentinthesoutheast quadrantdoesnotexhibitasignicantcorrelationwithany ofthesizemeasures.Thiscouldbebecausethesize measuresatthetimeoflandfallarecalculatedthrougha linearinterpolationofthedataavailableat00,06,12,and 18UTC.Ashalfofthehurricaneislocatedoverlandwhen theeyemakeslandfall,surfacefrictionthathelpsto decreasethetangentialspeedofthewindsisgreatestinthe forwardhalfofthestorm.Thus,theedgeofthevarious windthresholdsmaybeslightlyfartherfromorcloserto Table4 Spearman'sRhocorrelationcoefcientsbetweenmeasures ofstormsizeandtheextentoftheraineldineachquadrant SizemeasureRainNERainNWRainSWRainSE R17NE0.317 0.497 0.341 0.170 R17SE 0.4700.525 0.327 0.050 R17SW0.358 0.4550.540 0.029 R17NW0.329 0.4130.427 0.084 R26NE0.350 0.510 0.296 0.193 R26SE 0.4380.542 0.306 0.100 R26SW0.307 0.5640.531 0.121 R26NW0.251 0.4270.460 0.081 R33NE0.2790.3980.285 0.101 R33SE 0.433 0.4050.2650.004 R33SW0.225 0.4790.638 0.079 R33NW0.218 0.4540.638 0.065 ROCI 0.4090.5360.557 0.071 Italicisedvaluesarestatisticallysignicantat a = 0.01 142 C.J.Matyas123

PAGE 9

thecirculationcenteratthetimeoflandfallthanthe interpolatedvaluesindicate.Futureresearchshould examinetherelationshipbetweentheedgeoftheraineld andthewindeldspriortothetimewhenthewindelds interactwithlandtoachievebetterresults. WhensubtractingtheROCIfromtheraineldspanin eachquadrant,theROCIvalueislargerin90%ofthe comparisons,whichindicatesthattheROCItypically precedesthearrivaloftheheaviestrainfallaslandfall occurs.Aslocationsthatareexpectedtofalloutsideofthe ROCIhavealowchanceofreceivingheavyrainfallasthe hurricanemovesinland,thisresultsuggeststhattheROCI couldbeutilizedtodelineateregionsatriskfromheavy rainfallfromthosethatarenot.Theradiusofgale-force windsismostcloselyco-locatedwiththeedgeoftherain eldonaverage.Theraineldsinthenortheastand northwestquadrantstendtobe5075kminfrontofR17, whiletheedgeoftheraineldislocated8and39km closertothecirculationcenterthanR17inthesouthwest andsoutheastquadrants,respectively. 6Statisticalmodeling TherainelddistancesdiscussedinSect. 4 indicatethat theradiusoftheraineldvariesineachquadrantofa hurricane,andsuggestthatdifferentcombinationsof environmentalforcingmaybeaffectingthespatialdistributionofrainfallineachquadrantofthestorm.For example,whenhurricanetroughinteractionsoccur,verticalwindshearishigh,divergenceat200hPaisstrong,and theforwardmotionofthehurricaneincreases(Atallahand Bosart 2003 ;BosartandLackmann 1995 ;EvansandHart 2003 ;Hanleyetal. 2001 ;Hanley 2002 ;Harretal. 2000 ; Jonesetal. 2003 ;Pengetal. 2007 ;ShapiroandMoller 2003 ).Theseconditionsmayenhanceorinhibitconvection ondifferentsidesofthestorm.Otherstudieshaverelated themoisturecontentintheenvironmentsurroundinga hurricanetothedistributionofitsrainfall.HillandLackmann( 2009 )andJiangetal.( 2008a b )suggestthatTCs havelargerraineldswhenhighmoisturecontentexistsin theenvironmentsurroundingthestorm. Tofurtherexploretherelationshipsbetweenvariables representingtheenvironmentalconditionssurroundingthe stormsandtheradiusoftheraineldsatthetimeof landfall,thecurrentstudyutilizesstatisticalmodels. Employingmultiplelinearregressionanalysisallowsthe relativecontributionsofeachpredictortothemodeltobe assessedbycalculatingthepartialcorrelationsofeach predictortothedependentvariable.Thus,theresultsofthe analyseswillindicatewhichphysicalforcingmechanism(s)mayexertthedominantinuenceontheextentof theraineldineachquadrantofthehurricane.Theuseof regressionanalysisinTCresearchiswelldocumentedas researchershaveusedittopredicttheextentofTCwind elds(BellandRay 2004 ;Demuthetal. 2004 2006 ; Muelleretal. 2006 )andtoforecastTCintensity(DeMaria andKaplan 1994 ;DeMariaetal. 