<%BANNER%>

UFIR



A model of hurricane-caused tree debris in Houston, Texas
CITATION PDF VIEWER
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/IR00001207/00001
 Material Information
Title: A model of hurricane-caused tree debris in Houston, Texas
Series Title: Thompson, B. K., F. J. Escobedo, C. L. Staudhammer, C. J. Matyas, and Y. Qiu, 2011: A model of hurricane-caused tree debris in Houston, Texas. Landscape and Urban Planning, Published online 25 March, 2011, DOI:10.1016/j.landurbplan.2011.1002.1034.
Physical Description: Journal Article
Creator: Matyas, Corene
Publisher: Landscape and Urban Planning
Publication Date: 2011
 Notes
Abstract: Information onpost-hurricaneurbanforestdebrisintheSoutheasternUnitedStatesislimited.Astratified subset ofpermanentplotswithinthecityofHouston,Texas,originallyestablishedin2001wereselected and measuredforhurricaneimpactsonurbanforeststructureanddebrisgenerationfollowingHurricane Ike whichstrucktheHoustonregiononSeptember13,2008.Threestatisticalmodelsweredeveloped and inputparametersincludedmeasuredurbanforeststructuredata,landcoverdatafromexistingplots, the NationalHurricaneCenter’sH*Winddataset,andtheUnitedStatesGeologicalSurvey’sNationalLand Cover Database.Thestatisticalmodelsestimatedtreedebrisbasedonalignmentofpre-andpost-storm data. Landcoverwastestedasaproxyvariableforpre-stormurbanforestbiomassandpost-stormtree debris. Estimatesofpre-storm,urbanforestbiomasswereestablishedtotestthestatisticalrelationship between pre-stormbiomassandpoststormdebris.Testingofbothlandcoverasaproxyvariableand the biomass–debrisrelationshipwereperformedtosimplifyvolumetricestimatesofdebrisproduced through statisticalmodeling.Debrismodelswerespatiallyanalyzedtodeterminedebrisdistributionand to comparewithexistingliteratureandavailablevegetationdebrisestimationmodels.Resultssuggest that urbanforeststructurevariableshavegreaterinfluenceovervariationindebrisestimatesthanstorm- related variables.
Acquisition: Collected for University of Florida's Institutional Repository by the UFIR Self-Submittal tool. Submitted by Corene Matyas.
Publication Status: Published
 Record Information
Source Institution: University of Florida Institutional Repository
Holding Location: University of Florida
Rights Management: All rights reserved by the submitter.
System ID: IR00001207:00001

Downloads

This item is only available as the following downloads:

2thompsonetal2011 ( PDF )


Full Text

PAGE 1

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 LandscapeandUrbanPlanning xxx (2011) xxx…xxx Contentslistsavailableat ScienceDirectLandscapeandUrbanPlanning journalhomepage: www.elsevier.com/locate/landurbplan Modelinghurricane-causedurbanforestdebrisinHouston,TexasBenjaminK.Thompsona,FranciscoJ.Escobedoa,ChristinaL.Staudhammera , 1,CoreneJ.Matyasb, YouliangQiubaSchoolofForestResourcesandConservation,UniversityofFlorida,POBox110410,Gainesville,FL32611,USAbDepartmentofGeography,UniversityofFlorida,POBox117315,Gainesville,FL32611,USA articleinfo Articlehistory: Received30May2010 Receivedinrevisedform20February2011 Accepted22February2011 Available online xxx Keywords: UrbanForestEffectsmodel Hurricanedamage Urbanforestmanagement Emergencymanagementabstract Informationonpost-hurricaneurbanforestdebrisintheSoutheasternUnitedStatesislimited.Astrati“ed subsetofpermanentplotswithinthecityofHouston,Texas,originallyestablishedin2001wereselected andmeasuredforhurricaneimpactsonurbanforeststructureanddebrisgenerationfollowingHurricane IkewhichstrucktheHoustonregiononSeptember13,2008.Threestatisticalmodelsweredeveloped andinputparametersincludedmeasuredurbanforeststructuredata,landcoverdatafromexistingplots, theNationalHurricaneCentersH*Winddataset,andtheUnitedStatesGeologicalSurveysNationalLand CoverDatabase.Thestatisticalmodelsestimatedtreedebrisbasedonalignmentofpre-andpost-storm data.Landcoverwastestedasaproxyvariableforpre-stormurbanforestbiomassandpost-stormtree debris.Estimatesofpre-storm,urbanforestbiomasswereestablishedtotestthestatisticalrelationship betweenpre-stormbiomassandpoststormdebris.Testingofbothlandcoverasaproxyvariableand thebiomass…debrisrelationshipwereperformedtosimplifyvolumetricestimatesofdebrisproduced throughstatisticalmodeling.Debrismodelswerespatiallyanalyzedtodeterminedebrisdistributionand tocomparewithexistingliteratureandavailablevegetationdebrisestimationmodels.Resultssuggest thaturbanforeststructurevariableshavegreaterin”uenceovervariationindebrisestimatesthanstormrelatedvariables. 2011 Elsevier B.V. All rights reserved. 1.Introduction Hurricaneimpactstoforestedlandscapesandhumansettlementsconsistofphysicaldamages,socio-economicburdens,and ecologicaldisturbance( EverhamandBrokaw,1996;Pielkeand Landsea,1998 ).Physicaldamagestoforestedlandscapesinclude treedefoliation,breakage,anduprootingofindividualtrees( Foster andBoose,1992;FrancisandGillespie,1993;Duryeaetal., 2007a,b ).Damagetoforestedlandscapescanresultinstand-levelor regionalimpactswhichresultinbothshort-andlong-termchanges inecologicalhealth,productivity,andstructureofforestedecosystems( FosterandBoose,1992;EverhamandBrokaw,1996;Stanturf etal.,2007 ).Landscape-scaledisturbancescanalsodetrimentally affectsocietieswith“nancialcostsduetodamagemitigationand implementationofstormrecoveryefforts( PielkeandLandsea, 1998;Stanturfetal.,2007;Staudhammeretal.,2009 ). Correspondingauthor.Currentaddress:DepartmentofBiologicalSciences,UniversityofAlabama,Tuscaloosa,AL35487,USA.Tel.:+12053481538; fax:+12053481786. E-mailaddresses: bthompson@u”.edu (B.K.Thompson), fescobed@u”.edu (F.J.Escobedo), cstaudhammer@ua.edu (C.L.Staudhammer), matyas@u”.edu (C.J.Matyas), yqui@u”.edu (Y.Qiu).1Tel.:+13528463503;fax:+13528461277.Resultingdamagestohumansettlementsandassociatedcosts ofhurricanescaneasilyexceedUS$1millionpercommunity,per storm( BurbanandAndresen,1994;PielkeandLandsea,1998 ). Furthermore,cleanup,mitigation,andrecoveryeffortsrequire largeinvestmentsinhumanresourcesandcapital( Umpierreand Margoles,2005 ).Forexample,asof2010,HurricaneKatrinawas themostcostlynaturaldisasterinthehistoryoftheUnitedStates, costingUS$81billion( OswaltandOswalt,2008 ).Treedebrisisa signi“cantportionoftotalhurricanerecoverycosts( Umpierreand Margoles,2005 ).Duringthe2004…2005hurricaneseasons,communitiesinthestateofFloridaintheUnitedStates(US)spent betweenUS$3000and$4millionperstormonclean-upanddisposalofvegetationdebris,withanaverageofUS$704,045per community,perstorm( Staudhammeretal.,2009 ). Manystudiesofhurricane-vegetationinteractionsarebased onasingleeventandaretypicallyreviewsofstormobservations, post-stormassessments,evaluationsofstormrecoveryprocesses, oracombinationoftheseresearchapproaches( Everhamand Brokaw,1996;Kupferetal.,2008;OswaltandOswalt,2008 ).The extentandseverityofhurricanedamagestoforestecosystems aredeterminedbyhurricanecharacteristicsincludingsize,speed, direction,maximumwindspeedsandwinddistributionpatterns ( Stanturfetal.,2007;Kupferetal.,2008;Escobedoetal.,2009 ), resultinginvariablepatternsandscalesofdamagesacrossland-0169-2046/$…seefrontmatter 2011 Elsevier B.V. All rights reserved. doi: 10.1016/j.landurbplan.2011.02.034