2005 ). Toascertainwhichenvironmentalconditionsexhibitthe strongestrelationshipwithraineldsize,multiplelinear regressionmodelsaredevelopedtopredicttheextentofthe raineldineachquadrantofthestormatthetimeof landfall.Threesetsofmodelsaredeveloped;theyutilize predictorsatt0,t12,andt24.Modelsaredevelopedto predicttheraineldextentinquadrantsplacedaccording tocardinaldirection,stormmotion,andverticalwind shear.However,astheresultsfromthecardinaldirectionrelativequadrantsexhibithigherpercentagesofexplained varianceoverall,theresultsfromtheotherquadrant placementsarenotdiscussed.Thedeterminationofwhich conditionsmayhavethestrongestinuenceonraineld extentisbasedonhowfrequentlyeachpredictorassociated withaparticularconditionappearsinthe12regression models,andthevalueofitsstandardizedpartialcorrelation coefcients.Itisimportanttonotethatiftheresultsof thesemodelsareinagreementwithpreviousobservational andmodelingstudies,thentheGIS-basedmethodsof spatialanalysisperformedinthisstudyarevalidatedand canbeutilizedinfurtherresearch. Allvariablesshouldbenormallydistributedandmulticollinearitymustbeminimizedinorderforamultiple linearregressionanalysistobesuccessful,Accordingtothe resultsofaShapiro-WilksWtest(Kleinetal. 2000 ),allvariableslistedinTable 2 haveanormaldistribution. However,manyofthepredictorsarerelatedtooneanother astheyrepresentmeasurementsofconditionsinadjacent layersoftheatmosphere,suchastheuppertropospheric temperaturevariables.Whenperformingaregression analysis,multicollinearityofvariablesisundesirableasthe varianceinboththemodelandthecoefcientsofthe predictorsbecomesinated.Thus,beforetheregression modelsaredeveloped,across-correlationmatrixisconstructedforalldependentandindependentvariables.For groupsofvariablesexhibitinghighcorrelationvalues (| r | [ 0.33; a = 0.05),theonevariableinthegroupwith thehighestcorrelationtothedependentvariableforeach modelisenteredintotheregressionanalysis. Tofurtherreducethenumberofpredictorsutilizedin eachmodel,aforwardstepwiseapproachisutilized.At eachstep,theindependentvariablethathasthehighest correlationtothedependentvariableandthatisstatistically signicantat a = 0.05isaddedtothemodel,whilecontrollingforvariablesthatarealreadyincludedinthemodel. Thismethodrepeatsuntilnoneoftheremainingindependentvariablesisstatisticallysignicanttothemodel. Successfulmodelswillhaveahighamountofvariancein Associationsbetweenthesizeofhurricanerainelds 143123

PAGE 10

theradiioftheraineldsthatisexplainedbythepredictors ineachmodel(i.e.,ahigh R2value).Standardizingthe regressioncoefcients,whichrepresenttherelationshipof eachindependentvariabletothedependentvariable,rendersthemindependentofunitsofmeasurementtofacilitate theidenticationofthepredictor(s)thatcontributesthe mosttowardexplainingthevarianceinthedependent variable. 7Resultsofstatisticalmodeling Anexaminationofthepercentageofexplainedvariance acrossthe12regressionmodelsshowsthatraineldextent ispredictedbestforthenortheastquadrant,followedbythe northwest,southwest,andsoutheastquadrants(Tables 5 6 7 ).Researchexaminingtheeffectsofverticalwindshear onTCsfounda24-hlagbetweentheonsetofverticalwind shearandchangestothestructuresoftheirmodeledstorms (Bender 1997 ;FrankandRitchie 1999 2001 ).Yet,thet24 modelsinthecurrentstudyyieldthelowestpercentagesof explainedvariancewithinthethreetimeperiods.This resultcouldbeattributedtothelargechangesinvertical windsheardirectionandvelocityexperiencedbymostof theTCsduringthe24-hperiodpriortolandfallas18 stormsexperienceachangeinvelocitythatisgreaterthan 5ms1.However,themodelsthatutilizepredictorsfrom t12yieldhigher R2valuesthanthoseutilizingpredictorsat thetimeoflandfall,suggestingthatthereisa12-hlag