PAGE 2

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 2 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxxscapes( Tanneretal.,1991;FosterandBoose,1992;Stanturfetal., 2007 ).Areviewby Merryetal.(2009) listedtreespecies,proximity totheeyeofthehurricane,standandsitecharacteristics,speciesspeci“cresponsestostormsurges,andtopographicexposure astheprimarydriversofforestdamagefromtropicalcyclones inthesoutheasternUS.However,fewstudieshaveinvestigated landscape-scalehurricaneimpactstoforestsinurbanizedlandscapes( EverhamandBrokaw,1996;Duryeaetal.,2007a,b;Burley etal.,2008 ).Existingliteraturehasmostlydocumentedhurricane effectsonspeci“ctreesandindividualspeciesinurbanareas, butitremainsdif“culttocharacterizeurbanlandscapesbecause thevariationinindividualtreeandspecies-speci“churricane responsesatlargespatialscalesareunknown. Thereisagrowingscienti“cinterestintheecologicalfunctions andprocessesofurbanecosystems( Cadenassoetal.,2007;Nowak etal.,2002;Xiaoetal.,2004;Myeongetal.,2006;Staudhammer etal.,2011 ).However,therearefewstudiesthatmodellandscapescaletreedebriswithinurban,hurricane-affectedecosystemsof theSoutheasternUS.Urbanlandscape-scaleassessmentscanbe facilitatedbyGeographicInformationSystems(GIS)andused toanalyzeground-baseddatausingstatisticalmodeling,spatial analyses,andmodelingthroughcombinationsofrelatedspatial datasets( Salesetal.,2007;Zengetal.,2007;Blackardetal.,2008; HuandWang,2008 ).Studiesalsoshowthatclassi“cationofland coverisaneffectivewaytorepresenttheecologicalcomplexity andvariabilityofurbanlandscapes( Anderson,1977;Ridd,1995; Cadenassoetal.,2007;Buyantuyevetal.,2007 ).Assuch,temporal andspatialdistributionsofurbanforeststructurearein”uenced byland-useandlandcoverchange( Ridd,1995;Nowaketal.,1996; HuandWang,2008;Xiaoetal.,2004 ). Debrisestimationmethodsareproblematicduetothecomplexityandvariationofdamage-dependentvariablesacrosslandscapes ( Stanturfetal.,2007 ),butseveralmodelshavebeendeveloped toestimatehurricane-causedtreedebris( Table1 ).Thehurricane moduleoftheHazardsUSMulti-Hazard(HAZUS-MH)modelestimatesdebrisbasedontreedensityandtreeheightdata,andalso incorporatesspatialhurricanewinddatafromtheNationalHurricaneCenters(NHC)H*windModel.Treedamageestimatesare basedonastudyofwind-downed,naturalforesttrees,assuming alltreesrespondtowinduniformlyandthatonlytreesgreater than30feetinheightwillgeneratedebris( FederalEmergency ManagementAgency[FEMA],2006 ).InVirginiaandNorthCarolina,thismodelover-estimatedtreedebrisbyroughly90%and 41%,respectively( FEMA,2006 ). TheHurricaneDebrisEstimatingModel(HDEM;UnitedStates ArmyCorpsofEngineers)isbasedondatacollectedfromHurricanesFrederic(1979),Hugo(1989),andAndrew(1992).Model inputvaluesincludehousingdensity,vegetationcover,densityof commercialdevelopment,precipitation,stormcategoryaccording totheSaf“r…Simpsonscale,andwindspeed( FEMA,2007 ).GeographicInformationSystemsareusedinthismodeltoestimate windspeedsderivedfrommeteorologicaldataanddebrisestimateshaveareportedaccuracyof 30%( COES,2005;FEMA,2007 ). TheHurDETmodel( UmpierreandMargoles,2005 )isaGIS-based decisionsupporttoolthatestimatesdebrisfrombuildingandconstruction,vegetation,andsedimentfromstormsurgeeffects.The modelincorporatesspatialinformationfromlandcoverclassi“cations,categorizingdebrisaccordingtoassumeddebrisvolumes pertreeandproportionsoftreecoverinhardwoodtreesversus palms.Resultsofthismodelvarydependingonassumedratesof impactontreecoveraccordingtoSaf“r…Simpsonstormcategories. Resultsfrom Escobedoetal.(2009) wereusedtodeveloptheHurricaneAdaptationintheSTORMutilityofthei-Treeurbanforest managementsoftwarepackage(Version3.0).Thismodelisbased onFEMAtreedebrisdatafromthe2004…2005Floridahurricane seasons( Escobedoetal.,2009 ),andincorporatestreedensityand canopycover,developedurbancover,andsustainedwindspeed data.Duetoitsrecentdevelopment,ithasnotyetbeentestedwith actualhurricaneevents. Existingdebrisstudiesandmodels,withtheexceptionof Escobedoetal.(2009) ,areempiricalinnatureandhavenotbeen testedaspredictivetools.Allexistingmodelsrelyonspatialdata (e.g.landcoverorspatiallymodeledwinddata)togeneratevolumetricdebrisestimates;however,noneproducespatialestimates ofdebris.Additionalscienti“cresearchatthelandscapelevelis neededtoguidemanagementoftreesandforestsforhurricane effectsatvaryingspatialscales( Stanturfetal.,2007 ).Modelingof spatiallyexplicitpost-hurricaneurbantreedebriscouldimprove ef“ciencyofcommunityresponsetohurricanesthroughidenti“cationofequipmentstagingsites,locationofdebriscollection centers,moreappropriateprocurementofcontractors,enlistment ofadditionalhumanandlaborresources,andestimationofcosts associatedwithclean-upanddisposalofhurricanedebris( Burban andAndresen,1994;PielkeandLandsea,1998;Balsillie,2002; UmpierreandMargoles,2005;Stanturfetal.,2007 ). Theobjectiveofthisstudywastodevelopamodelthatcanbe usedtoestimatehurricane-causedtreedebrisinothercitiesin theSoutheasternUS.Speci“cally,datafromtheCityofHouston, TXwereusedtodevelopamodelwhichcanbeappliedtospatiallyanalyzeandexploreestimatesofpost-stormtreedebris.First, wedevelopedastatisticaldebrismodelutilizinggeo-referenced groundandstormdataasmodelinputstoestimatetreedebrisas afunctionofpre-stormurbanforeststructurevariables,spatially explicitstormcharacteristics,andland-cover.Wethenspeci“cally testedthehypothesesthat:(1)pre-stormtreebiomassispositively andsigni“cantlycorrelatedwithpost-stormtreedebris,and(2) landcoverisastatisticallysigni“cantproxyvariableforcharacterizingpre-stormtreebiomassandpost-stormtreedebris.Finally, weveri“edourmodelusinggeostatisticalanalyses,otheravailable debrisestimationmodels,andpost-hurricanedebrisinformation providedbytheCityofHouston. 2.Materialsandmethods 2.1.Studysite Thestudyareaencompassedtheeightcountiessurrounding theHouston,Texasmetropolitanarea( Fig.1 ),roughlycentered at294548northand952148west.Houstonhasanaverage annualtemperaturebetween15and20C(60…70F)andannual precipitationaverages1020…1530mm(40…60)( Bailey,1995 ).The westerntwo-thirdsofthestudyareaareonthePrairiepark-Table1 Summaryofexistinghurricanevegetationdebrisestimationmodels. ModelInputs Modelaccuracy HazardsU.S.Multi-Hazard(HAZUS-MH)Treedensity,treeheight,spatialhurricanewinddata41…90% TheHurricaneDebrisEstimatingModel(HDEM)Housingdensity,vegetationcover,densityofcommercial development,precipitation,stormcategory,windspeed 30% HurDETDebrisvolumespertree,proportionsoftreecoverin hardwoodsversuspalms,stormcategory Variedbystorm category FloridaHurricaneadaptationofi-TreeSTORMRoadmiles,debrisrate,treeanddebrisremovalcostsNotavailable

PAGE 3

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx 3 Fig.1. TheHouston,Texasstudyarea,relativestrengthofHurricaneIkeatlandfall,andlocationof2008re-measurementplots.NotedBZindicatesDecibel sofZorradar echointensityandre”ectivity.landprovinceecoregionandarecharacterizedbygentlyrolling, ”atplains,prairies,andoaksavannas.TheSouthernmixedforestprovinceecoregioncomprisestheeasternthirdofthestudy areaandischaracterizedby”at,coastalplainsandpineevergreenandbroadleafdeciduousforests( Bailey,1995 ).Therapidly urbanizingmetropolitanareahada2006populationof5.5million ( www.census.gov ; Campbell,1997 ). ThetropicalcyclonereportcompiledbytheNHC,showsthat HurricaneIkemadelandfallonGalvestonIsland,Texasat0700 CoordinatedUniversalTime(UTC;2:10AMCentralDaylightTime) September13,2008asastrongcategory2hurricane(onthe Saf“r…SimpsonHurricaneScale)with49msŠ 1(95kt)windsanda minimumcentralpressureof950hPa( NationalHurricaneCenter, 2008 ).The0300UTCforecastadvisoryissuedbytheNHCstated thathurricane-forcewindsextendedoutwardnearly200kmfrom theeye,whosediameterwasestimatedat74km.Acomparison ofthese“gureswiththesizeclimatologyforAtlanticbasintropicalcyclonescompliedby KimballandMulekar(2004) placesIkein thetop10%oflargesttropicalcyclonesintheAtlanticbasin.Thus, althoughnotclassi“edasamajorhurricane,Ikewasunusuallylarge insizesothathurricane-forcewindsaffectedalargerregionthan wouldbeexpectedforacategory2hurricane.Stormsurgeheights topped15feetalongtheBolivarPeninsulaandthe$US8.9billionin damagecausedbyIkeranksitasthefourthcostliestUShurricane ( CutterandSmith,2009 ). 2.2.Datacollection Permanentplotdatacollectedin2001providedameasureof pre-stormstructuredataontheHoustonareasregionalforests. The332plotsweresystematicallylocated,andcomprised0.067ha (0.167acre)circularplots( Nowaketal.,2005 ).Aspartofaremote sensingand“eldassessmentofHoustonsregionalurbanforests, theTexasForestServicedividedtheeight-countystudyareaintoa systematicgrid,andfourrandompointswerelocatedwithineach squareofthegridforamoreef“cientsamplingdesign(personal communication,M.Merritt,TexasForestService,2009).Inurban areas,threeofthefourpointswererandomlyselectedasplotcenters;insuburbanareas,twoofthefourpointswererandomly selectedasplotcenters;andinrangeland/agriculturalareas,one ofthefourpointswasrandomlyselectedasaplotcenter(personal communication,M.Merritt,TexasForestService,2009).Thefollowingtree-levelandplot-levelvariableswerecollectedfortreeswith diameteratbreastheight(DBH)greaterthan5inches(12.7cm): € Plotlocationdata:geographiccoordinatesofplotcenter,street address,anddistanceanddirectionofplotcenterrelativetopermanentreferenceobjectssuchas“rehydrants,streetlights,or cornersofbuildingsforplotre-locationpurposes; € Percentplotsurfacecovers; € Treemeasurements:species,numberofstems,DBH(measuredat 1.37m,or4.5feet),height,crowndimensions,%missingcrown, %crowndieback,crownlightexposure,distanceanddirection fromplotcenter; € Distanceanddirectionoftreesrelativetobuildingswithin18m following Nowaketal.(2002) The2001datawereusedintheUrbanForestEffects (UFORE/ECO)modeltosummarizeurbanforeststructureandfunctionvariables,including:averagetreesize(DBH;cmandheight;m), percentcrowndiebackasaproxyforcondition,leafarea(m2),leaf biomass(kg),andcarbonstorage(kg; Nowaketal.,2002 ). Nowak etal.(2002) providesalistofthetree-speci“callometricequations usedinthemodeltoestimatetreebiomassandcarbonstorage.