betweenconditiononsetandtheresultingextentofconvectiverainfall. Thereare27statisticallysignicantpredictorscontained withinthe12multiplelinearregressionmodelsthatpredict theraineldextentineachquadrantatthetimeoflandfall. AnexaminationofthePearsoncorrelationcoefcients amongthesepredictorsrevealsthattheycanbegrouped intofourmaininuencesontheraineldextent.Examiningthefrequencyandthestrengthwithwhichpredictors contributetotheexplainedvarianceinraineldextent allowstheseinuencestoberankedasfollows:(1)vertical windshear,(2)moisture,(3)intensity,and(4)location. Verticalwindshearexertsthestrongestinuenceonthe extentofconvectionaboutthecirculationcentersofhurricanesatthetimeoflandfall.Ofthe27predictorscontainedwithinthe12regressionmodels,11arerelatedtothe strengthand/ordirectionoftheverticalwindshear(ShDN, ShDE,ShDV,U200).Thisresultsupportsthoseofother studies(e.g.Cecil 2007 ;Chenetal. 2006 ;Lonfatetal. 2004 2007 ;Matyas 2010 ;Rogersetal. 2003 )suggesting thatstrongverticalwindshearisaleadingcauseofconvectionthatisasymmetricallydistributedinTCs.Studies byCorbosieroandMolinari( 2002 ),Chenetal.( 2006 ),and Cecil( 2007 )showthatincreasedverticalwindshearresults inashiftofconvectiontowardthedownshearquadrants, andreducedconvectionintheupshearquadrants.The resultsofthisstudyaddtopreviousresearchasthey Table5 Resultsofmultiplelinearregressionanalysesperformedto predicttheradiusoftheraineldsofhurricanesatthetimeoflandfall usingpredictorsatthetimeoflandfall Dependent variable R2Independent variable Standardized coefcient Northeastquadrant0.782ShDN0.884 Northwestquadrant0.646DirN0.624 T2000.331 RhHi0.324 Southwestquadrant0.598LatM 0.592 U200 0.494 RhLo0.254 Southeastquadrant0.379T0000.545 ShDE0.331 Table6 Resultsofmultiplelinearregressionanalysesperformedto predicttheradiusoftheraineldsofhurricanesatthetimeoflandfall usingpredictorsfrom12hpriortolandfall Dependent variable R2Independent variable Standardized coefcient Northeastquadrant0.784ShDN0.813 ShDE0.189 Northwestquadrant0.727ShDN0.563 T1500.382 RhHi0.258 Southwestquadrant0.629U200 0.563 Vmax0.316 RhHi0.307 Southeastquadrant0.297Lon 0.409 RhLo0.332 Table7 Resultsofmultiplelinearregressionanalysesperformedto predicttheradiusoftheraineldsofhurricanesatthetimeoflandfall usingpredictorsfrom24hpriortolandfall Dependent variable R2Independent variable Standardized coefcient Northeastquadrant0.548ShDN0.740 Northwestquadrant0.622 Vmax0.642 Lon0.316 ShDV0.296 Southwestquadrant0.519U200 0.539 RhHi0.419 Southeastquadrant0.292Lon 0.368 RhLo0.329 144 C.J.Matyas123

PAGE 11

demonstratethattheextentofconvectionextendsoutwards withincreasingverticalwindshear.Strongverticalwind shearoccurswhenhurricanesinteractwithmiddlelatitude weathersystems,upper-levelcut-offlows,oroutowfrom othertropicalcyclonesororganizedconvectivesystems (Hanleyetal. 2001 ;Jonesetal. 2003 ;RitchieandElsberry 2001 ). Thethreehigheststandardizedcoefcientsofthe27 predictorsareforthemagnitudeofthenorthwardvertical windshearcomponent(ShDN)att24,t12,andt0asa predictoroftheraineldspaninthenortheastquadrant. Thestrongsouthwesterlyverticalwindshearexperienced bymosthurricanesinthecurrentstudy(Fig. 4 )placesthis enhancedconvectioninthenortheastquadrant,whichtends tobeco-locatedwitheitherdownshearquadrant(Table 3 ). Whenshearhasastrongereastwardcomponent,the northeastandsoutheastquadrantsareplacedinthedownsheardirection,thusaccountingforthepositivecorrelation ofvariableShDEwithraineldextentatt12andt0 (Tables 5 6 ).Zonalwindat200hPaisalsorelatedtorain eldextent(Tables 5 6 7 ).Increasedzonalwindat 200hPaiscloselyassociatedwithanincreasedwesterly componentofthe850200hPadeep-layerverticalwind shear.Thus,itisnotsurprisingthatwhenU200islarge,the extentoftheraineldinthesouthwestquadrantissmallas thisquadrantislocatedintheupsheardirection. Moistureisthesecondmostimportantpredictorofrain eldextentatthetimeoflandfall.VariablesRhLoand RhHiappearin7ofthe12regressionmodelsandcontributetotheexplainedvarianceinraineldextentinthe northwest,southwest,andsoutheastquadrants(Tables 5 6 7 ).Inpreviousstudies,increasedmoistureavailableinthe regionsurroundingtheTChasbeenrelatedtoincreases convectivecloudformationandrainrates(Jiangand Halverson 2008 ;Rodgersetal. 