PAGE 4

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 4 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxxAportionofthepermanentplotswerere-visitedin October…November2008,approximatelyonemonthafterHurricaneIkemadelandfall,tomeasurepost-stormurbanforest damage.Astrati“edrandomsubsetofthesepermanentplotsthat containedtrees(approximately10%oftheoriginalplotsmeasured) wereselectedforre-measurementusingtwocriteria.First,we selectedplotswhereatleastonetreetallerthan6mwaslocated nomorethan15mfromabuilding.Thisresultedinasubsetof plotsclusteredwithinHoustoncitylimits,locatedwestandon theleftsideofthestormtrack( Fig.1 ).Asecondtieraimedata betterrepresentationofthegeographicaldistributionofplotsin relationtothepathofthehurricanewasobtainedbyselecting plotseastofthestormtrackoroutsidetheHoustoncitylimitswith atleastonetreetallerthan6m.Plotswithaccessorsafetyissues andthoseseverelydamagedbyhurricanestormsurgeeffects werenotre-measured.Threeplotsthatunderwentrecentmajor land-usechangebetween2001and2008werealsoexcludedfrom theanalysis.Thisprocessresultedinthirtyfourpost-hurricane plotsbeinganalyzed( Fig.1 ).Elevenofthesethirtyfourplotswere withinthecitylimitsofHouston,Texas. Post-hurricanedatacollectedincludedre-measurementoftreeandplot-levelvariablescollectedintheoriginal2001Houston UFOREstudyplusadditionalvolumetricmeasurementsofpoststormtreedebrisutilizingUSDAForestService…UrbanForestry SouthDebrisVolumeEstimationTables(personalcommunication, D.Hartel,USDAForestService,2008).IndividualUFOREtreecarbon storageestimateswereconvertedtogreenweightbiomassassuming50%carbonand50%moisturecontent( Nowaketal.,2002 ). Individualtreebiomassvaluesfrom2001and2008,andvolumetricmeasurementsofpost-stormtreedebrisweresummedbyplot toobtainmathematicalestimatesofpre-andpost-stormstanding biomassandpost-stormdownedbiomassperplot.The2008data werematchedtreebytreewiththeoriginal2001datatoanalyze changes. TheH*Winddatausedforthisanalysisareproducedbythe NHCandcharacterizehurricanewindsatvaryingspatialscales. Estimatesareassignedtodiscretespatialgridlocationsacross landscapesbasedonanalysesofdatafromaircraft”yovers,ships, ground-basedtowers,andresearchbuoys,andareproducedevery 6hwhilestormsareoveropenoceanandevery3honcea stormmakeslandfall( Powelletal.,2004 ).TheH*Windvariables include:maximumsustainedsurfacewindspeed(inm/s,knots, andmph),winddirection(indegrees),windsteadiness(indexed valuebetweenzeroandone),andwindduration(min). Maximumsustainedsurfacewindspeedreferstowindsblowingataconsistentrateofspeedforaminimumof1min,ata distanceof10mfromthegroundsurface,andthedirectionvariablereportsdirectionofthemaximumsustainedwindsatthetime theywererecorded( PowellandHouston,1996 ).Steadinessvalues measureconsistencyofthemaximumsustainedwindsdirection forallrecordedwindspeeds,wherelowvaluesindicateincreased variationinwinddirectioncomparedtohighvalueswhichindicaterelativewindsteadiness( Dunionetal.,2003;Kupferetal., 2008 ).Durationreferstotheamountoftimemaximumwindswere blowingatthetimetheywererecorded;howeverdurationvalues areonlyassignedtoH*Winddatapointsformodeledwindspeeds greaterthantheminimumwindspeedthresholdfordesignation asacategory1hurricaneontheSaf“r…Simpsonscale(e.g.,119km perhour,or74mph).Attheconclusionofeachstorm,awindswath datasetisproducedthatcontainsmaximumvaluesforeachpoint ofthegridrecordedthroughoutthestormsduration( Powelland Houston,1996 ). TheUnitedStatesGeologicalSurveys(USGS)2001National LandCoverDataset(NLCD)wasusedtolinklandcovertoposthurricanetreedebris.Classi“cationsarerepresentedby30m2pixelsdevelopedusingmulti-seasonalLandsat5andLandsat7 satelliteimagery,USDAForestServiceForestInventoryandAnalysis(FIA)data,NationalAgriculturalStatisticsServicecropland data,andancillarydataincludingmaps,orthoimagery,andthe 1992USGSNationalLandCoverDataset( Homeretal.,2008 ).Five of16landcoverclassi“cationswithintheeight-countyresearch areaweresampledforthisstudy:DevelopedHighIntensity(DH), DevelopedMediumIntensity(DM),DevelopedLowIntensity(DL), DevelopedOpenSpace(DO)andWoodyWetlands(WW).These landcovers,representingthelandareawithinHoustonCityLimits,werefurthercondensed,resultingintheeliminationoftheDH category,whichwasrepresentedbyonlyfourplots.Tofacilitate statisticalanalyses,thesefourplots,locatedoncommercialand industrialareas,wereassignedtotheDMcategory. SpatialdataforCountyandCityboundaries( www.glo. state.tx.us )weredisplayedinAlbersprojectionusingNorthAmericanDatum1927.TheUFORE/ECOvariablesweresummarizedat theplotlevelandoverlaidwithadministrativeboundaries,H*Wind data,andlandcoverdata.Theothervariables,suchasthemaximum windspeedandlandcovertypewereassignedbasedonproximitytore-measuredUFOREplotsandadistance-to-trackŽvariable wasderivedfromthelayerofpointsrepresentingthebesttrackof HurricaneIkeascontainedwithintheof“cialHurricaneDatabase (HURDAT; NationalHurricaneCenter,2009 ). Theplot-levelvariablesincludedinthemodelweretotalnumberoftrees,averageDBH,averageheight,totalcarbonstorage,fresh weightbiomass,andpercenttreecover;the2008re-measurement alsoincludedplot-leveldebrisestimates(incubicyards).Other UFOREvariablessuchastreecrowndimensionsandtypeandpercentageofsurfacecoverswereexcludedfromanalysisduetoa lackofevidenceintheliteraturedemonstratinglandscape-scale relationshipsbetweenthesevariablesandeitherbiomassordamageanddebrisestimation,especiallyinurbanlandscapes( Everham andBrokaw,1996;Salesetal.,2007;Duryeaetal.,2007b;Kupfer etal.,2008 ).Treeconditionandpercentageofmissingfoliage, thoughshowntobeimportanttopredictdamageinpreviousstudies( Duryeaetal.,2007b;Kupferetal.,2008 ),werenotincludedin theanalysisduetothedif“cultyinscalingthesevariablesuptothe plotlevel,thescaleatwhichtreedebriswasmeasured.Additionally,changestourbanforestbiomassduringthesevenyearperiod betweenpre-andpost-stormmeasurementswerenotaccounted for,andtheeffectsofurbanization,treegrowth,pruning,mortality,andplantingsthatoccurredbetween2001and2008couldhave affectedourestimates. ThewinddirectionvariablefromtheH*Winddatawasincluded inthemodelintwoforms.Becausedirectionmeasurementswere con“nedtotherange135…285,weincludedthisvariableinan un-transformedformat.Inrecognitionofthecircularnatureofthis variable,wealsobrokethisvariableintoitscomponentpartsfor east…west(cosine)andnorth…south(sine)directions.Thisvariable, however,wasultimatelydroppedsinceitstransferabilitytoother citiesandstormsisproblematic.Signi“canceassociatedwithwind directionisspeci“callyrelatedtotheuniqueangleanddirectionof HurricaneIkestrackinrelationtotheCityofHoustonsorientationtotheGulfcoastline.Thustheresultsofananalysisincluding directionwouldlimittheapplicationofthisstudysresults.The winddurationvariablefromtheH*Winddatasetwasnotconsideredinanymodelduetomissingvalues.Durationvaluesareonly assignedtoH*Winddatapointswithwindspeedsgreaterthan theSaf“r…Simpsoncategory1thresholdof119kph( Powelletal., 2004 ),andtheirinclusionwouldhaveeliminatednineplotsfrom ouranalysis. 2.3.Dataanalysis StatisticalanalyseswereperformedusingtheproceduresPROC CORRandPROCGLM(SASversion9.2).Analyseswereperformed