1994 1998 2000 ;Rodgers andPierce 1995 ).Recently,HillandLackmann( 2009 ) foundthatenvironmentalrelativehumidityisanimportant controlontheextentofboththeouterspiralrainbandsand thewindeldsoftheirmodeledTCs.Thestatistically signicantpositivecorrelationsbetweenrelativehumidity andraineldextentfoundinthecurrentstudyprovide observationalsupporttothendingsofHillandLackmann ( 2009 ).RelativehumidityvaluesforTCsinthecurrent studyrangedfrom52to76%inthelowertroposphereand from33to66%intheuppertroposphere.Thesevalues comparewellwiththoseutilizedbyHillandLackmann ( 2009 ),whosetrelativehumidityvaluesat80%forthe inner150kmofthestorm,and20,40,60,and80%forthe regionoutsideof150km.Dryairadvectionfromcontinentalairmasses(e.g.,Cubukcuetal. 2000 ),middlelatitudeweathersystems(e.g.Rodgersetal. 1991 ),orthe Saharanairlayer(e.g.DunionandVelden 2004 )can reduceprecipitationinonepartofthestorm.Astherelative humidityvariablesintheSHIPSdatabaserepresentan averagedvalueovera200800kmradialdistancefromthe TC'scirculationcenter,differencesinrelativehumidity valueswithineachquadrantcannotbequantied.Higher correlationcoefcientsbetweenraineldextentandrelativehumiditycouldbeattainedifmoistureisexaminedin eachquadrantofthestormutilizingNorthAmerican RegionalReanalysisdata. Previousresearchhasshownthatstormintensityis relatedtorainfallratesproducedbythestorm(Jiangetal. 2008a b ;Lonfatetal.2004 ;Tuleyaetal. 2007 ).Thecurrentstudyndsthatstormintensityiscorrelatedwiththe extentoftheraineldonthewesternsideofthehurricanes,indicatingpositiverelationshipsamongstorm intensity,raineldsize,andstormsymmetry.Higher maximumsustainedwindspeediscorrelatedtoalarger raineldextenttowardthenorthwestatt24andsouthwest att12(Tables 6 7 ).Temperaturesintheuppertroposphere (T150,T200),whichareknowntobecorrelatedwithTC intensity(e.g.Gray 1979 ),arepositivelycorrelatedwith theextentoftheraineldtowardthenorthwestatt0and t12.Stormsymmetryisanimportantindicatorofintensity asindicatedbyDvorak( 1975 ).Fastertangentialwindscan advectmoisturefromtheeasterntothewesternsideofthe hurricanetoallowconvectiontoformatgreaterdistances fromthecirculationcentertocreateamoresymmetrical TC.Weakertangentialwindsmaynotallowforthe advectionoftropicalmoisturecounterclockwiseintothe southwesternquadrantofthestorm,creatingasmaller extentoftheraineldinthislocationandproducingan asymmetricalraineld. Additionally,thelatitudeatwhichTCsreachtheir maximumintensityisimportantwhenpredictingtherain eldextenttowardthesouthwest.OnceTCsreachtheir maximumintensity,theygrowinsizeasmoreangular momentumentersthesystemthanisneededtomaintainits circulation(Merrill 1984 ).DunnandMiller( 1960 )and Maclayetal.( 2008 )foundthatmanyTCscontinuetogrow insizeafterreachingtheirmaximumintensity.Theresults ofthecurrentstudyindicatethathurricanesreachingtheir maximumintensityatlowerlatitudestendtohavealarger raineldextentinthesouthwesternquadrantatthetimeof landfallthanthosereachingmaximumintensityathigh latitudes. Stormlocationpriortothetimeoflandfallisalsocorrelatedwithraineldsize.Hurricaneslocatedfartherwest (east)haveasmallerraineldextenttowardthenorthwest (southeast).Theninestormslocatedwestof88 Wlongitudeatt24havelargerraineldstowardthenortheastand southeastthantowardthenorthwest,andtheextentisless than210kmtowardthenorthwestineightofthesecases. OnlyHurricaneRita(2005)featuresanorthwestextent greaterthan300kminthisregion.Incontrast,justeightof Associationsbetweenthesizeofhurricanerainelds 145123

PAGE 12

thecaseslocatedeastof88 Wfeatureraineldspans towardthenortheastandsoutheastlargerthanthattoward thenorthwestquadrant.Inthedaypriortolandfall,the entrainmentofdriercontinentalairmassesintothewestern sidesofTCswhenlocatedinthewesternGulfofMexico couldaccountforthesmallernorthwestspansofTCsas theymakelandfallinTexasorLouisiana. Atbotht12andt24,84 Wlongitudeseparatessmall fromlargeraineldextenttowardthesoutheast.Sevenof thetensoutheastwardspansthatarelessthan200km belongtostormslocatedeastofthislongitude,andallof thesestormshavesmallerextentstowardthesouththan towardthenorth.SixofthesesevenTCseventually becomeextratropicalandmovenorthwestatt24followed byaturntowardthenorthornortheastpriortolandfall.As TCsmovepoleward,outowincreasesandinteractionwith amid-latitudebarocliniczonefavorsenhancedprecipitationnorthofthecirculationcenter(Jonesetal. 2003 ).At thesametime,moisturedecreasesonthesouthsideofthe storm(Kleinetal. 2000 )andtheextentoftheraineldis reduced.Onlytwoofthetensoutheastwardspansover 260kmsoutheastbelongtostormsthatarelocatedeastof 84 W;HurricanesCharley(2004-1)andKatrina(2005-1) arelocatedsouthof29 Nlatitude,andmaynotbe encounteringthedryairadvectiononthesouthsideofthe stormthatisexperiencedbystormslocatednorthof30 N. 8Conclusionsandfutureresearch Whilemanypreviousstudieshaveexaminedlocations whereconvectiverainfalldevelopswithinTCsinresponse todifferentenvironmentalconditions,thisstudyinvestigatedenvironmentalinuencesontheradialextentofthe raineldsatthetimeoflandfall.Thirty-oneUShurricane landfallswerestudied.Foreachquadrantofthestorm,the distancefromthecirculationcentertotheedgeoftherain eldwascalculatedthroughananalysisofradarreectivity datausingGIS.Theresultsofthestudyshowedthathurricanes,whichdidnotbecomeextratropicalwerethemost symmetricallyshapedatthetimeoflandfall,whilethose within2daysofbecomingextratropicalwerethemost asymmetrical.Althoughtheedgesoftheraineldsdidnot preciselyalignwithanyofthemeasuresofstormsize,they weremostcloselyco-locatedwiththeradiusofgale-force windsineachquadrant.TheROCIencompassedtherain elds90%ofthetime,indicatingthatitcanbeauseful delineationofregionsthatreceivedconvectiverainfall producedbyTCsatlandfall. Todeterminewhichatmosphericconditionsweremost likelyresponsibleforraineldsizeatthetimeoflandfall, multiplelinearregressionsmodelspredictedtheraineld extentineachquadrant.Eachsetoffourmodelsutilized predictorsfromthetimeoflandfall,or12or24hpriorto landfall.Themodelspredictedtheextentoftherainelds bestwhenutilizingvariablesfromt12,indicatingthatit takesabout12hforatmosphericconditionstofully inuenceraineldsize.Examiningthefrequencywith whicheachvariablewasutilizedasapredictorinthe12 modelsandthevaluesoftheirstandardizedcorrelation coefcients,itwaspossibletodeterminewhichatmosphericconditionslikelyexertedthemostinuenceonrain eldsize.Verticalwindshearwasthemostimportant predictorofraineldsizeasstrongsouthwesterlywind shearwashighlycorrelatedtoalarge(small)extenttoward thenortheast(southwest).Moisture,intensity,andlocation werecorrelatedwithraineldextenttowardthenorthwest, southwest,andsoutheast. Itisclearthattransitioningintoanextratropicalsystem exertsastronginuenceontheasymmetricalextentofthe raineldsinthefourquadrantsofthehurricanes.Analysis offuturehurricanelandfallsshouldincreasesamplesizeso thatstormsthatdomakeanETcanundergostatistical analysisseparatefromthosethatdonot.Forhurricanesthat donotmakethistransition,conditionssuchastheshape