PAGE 5

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx 5 Table2 DescriptionofdatasourcesandvariablesusedinurbanforesteffectsandhurricaneanalysesforHouston,Texas(USA)andHurricaneIke. AnalysisDatasourceVariablesused Predictivemodelofdebris 2001HoustonUFORE( n =34)Plotlocation,plot-levelaveragetreeDBHandheight, treebiomass,treedensity,%treecover 2008HoustonUFORE( n =34)Treedebris(CY) NOAAH*WindMaximumwindspeed,windsteadiness,directionof maximumwindspeed USGSNationalLandCoverDataset(NLCD)Landcoverclassi“cation HURDAT( NationalHurricaneCenter,2009 )Hurricanetrack(distancetotrackderivedvia GeographicalInformationSystem) Plot-speci“cvolumetricestimatesofdebris 2001UFORE( n =332)Landcover,treeDBHandheight NOAAH*WindMaximumwindspeed USGSNationalLandCoverDataset(NLCD)Landcoverclassi“cation InterpolatedvolumetricestimatesofdebrisPlot-speci“cvolumetricestimatesofdebrisUSGSNationalLandCoverDataset(NLCD) ComparisontoFEMAFEMAprojectworksheet(PW)recordsTotaldebrisremoved ComparisontoU.S.ArmyCorpsofEngineersHurricane DebrisEstimationmodel UnitedStates2000CensusdataNumberofhouseholds Saf“r…SimpsonscaleratingDebrismultiplier FEMA(2007) Vegetationandcommercial/industrialdensity multipliers NHCprecipitationrecordsStormprecipitationmultiplier Comparisontoi-TreeSTORM CityofHoustontransportationGISdataTotalnumberofstreetmiles Saf“r…SimpsonscaleratingDebrisrate UFORE,UrbanForestEffects;NOAA,NationalOceanicandAtmosphericAdministration;DBH,diameteratbreastheight;USGSUnitedStatesGeological Survey;FEMA,Federal EmergencyManagementAgency;NHC,NationalHurricaneCenter.usingEnglishunitsratherthanmetricunitsbecausethedependent variable,debris,iscommonlyreportedandanalyzedbyemergency managementagenciesintheUnitedStatesinunitsofcubicyards ( FEMA,2006,2007 ),andthus,modelveri“cationisfacilitatedwith theseunits.Spatialanalyseswereperformedwiththesoftware packageArcGIS9.3andtheSpatialAnalystextension. Exploratorydataanalysesincludedthecalculationofdescriptive statisticsandtwo-waycorrelations.Sinceinitialanalysesindicated thatdebrisdatawerepositivelyskewed,analyseswereperformed usingthenaturallogtransformationofdebrisvalues.Statistical modelswereestimatedviaastepwiseregression,anddiagnostic testswereperformedtoensuremodelassumptionsofnormality andheteroskedasticityofresidualsweremetandtoidentifypotentialoutlyingobservations.Modelswerecomparedbyanalyzing residuals,aswellaswithgoodnessof“tmeasuressuchas R2and RootMeanSquareError(RMSE)values.Wheresigni“cantdifferenceswerefound( =0.10),meanseparationtestswereperformed viaScheffestest. 2.4.Veri“cationandcomparisonofresults Developingstatisticalmodelsandtheuseofspatialanalyses providestheopportunityforseveraldifferenttypesofanalyses. First,therelationshipbetweenhurricanedebrisandhurricane variables,pre-stormurbanforeststructure,andlandcovercan beanalyzedwithre-measuredplot-leveldata( n =34plots)to developastatisticalmodelofdebris.Second,geostatisticalinterpolationsof2001pre-stormbiomassandvolumesand2008 post-stormdebriscanbecomparedwithinHoustoncitylimitsto qualitativelyunderstandbiomass…debrisrelationships( n =11of 34re-measuredplots).Third,thestatisticalmodelcanbeapplied to2001,pre-stormUFORE/ECOpermanentplotdatatoestimate potentialpost-hurricanedebrisacrossthestudyarea( n =332permanentplots)andwithincitylimits( n =48of332plots).Finally, ourstatisticalmodelofpost-HurricaneIkedebriscanbeveri“ed usingotheravailabledebrisestimationmodelsandactualpostHurricaneIkedebrisinformationprovidedbytheCityofHouston. Asummaryofdatasourcesandvariablesusedforeachanalysisis foundin Table2 Allspatialinterpolationswere“rstmadeusingpre-andpoststormdatafromthe34plotsre-measuredin2008.Predictedvalues forallplotswerespatiallyinterpolatedbyordinarykrigingassumingaGaussiansemi-variogrammodel.Thecellsizeoftheoutput rasterwassetto49.3msothateachpredictedcellwouldmatch the0.067ha(0.167acre)sizeoftheUFOREplotsfromwhichthe predictedvalueswereinterpolated.Rastercellvalueswithinthe HoustonCityLimitswerethenextractedusingthecitylimitslayer. Cellswereconvertedtopointssothatanattributetableofdebris valuescouldbeexportedtocalculatedebrisacrossthelandscape. Themodelwasbasedonthefourdistinctlandcoverclassi“cations withintheHoustoncitylimits.Debrisestimatesforun-sampled landcoverswereassignedasfollows:(1)un-sampledlandcovershavingmeasuredUFOREplotvaluesforaveragetreeDBH andheightwereassignedthemeanvalueofsampledlandcover parametervalues;or(2)un-sampledlandcoverswithoutmeasured UFOREplotvaluesforaverageDBHandheight(i.e.,UFOREplots withouttrees)wereassignedadebrisvalueofzero. Theseresultswereveri“edbycomparingthemtoprojectworksheet(PW)recordssubmittedtoFEMAbytheCityofHouston. ResultswerealsocomparedtooutputfromtheUSArmyCorps ofEngineersHDEMmodeland Escobedoetal.s(2009) resultsas incorporatedintoi-TreeSTORM.TheHDEMrequiresthenumber ofhouseholdsasaninputandmultipliervaluesforthefollowing:debris(incubicyards)asindexedbySaf“r…Simpsonstorm category,vegetationasindexedbydensity,commercial/industrial developmentasindexedbydensity,andstormprecipitationas indexedbyrainfallcharacteristics( FEMA,2007;NationalHurricane Center,2008 ).Thei-TreeSTORMtoolrequirestotalnumberofstreet milesandauser-de“neddesignationofthestormeventsdebris generatingcapacityreferredtoasdebrisrateŽ.Thetotalnumber ofstreetmilesinHoustonwasdeterminedtobe26,597according totransportationGISdataobtainedfromtheCityofHouston.The user-de“neddebrisratewassettothedefaultvalueofmediumŽ basedonthedesignationofHurricaneIkeasastrongcategory2 storm. 3.Results 3.1.Correlationanalysis Two-waycorrelationsrevealedthatwindsteadiness( r =0.676), directionofmaximumwindspeed( r = Š 0.457)anditscosine ( r =0.547),biomass( r =0.453),andnumberoftreesperacre ( r =0.419)hadthehighestabsolutecorrelationswithtreedebris