andorientationofthecoastlinemaybecomeimportant indicatorsofraineldsize.Quantifyingatmospheric moistureineachquadrantofthestormmayimprovethe correlationbetweenmoistureandraineldsize.Asdata pertainingtoatmosphericconditionssurroundingweaker TCsarenotdistributednormally,non-parametricstatistical testingcouldbeemployedtofurtherexaminetheinuencesoftheenvironmentontheraineldsizesofthese systems.Additionally,tobetterdeterminehowcloselythe edgesoftheraineldsalignwiththewindelds,analysis shouldcommencepriortothelandfalloftheradiusofgaleforcewinds.Acknowledgments TheauthorwishestothankKevinAshandDr. MarkDeMariafortheircommentsonearlierdraftsofthismanuscript. Thecommentsoftworeviewerswereextremelyhelpfulinproducing thenaldraftofthemanuscript.ReferencesAnsariS,DelGrecoS(2005)GIStoolsforvisualizationandanalysis ofNEXRADradar(WSR-88D)archiveddataattheNational ClimaticDataCenter.21stinternationalconferenceonInteractiveInformationProcessingSystems(IIPS)forMeteorology, Oceanography,andHydrology.AmericanMeteorologicalSociety,SanDiego AtallahEH,BosartLR(2003)Theextratropicaltransitionand precipitationdistributionofHurricaneFloyd(1999).MonWea Rev131:10631081 AtallahEH,BosartLF,AiyyerAR(2007)Precipitationdistribution associatedwithlandfallingtropicalcyclonesovertheeastern UnitedStates.MonWeaRev135:21852206 BellK,RayPS(2004)NorthAtlantichurricanes197799:surface hurricane-forcewindradii.MonWeaRev132:11671189 146 C.J.Matyas123

PAGE 13

BenderMA(1997)Theeffectofrelativeowontheasymmetric structureintheinteriorofhurricanes.JAtmosSci54:703724 BlackML,GamacheJF,MarksFD,SamsuryCE,WilloughbyHE (2002)EasternPacicHurricanesJimenaof1991andOliviaof 1994:theeffectofverticalshearonstructureandintensity.Mon WeaRev130:22912312 BlackwellKG(2000)EvolutionofHurricaneDanny(1997)at landfall:Doppler-observedeyewallreplacement,vortexcontraction/intensication,andlow-levelwindmaxima.MonWeaRev 128:40024016 BosartLF,LackmannGM(1995)Postlandfalltropicalcyclone reintensicationinaweaklybaroclinicenvironmentacasestudyofHurricaneDavid(September1979).MonWeaRev 123:32683291 CecilDJ(2007)Satellite-derivedrainratesinverticallysheared tropicalcyclones.GeophysResLett34:L02811 ChanJCL,LiuKS,ChingSE,LaiEST(2004)Asymmetric distributionofconvectionassociatedwithtropicalcyclones makinglandfallalongtheSouthChinacoast.MonWeaRev 132:24102420 ChenSYS,KnaffJA,MarksFD(2006)Effectsofverticalwindshear andstormmotionontropicalcyclonerainfallasymmetries deducedfromTRMM.MonWeaRev134:31903208 CorbosieroKL,MolinariJ(2002)Theeffectsofverticalwindshear onthedistributionofconvectionintropicalcyclones.MonWea Rev130:21102123 CorbosieroKL,MolinariJ(2003)Therelationshipbetweenstorm motion,verticalwindshear,andconvectiveasymmetriesin tropicalcyclones.JAtmosSci60:366376 CubukcuN,PfefferRL,DietrichDE(2000)Simulationoftheeffects ofbathymetryandland-seacontrastsonhurricanedevelopment usingacoupledocean-atmospheremodel.JAtmosSci57:481 492 DeMariaM,KaplanJ(1994)Astatisticalhurricaneintensity predictionscheme(SHIPS)fortheAtlanticbasin.WeaForecast 9:209220 DeMariaM,KaplanJ(1999)AnupdatedStatisticalHurricane IntensityPredictionScheme(SHIPS)fortheAtlanticandeastern NorthPacicbasins.WeaForecast14:326337 DeMariaM,MainelliM,ShayLK,KnaffJA,KaplanJ(2005)Further improvementstotheStatisticalHurricaneIntensityPrediction Scheme(SHIPS).WeaForecast20:531543 DemuthJL,DeMariaM,KnaffJA,VonderHaarTH(2004) Evaluationofadvancedmicrowavesoundingunittropicalcycloneintensityandsizeestimationalgorithms.JApplMeteor 43:282296 DemuthJL,DeMariaM,KnaffJA(2006)Improvementof advancedmicrowavesoundingunittropicalcycloneintensity andsizeestimationalgorithms.JApplMeteorolClimatol 45:15731581 DunionJP,VeldenCS(2004)TheimpactoftheSaharanairlayeron