PAGE 6

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 6 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx( Table3 ).Otherhighcorrelationswererecordedfornumberoftrees peracrewithbiomassand%treecover,biomasswith%treecover andheight,andmaximumwindspeedwithdirectionofmaximum windspeedaswellasitscosineandsinecomponents.Because thedirectionofmaximumwindspeedvaluesoccurredmostlyin anarrowrange(225…265),thisvariablewashighlycorrelated withitscosineandsinevalues.Theextremelyhighcorrelation betweenwindsteadinessandbiomass( r =0.818)seemstoindicate thatareaswithgreatertreebiomasswerelocatedontheedgeof thewindswath(wheresteadinessvalueswouldbehigher),rather thantowardsthemiddleofthestormtrack(wheresteadinessvalueswouldbeverylow).Thus,windsofamoreconsistentdirection movedthroughareasofgreatestbiomass. 3.2.Debrismodels Threestatisticalmodelsofdebrisweredevelopedandare referredtohereafterastheFull,Final,andAlternateModel.The FullModelincludesallvariablesselectedforanalysisincluding landcover(categoricalvariable),distancetothehurricanetrack, numberoftrees,treebiomass,treediameter,treeheight,%tree cover,maximumwindspeed,windsteadiness,andthecosine andsinecomponentsofthedirectionofmaximumwindspeed. Toavoidmulticollinearity,theunmodi“eddirectionofmaximum windspeedwasnotincluded.TheFinalModelwasderivedfrom theFullModel,selectingonlythosepredictorvariablesthatwere signi“cantpredictorsofdebrisorcontributedtomodelstability. Thevariablesassociatedwithdirectionofmaximumwindspeed werealsodroppedduetoitslackoftransferabilitytoothercities andhurricanes.TheAlternateModelisacompetingmodelwhich usesvariablesfromeachofthedatasetstoexplainvariabilityof debris. Landcoverandtreeheightwerethemostsigni“cantvariables explainingvariationindebrisinthefullmodel( Table4 ).Thefull modelhadanRMSEof1.57andthe R2valueindicatedthatthe modelexplained64%ofthevariabilityinlogofdebris.The“nal modelcontainedjustfourvariables,withlandcoverandtreeheight beingthemostimportant( Table4 ).Althoughdistancetotrack wasnotsigni“cant( P =0.15)inthe“nalmodel,itwaskeptinthe modelasitstabilizedmodelresidualsintermsofhomoscedasticity. Despitethehighcorrelationsbetweendebrisandwindsteadiness, biomass,andnumberoftreesperacre,noneoftheseexplanatory variableswasincludedinthe“nalmodel.Theindependentvariablesinthe“nalmodelexplained9%less(53%)ofthevariation inlogofdebris,withaslightlyhigherRMSE(1.74)versusthatof theFullModel.Thealternatemodelincludedexplanatoryvariables forlandcover,windspeed,DBHandheight,althoughwindspeed wasnotsigni“cant( P =0.583; Table4 ).Theindependentvariables explained49%ofthevariationinlogofdebris,andhadaslightly largerRMSE(1.80)versusthatoftheothertwomodels.Itshouldbe notedthatwhenwindsteadinesswassubstitutedforwindspeed inthealternatemodel,theRMSEimprovedslightlyto1.76and 51%ofthevariationinlogofdebriswasexplainedbythismodel; whenwinddirection(intheformoftheeast…westcosinecomponent)wasaddedtothemodel,theRMSEimprovedto1.57and R2increasedto63%.However,sincewindsteadinessanddirectionof maximumwindspeedaremoredif“culttoaccuratelypredictprior tostormlandfall,weincludedonlywindspeedinthealternate model. Theresidualanalysesdidnotindicateanyproblemsassociated withoutliers;however,allthreemodelsshowedheteroskedasticity,whichwascausedbyobservationsofzerodebris.Alternate transformationofthedatadidnotmitigatethisprobleminanyof thethreemodels.Whilethisheteroskedasticitywillcausetestsof signi“cancetobesomewhatsensitive,predictionsfromthismodel arestillexpectedtobeunbiased( Neteretal.,1996 ).Table3 Two-waycorrelationsbetweenquantitativemodelvariables( n =34)forHouston,Texas(USA)andHurricaneIke. Debris#oftreesacreTreeDBHTreeheightDistanceto track Treebiomass%TreecoverMaximum windspeed SteadinessDirectionof maximum windspeed Cosineof directionof maximum windspeed #oftreesperac0.419*TreeDBH Š 0.088 Š 0.259 Treeheight0.1580.406*0.571*Distancetotrack0.1150.0520.0430.216 Treebiomass0.453*0.767*0.1810.666*0.177 %Treecover0.2880.789*Š 0.0080.527*0.2360.818*Maximumwindspeed0.2170.11 Š 0.088 Š 0.0260.0080.1590.006 Windsteadiness0.676*0.2590.0090.2140.1630.401*0.33 Š 0.202 Directionofmaximumwindspeed Š 0.457*Š 0.0780.003 Š 0.1140.019 Š 0.2100.019 Š 0.767*0.013 Cosineofdirectionofmaximumwindspeed0.547*0.1960.0170.247 Š 0.0120.3600.1320.597*0.192 Š 0.947*Sineofdirectionofmaximumwindspeed Š 0.1200.1090.0250.0890.0820.0700.245 Š 0.807*0.4740.864*Š 0.678* DBH,diameteratbreastheight.*Signi“canceat P <0.05level.

PAGE 7

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx 7 Table4 Analysisofvariance(ANOVA)forthreepost-hurricanedebrismodelsusingdatafromUFOREandUSGSinHouston,Texas(USA)andHurricaneIke(HURDAT) SourceDFFullmodelFinalmodelAlternatemodel F valuePr> FF valuePr> FF valuePr> F Landcover32.710.070*5.360.005*3.950.019*Distancetotrack11.060.3152.200.150 Numberoftrees10.320.580 TreeDBH11.160.2933.590.069*3.210.084*Treeheight13.00.097*6.100.020*4.590.041*Treecover10.380.543 Treebiomass10.200.661 Maximumwindspeed10.770.3880.310.583 Steadiness12.780.110 Cosineofdirectionofmaximumwindspeed16.860.015*Sineofdirectionofmaximumwindspeed10.970.333 DBH,diameteratbreastheight.*Signi“canceat P <0.10level.Partial R2valuesrangingfrom30%to40%indicatedthatland covercontributedsubstantiallyinexplainingvariationindebris.In boththe“nalandthealternatemodel,meanseparationtestsindicatedthatDM(DevelopedMediumIntensity)andDL(Developed LowIntensity)werenotstatisticallydifferent.TheDO(Developed OpenSpace)landcoverwas,however,statisticallydifferentfrom DMandDL,andWW(WoodyWetlands)wasstatisticallydifferent fromallotherlandcovers. Tofurtherexploretherelationshipbetweenpre-stormbiomass andpost-stormdebris,pre-stormbiomassvalueswereconverted frompoundsperacretocubicyardsperacreusingUSDAForestServicedebrisvolumeestimationtables.Then,usingdata fromthe11plotsre-measuredin2008thatwereinsideHoustonCitylimits,the2001pre-stormstandingtreebiomass/volume and2008post-hurricanetreedebriswereseparatelyinterpolated ( Figs.2and3 ,respectively).Thisqualitativeandvisualcomparisonindicatesthatareasofhighertreebiomassnorthandeastof Houstonwerethesameregionsthatexperiencedgreaterdebris. Thepositivecorrelationbetweenpre-stormtreebiomassandpoststormtreedebrisisfurtherillustratedin Table5 ,whichshowsthe calculateddifferencebetweenpre-stormbiomassandpost-storm debris. 3.3.Developmentofspatiallyexplicitdebrisestimates TheAlternateModelwasappliedtoall332permanentplots measuredin2001toestimatepotentialdebrisacrossthestudy area.Parameterestimatesgeneratedforeachlandcovercannot beappliedtountestedlandcoversandthereforeparameterestimatesfromsampledlandcoverswereaveragedandappliedto un-sampledlandcoverswithplotdata.Un-sampledlandcoverswithoutassociatedUFOREdata(i.e.,UFOREplotswithout trees)wereassumedtohavezerodebris.Thisanalysisresulted inanaverageof15.28cubicyardsofdebrisperacrepotentially Fig.2. Interpolationmapusingordinarykrigingofpre-storm,standingtreebiomassincubicyards/acre( n =34)forHouston,Texas(USA).

PAGE 8

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 8 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx Fig.3. Interpolatedmapusingordinarykrigingofmeasuredpost-storm,downedtreedebrisincubicyardsperacre( n =34)forHouston,Texas(USA). Table5 Interpolatedcomparisonof2001pre-stormbiomassand2008post-stormdebrisbylandcoverclassi“cationwithintheCityofHouston,Texas(USA)fol lowingHurricaneIke. Landcoverclassi“cation2001pre-storm biomass-convertedtovolume (avg.cy/ac) 2008post-stormmodeled debris(avg.cy/ac) %Post-stormdowneddebrisby landcover(post-storm debris/pre-stormbiomass) DevelopedLowIntensity80.218.2410 DevelopedMediumandHighIntensity39.475.1113 DevelopedOpenSpace75.133.565 WoodyWetlands137.72169.42123aOther(averagedvaluesfromsampledland coverclassi“cations) 83.1346.5956 aIncreasedvalueslikelyindicatetheeffectoftreebiomassincreasesduringtheanalysisperiod.beinggeneratedacrossthestudyareafollowingastormsimilartoHurricaneIke( Fig.4 ).Multiplyingbythetotalacreage ofHouston,TX,(407,465acres,or164,895ha)yielded6,227,691 totalcubicyardsofpotentialpost-stormdebrisgeneratedwithin citylimits.Anadditionalanalysisusing48oftheoriginal2001 plotslocatedwithinHoustoncitylimitsresultedinanaverageof14.05cubicyardsofdebrisperacre.Multipliedbythe acreageinHouston,thisresultedinanestimateof5,726,587cubic yardsoftreedebris.Incomparison,PWinformationprovided bytheCityofHoustonreported6,116,000cubicyardsofdebris removedwithintheHoustonCityLimits,withanestimatethat 75%ofthattotal,or4,587,000cubicyardswasstrictlyvegetation debris(personalcommunication,V.Ayres,CityofHouston,2009; Table6 ).Table6 ModeleddebrisestimatesbasedonalternatemodelandcomparisonwithothermodelestimatesanddebrisvolumesreportedonPWdatafromtheCityofHou ston,Texas (USA). DebrisassessmentwithinCityofHoustonAveragedebris (cy/acre) Totaldebris(1000cy)ComparisontoPWdata (vegetationandmixed)aComparisontoPWdata (vegetation)a Projectworksheetdebris(mixedandvegetation)15.016116.000N/A Projectworksheetdebris(vegetation)11.264587.00N/A0 Modeleddebris(statistical)14.055726.43 Š 6%+25% Modeleddebrisusingspatialinterpolation15.286227.34+2%+36% HDEMb8.583494.60 Š 43% Š 24% i-TreeSTORMc15.276221.00+2%+36% aComparisonsrepresentthe%differenceofmodeleddebrisestimatesfromPWdata.bTheHurricaneDebrisEstimationModel(HURDEM; FEMA,2007 )withthefollowingvaluesasmodelinputs:#ofhouseholds=718,000;debrismultiplie r=8cy;vegetation multiplier=1.3;commercial/industrialmultiplier=1.2;precipitationmultiplier=1.3.ci-TreeSTORMwiththefollowingvaluesasmodelinputs:total#ofstreetmiles=26,597;debrisrate=mediumŽ.