Atlantictropicalcycloneactivity.BullAmMeteorSoc85:353 DunnGE,MillerBI(1960)AtlanticHurricanes.LouisianaState UniversityPress,BatonRouge,p326 DvorakVF(1975)Tropicalcycloneintensityanalysisandforecasting fromsatelliteimagery.MonWeaRev103:420430 ElsberryRL(2002)Predictinghurricanelandfallprecipitation: Optimisticandpessimisticviewsfromthesymposiumon precipitationextremes.BullAmMeteorolSoc83:13331339 EvansJL,HartRE(2003)Objectiveindicatorsofthelifecycle evolutionofextratropicaltransitionforAtlantictropical cyclones.MonWeaRev131:909925 FrankWM(1977)Thestructureandenergeticsofthetropicalcyclone I.Stormstructure.MonWeaRev105:11191135 FrankWM,RitchieEA(1999)Effectsofenvironmentalowupon tropicalcyclonestructure.MonWeaRev127:20442061 FrankWM,RitchieEA(2001)Effectsofverticalwindshearonthe intensityandstructureofnumericallysimulatedhurricanes.Mon WeaRev129:22492269 GrayWM(1979)Tropicalcycloneintensitydeterminationthrough upper-troposphericaircraftreconnaissance.BullAmMeteorSoc 60:10691074 HanleyDE(2002)Theevolutionofahurricane-troughinteraction fromasatelliteperspective.WeaForecast17:916926 HanleyD,MolinariJ,KeyserD(2001)Acompositestudyofthe interactionsbetweentropicalcyclonesandupper-tropospheric troughs.MonWeaRev129:25702584 HarrPA,ElsberryRL(2000)Extratropicaltransitionoftropical cyclonesoverthewesternnorthpacic.PartI:evolutionof structuralcharacteristicsduringthetransitionprocess.MonWea Rev128:26132633 HarrPA,ElsberryRL,HoganTF(2000)Extratropicaltransitionof tropicalcyclonesoverthewesternNorthPacic.PartII:the impactofmidlatitudecirculationcharacteristics.MonWeaRev 128:26342653 HartRE,EvansJL(2001)Aclimatologyoftheextratropical transitionofAtlantictropicalcyclones.JClimate14:546564 HillK,LackmannGM(2009)Inuenceofenvironmentalhumidity ontropicalcyclonesize.MonWeaRev137:32943315 JarvinenBR,NeumannCJ,DavisMAS(1984)Atropicalcyclone datatapeforthenorthAtlanticbasin,18861983.NOAATech MemoNWSNHC22,Miami,p21 JiangHY,HalversonJB(2008)Onthedifferencesinstormrainfall fromHurricanesIsidoreandLili.PartII:waterbudget.Wea Forecast23:4461 JiangHY,HalversonJB,SimpsonJ,ZipserEJ(2008a)Hurricane RainfallPotential''derivedfromsatelliteobservationsaids overlandrainfallprediction.JApplMeteorolClimatol47:944959 JiangHY,HalversonJB,ZipserEJ(2008)Inuenceofenvironmental moistureonTRMM-derivedtropicalcycloneprecipitationover landandocean.GeophysResLett35.doi: 10.1029/2008 gl034658 JonesSC,HarrPA,AbrahamJ,BosartLF,BowyerPJ,EvansJL, HanleyDE,HanstrumBN,HartRE,LalauretteF,SinclairMR, SmithRK,ThorncroftC(2003)Theextratropicaltransitionof tropicalcyclones:forecastchallenges,currentunderstanding, andfuturedirections.WeaForecast18:10521092 KaplanJ,DeMariaM(1995)Asimpleempirical-modelforpredicting thedecayoftropicalcyclonewindsafterlandfall.JApplMeteor 34:24992512 KidderSQ,KusselsonSJ,KnaffJA,FerraroRR,KuligowskiRJ,Turk M(2005)Thetropicalrainfallpotential(TRaP)technique.PartI: descriptionandexamples.WeaForecast20:456464KimballSK(2008)Structureandevolutionofrainfallinnumerically simulatedlandfallinghurricanes.MonWeaRev136:38223847 KimballSK,MulekarMS(2004)A15-yearclimatologyofNorth Atlantictropicalcyclones.PartI:sizeparameters.JClimate 17:35553575 KleinPM,HarrPA,ElsberryRL(2000)Extratropicaltransitionof westernNorthPacictropicalcyclones:anoverviewand conceptualmodelofthetransformationstage.WeaForecast 15:373395 KonradCE(2001)Themostextremeprecipitationeventsoverthe easternUnitedStatesfrom1950to1996:considerationsofscale. JournalofHydrometeorology2:309325 LiuKS,ChanJCL,ChengWC,TaiSL,WongPW(2007) Distributionofconvectionassociatedwithtropicalcyclones makinglandfallalongtheSouthChinacoast.MeteorolAtmos Phys97:5768 LiuGR,ChaoCC,HoCY(2008)Applyingsatellite-estimatedstorm rotationspeedtoimprovetyphoonrainfallpotentialtechnique. WeaForecast23:259269 Associationsbetweenthesizeofhurricanerainelds 147123