PAGE 9

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx 9 Fig.4. Interpolatedmapusingordinarykrigingofpost-stormtreedebrisestimatesbasedonthisstudysAlternatemodelpredictions( n =332)forHouston,Texas(USA).4.Discussion OurFullModelexplainedthelargestpercentageofdebris variation;however,becauseindependentvariablesoverlappedin explainingvariability,theFullModelerrorwashigherthanthose oftheothertwomodels.ThecomparativelylowerRMSEvalues fromboththeFinalandAlternateModelsindicatedlowerlevels oferrorassociatedwithestimatesfromthesemodels.Moreover, thesemodelswerefarmoreparsimoniousandindicatedgood “t. TheAlternateModelhadalower R2andahighererrorcomparedtotheFinalModel.However,theAlternateModelmaybe moreappropriateforestimatingdebrisbasedonitsinclusionof windspeedasavariableinplaceofthedistancetotrackvariableoftheFinalModel.Whiledistancetohurricanetrackandthe east…westcomponent(cosine)ofwinddirectionweresigni“cantin theFinalModel,theirutilityisquestionableduetoHurricaneIkes uniquemorphologyandlargeradiusofmaximumwinds( National HurricaneCenter,2008 ).Windspeed,asmodeledthroughoutthe temporalcourseofhurricaneevents,maythereforebebettersuited forinclusionasanindependent,storm-speci“cvariableinapredictivemodel. Acomparisonofstatisticalandspatialdebrisestimatesusingthe AlternateModelprojectedfrom2001permanentplotdataandPW debrisinformationprovidedbythecityindicatesthatourmodel over-predictedtotaldebrisgeneratedasaresultofHurricaneIkeby 2…27%acrossthestudyarea( n =332plots)andby7…20%insidethe citylimits( n =48plots).AccuracyoftheHoustonPWdatamight besuspect(personalcommunication,V.Ayres,CityofHouston, 2009)sincedebrisremovedinHoustoncouldhavebeenalteredby homeownersblendingyarddebriswithhurricanedebris,debris remaininguncollectedinnaturalareasordif“cult-to-accesssites, orprivateremovalofdebris( FEMA,2007 ).However,theperceived accuracyofthespatialestimateisencouragingbecauseitlends validitytoaccuracyofmodeleddebrislocationinadditiontodebris volume.TheAlternatemodelsresultsareinlinewithPWrecordsof treedebrissincedatausedtodeveloptheAlternateModelarebased onpost-stormconditionsresultingfromtheimpactsofHurricane Ike. Wealsocomparedotheravailabledebrisestimationmodels withPWinformation.Estimateddebrisfromi-TreeSTORMdiffered fromthetotaldebrisrecordedonPWsby2%.However,unlikethe AlternateModel,theSTORMmodelwasdevelopedwithPWand weatherdatacollectedfromthe2004…2005hurricaneseasoninthe stateofFlorida.Therefore,itispossiblethatthedatausedtodevelop the Escobedoetal.(2009) modelaremoresimilarto,ormorerepresentativeoftheconditionsresponsibleforhurricane-causedtree debrisgenerationinHouston. TheHDEMunderestimateddebriscomparedwithPWdataand estimatesfromi-TreeSTORMandtheAlternateModel.Thismight bearesultofusingthedebrisvaluemultiplierofeightcubicyards whichisassignedbasedonthedesignationofHurricaneIkeasa