PAGE 14

LonfatM,MarksFD,ChenSYS(2004)Precipitationdistributionin tropicalcyclonesusingtheTropicalRainfallMeasuringMission (TRMM)MicrowaveImager:aglobalperspective.MonWea Rev132:16451660 LonfatM,RogersR,MarchokT,MarksFD(2007)Aparametric modelforpredictinghurricanerainfall.MonWeaRev 135:30863097 MaclayKS,DeMariaM,VonderHaarTH(2008)Tropicalcycloneinnercorekineticenergyevolution.MonWeaRev136:48824898 MarksFD(1985)Evolutionofthestructureofprecipitationin HurricaneAllen(1980).MonWeaRev113:909930 MatyasCJ(2010)Ageospatialanalysisofconvectiverainfallregions withintropicalcyclonesafterlandfall.IntJApplGeospatRes (inpress) MedlinJM,KimballSK,BlackwellKG(2007)Radarandraingauge analysisoftheextremerainfallduringHurricaneDanny's(1997) landfall.MonWeaRev135:18691888 MerrillRT(1984)Acomparisonoflargeandsmalltropicalcyclones. MonWeaRev112:14081418 MerrillRT(1988)Environmental-inuencesonhurricaneintensication.JAtmosSci45:16781687 MillerBI(1958)RainfallratesinFloridahurricanes.MonWeaRev 258-264 MilradSM,AtallahEH,GyakumJR(2009)Dynamicaland precipitationstructuresofpoleward-movingtropicalcyclones ineasternCanada,19792005.MonWeaRev137:836851 MoyerAC,EvansJL,PowellM(2007)Comparisonofobservedgale radiusstatistics.MeteorolAtmosPhys97:4155 MuellerKJ,DeMariaM,KnaffJ,KossinJP,VonderHaarTH(2006) Objectiveestimationoftropicalcyclonewindstructurefrom infraredsatellitedata.WeaForecast21:9901005 PengMS,MaueRN,ReynoldsCA,LanglandRH(2007)Hurricanes Ivan,Jeanne,Karl(2004)andmid-latitudetroughinteractions. MeteorolAtmosPhys97:221237 PowellMD(1987)Changesinthelow-levelkinematicandthermodynamicstructureofHurricaneAlicia(1983)atlandfall.Mon WeaRev115:7599 RappaportEN(2000)LossoflifeintheUnitedStatesassociatedwith recentAtlantictropicalcyclones.BullAmMeteorSoc81:2065 2073 RitchieEA,ElsberryRL(2001)Simulationsofthetransformation stageoftheextratropicaltransitionoftropicalcyclones.Mon WeaRev129:14621480 RodgersEB,PierceHF(1995)EnvironmentalinuenceonTyphoonBobbieprecipitationdistribution.JApplMeteor34:25132532 RodgersEB,ChangSW,StoutJ,SterankaJ,ShiJJ(1991)Satelliteobservationsofvariationsintropicalcycloneconvectioncaused byupper-tropospherictroughs.JApplMeteor30:11631184 RodgersEB,BaikJJ,PierceHF(1994)Theenvironmentalinuence ontropicalcycloneprecipitation.JApplMeteor33:573593 RodgersEB,OlsonWS,KaryampudiVM,PierceHF(1998)Satellitederivedlatentheatingdistributionandenvironmentalinuences inHurricaneOpal(1995).MonWeaRev126:12291247 RodgersE,OlsonW,HalversonJ,SimpsonJ,PierceH(2000) EnvironmentalforcingofsupertyphoonPaka's(1997)latentheat structure.JApplMeteor39:19832006 RogersRF,ChenSS,TenerelliJ,WilloughbyHE(2003)Anumerical studyoftheimpactofverticalshearonthedistributionofrainfall inHurricaneBonnie(1998).MonWeaRev131:15771599 ShapiroLJ(1983)Theasymmetricboundary-layerowundera translatinghurricane.JAtmosSci40:19841998 ShapiroLJ,MollerJD(2003)Inuenceofatmosphericasymmetries ontheintensicationofHurricaneOpal:piecewisePVinversion diagnosisofaGFDLmodelforecast.MonWeaRev131:1637 1649 TuleyaRE,DeMariaM,KuligowskiRJ(2007)EvaluationofGFDL andsimplestatisticalmodelrainfallforecastsforUSlandfalling tropicalstorms.WeaForecast22:5670 VieuxBE,BedientPB(1998)Estimationofrainfallforood predictionfromWSR-88Dreectivity:acasestudy,1718 October1994.WeaForecast13:407415 WilloughbyHE,MarksFD,FeinbergRJ(1984)Stationaryand movingconvectivebandsinhurricanes.JAtmosSci41:3189 3211 ZhengX,DuanYH,YuH(2007)Dynamicaleffectsofenvironmental verticalwindshearontropicalcyclonemotion,structure,and intensity.MeteorolAtmosPhys97:207220 148 C.J.Matyas123