PAGE 10

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 10 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxxSaf“r…Simpsoncategorytwohurricane( NationalHurricaneCenter, 2008 ).However,HurricaneIkewasastrongcategorytwostorm onthevergeofthecategorythreewindspeedthresholdatlandfall.ThustherelativestrengthofHurricaneIkeasacategorytwo stormcouldhavegeneratedgreaterdebristhanwasestimatedby theHDEMusingeightcubicyardsasthedebrisratemultiplier. HurricanewindvariablesderivedfromtheH*windmodelmade onlyminorcontributionstoexplanationsofdebrisvariation.Wind isoftenassumedtobetheprimarydriverofdebrisgeneration, yetthewindspeedvariableisoftenreportedasapoorpredictor ofdebris( Kupferetal.,2008;Staudhammeretal.,2009 ).Other studiesofhurricane-forcewindsonforestedecosystemssupport “ndingsthatstorm-relatedvariablesexplainlessvariationofdamagethannon-stormvariables( Stanturfetal.,2007;Kupferetal., 2008;Staudhammeretal.,2009 )andindicatethatinteractions betweenforestdamageandhurricanewindsarecomplexanddif“culttoidentifystatistically.Hurricanevariablessuchassteadiness andwindspeedexplainedevenlessvariabilityindebrisestimates whencomparedtoforeststructurevariables.Onepossibleexplanationisthattreefailurerelatedtotreedecayorpoorstructure mightoccuratwindspeedsfarbelowmaximumvaluesmodeledby H*Wind,andthereforethemodelmaynotcapturethisrelationship usingextremevaluesforwindasmodelinputs. Althoughtherewasasigni“cantpositivecorrelationbetween pre-stormurbanforestbiomassandpost-stormdebrisresults, biomasswasnotasigni“cantpredictorofdebrisinanyofthe models.Thisislikelyduetothecomplexmultivariaterelationshipbetweenotherexplanatoryvariablesincludedinthemodel; biomasswasalsosigni“cantlycorrelatedwithseveralothervariables(%treecover,height,numberoftreesperacre).Thepositive correlationimplies,however,thatfuturedevelopmentofaspatialdebrisestimationmodelsmaybeabletoutilizemeasuresof pre-stormbiomasstopartitionpost-stormdebrisestimatesasa percentageoftreecrownandstembiomass. Landcoverwasresponsibleforroughly40%ofthevariabilityof debrisestimatedbythemodels.Moreover,thereweresigni“cant differencesintheamountofdebrisgeneratedbylandcover,inparticularwithrespecttowoodywetlands.HurricaneIkegenerated 3…9cubicyardsperacre(5…13%oftotalbiomass)inthedeveloped landcovers,whereas164cubicyardsofdebrisperacrerepresenting 123%ofthetotalbiomassperacreweregeneratedinwoodywetlands.Debrisestimatesexceedingthetotalbiomassbysuchalarge amountseemimprobableandarelikelyaconsequenceofsampling protocolsandgreaterannualbiomassincreasesinwoodywetlands areasrelativetootherlandcovers.However,landcoverwasan importantvariableinrelatingurbanforestpre-stormbiomassand post-stormdebris,andwarrantsfutureresearch. Resultsobtainedbyapplyingthemodeldevelopedinthisstudy toexistingUFORE/ECOplotdataarecorroboratedbyotherliteratureandpost-hurricanedebrisestimates( FEMA,2006;Escobedo etal.,2009;Staudhammeretal.,2009 ).Aqualitativecomparisonof interpolatedmapsofmeasuredbiomassanddebris( Figs.2and3 ) revealedthatareasofgreaterpre-stormbiomasstothenorthand eastoftheCityofHoustonalsoyieldthegreatestmeasuresofpoststormdebris.Theseresultsarecompatiblewiththeinitial2001 HoustonUFOREstudy,wherenorthernandsouthernregionswere separatedbasedonecologicaldifferencesbetweenthetype,percentage,anddensityoftreeandforestcoverwithinthoseregions ( Nowaketal.,2005 ).Thenorthernregionismoreruralandhas greaterforestdensity,whereasthesouthernregionismoreurbanizedwithlowertreedensityandpercentcanopycover( Nowak etal.,2005 ).Similarly,greaterdebriswasmeasuredrightofthe hurricanetrack( Fig.4 )wheregreaterwindspeedsaregenerally found,inthenorthernhemisphere,asthevelocityofstormmotion isaddedtothevelocityofthetangentialwindsontherightsideof thestorm. MeantreeheightandDBHwerethestrongestexplanatory variablesfordebrisvariationcomparedwithothertreeplot-level measurementsasobservedbyotherstudies( EverhamandBrokaw, 1996;Kupferetal.,2008;OswaltandOswalt,2008 ).However,contrarytootherstudies( Escobedoetal.,2009 ),measuresofforest density(%treecoverandnumberoftrees)showedlittlecontributiontoexplanationsofdebrisvariation.Onepossiblereasonforthis isthatdebrisvariationexplainedbymeasuresofforestdensitymay havebeenovershadowedbythelandcovervariable.Our“ndings withrespecttohurricanevariablesweresimilartothoseof Kupfer etal.(2008) ,whofoundasigni“cantrelationshiptodistanceto trackandthattheH*Windvariablesexplainedverylittlevariation ofdebris.Thisresult,however,maybearesultoftheuniquemorphologyofHurricaneIke.TheeyeofHurricaneIkewasestimatedat 74kmindiameter( NationalHurricaneCenter,2008 ),inthetop90% ofeyesizesaccordingtotheclimatologyof KimballandMulekar (2004) .Thislargediameterextendedtheradiusofdamagingwinds andmayhavebeenresponsibleforpatternsofdamageobserved. Inthisstudy,non-stormvariables(i.e.,urbanforeststructure)weremoreimportantpredictorsofdebristhanwerestorm variables,a“ndingsupportedbyboth Kupferetal.(2008) and Staudhammeretal.(2009) .This“ndingisalsoinlinewith Duryea etal.(2007a) ,whosuggestedthathurricane-causedtreedamagecanbeattributedtotreesbeingimproperlylocated,poorly maintainedorofaspecieswithlowtolerancetohurricaneforce winds.Implicationsarethaturbanforestmanagersmightbeableto proactivelymitigatetheproductionofpost-hurricanetreedebris. Hurricanescannotbeprevented,whereasurbanforeststructure andcompositioncouldbealteredusingmanagementstrategiesand effectiveordinancesandpolicies. Limitationsofthisstudyarethatpost-stormdatacollection tookplaceapproximatelytwenty-“vedaysafterlandfallofHurricaneIke.Duringthistimeframe,muchofthedebrismighthave beenmovedorremovedfromplotsduringpost-hurricanecleanupanddisposaloperations.Asaresult,lessvolumeofdebrisare likelytohavebeenrecordedinmoreurbanizedplotsinthestudy area.Anotherlimitationisthesmallsamplesizeofpost-stormmeasurementswhichwerelimitedbysafetyandaccessissues.When samplesizesaresmall,interactionsbetweendebris-dependent variablescannotbeadequatelyquanti“ed.Finally,thereisaseven yeargapbetweenpre-andpost-stormdata,andthereforechanges tourbanforestconditionandstructure(e.g.,urbanization,tree growth,pruning,mortality,planting,and/orremoval)thatoccurred between2001and2008cannotbeaccountedfor.However,aspart ofaparallelstudyusingthesesameplotdata, Staudhammeretal. (2011) foundalossofonly0.24squaremetersofbasalareaperyear andanoverallnetchangeinnumberoflivetreesof2.9%duringthis 2001…2008analysisperiod. 5.Conclusion Ourstudyshowedthaturbanforeststructurevariables explainedgreatervariationintreedebristhanstormrelatedvariables,implyingthattreefailureandresultingdebrisaremore closelyassociatedwithtree-orsite-speci“cfactorsthanmeteorologicalvariables.Findingsmightsuggestthatmosttreefailures occurredatwindsspeedsbelowthemodeledmaximumvalues offeredbytheH*Winddataset,andthereforethedatadidnotaccuratelycapturethefullextentofrelationshipsbetweenwindspeed andhurricane-causedtreedebrisgeneration.Theseresultswere corroboratedbysimilar“ndingsfromrelatedstudiesanddebris volumesreportedbytheCityofHouston.Wedid“ndasigni“cantrelationshipbetweentreebiomassanddebris;however,this relationshipwasmaskedwhenincludingotherforestattribute variables.Inparticular,landcoverwasanimportantvariablein

PAGE 11

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx 11explainingthepatternsofdebrisgeneratedacrossthestudyarea. WhiletheHDEMproducedarelativelypoorestimateofdebris inHouston,TX,outputfromthei-TreeSTORMtoolandestimated treedebrisfromourAlternateModelwerewithinarelativelysmall marginoferrorcomparedtoPWdata.Thus,thismodelcouldbe appliedtoexistingUFORE/ECOplotandlandcoverdatatoobtain post-hurricanedebrisestimates.However,accuracyoftheHoustonPWdatausedtoverifyourresultsmightbesuspectdueto homeowneralterations.Moreover,bothmodelsshouldbefurther veri“edusingdatafromfuturepost-stormassessmentsinother hurricane-affectedcites.Inparticular,landcoverisameasureofthe amountofforestcoverinHouston;however,inadifferenturban area,wherethepopulationincludedurbanareaswithverydifferentlevelsofurbanforestcover,thisvariablemightnotbeasgood apredictor. Modelaccuracyandprecisioncanbeimprovedbyusingadditionaldatasuchaspost-stormdebrismeasurementsfromfuture hurricaneeventsand“nerscalerepresentationsofbothwind behaviorandlandcovervariation.Still,thispreliminarystudyis astartingpointforfurtherresearchcombiningurbanforeststructuredatathatiscommonlycollectedincitieswithotherspatially explicitdatasetstopredictpatternsofdebrisinurbanlandscapes. Moreover,futureanalyseswithlargerdatasetscouldincorporate othergeographicvariables(e.g.,soiltype,topography,groundsurfacecover),stormvariables(e.g.,rainfall,stormsize,forwardstorm speed,eyewalldiameter),and/ortreeandforestvariables(e.g.rootingspace,treecondition,wooddensityandelasticity,andtree crowndensityandcondition).Althoughspecies-speci“cvariables mightbedif“culttoextrapolatetovaryingspatialscales,results fromsuchanalysescouldproducemoreaccuratedebrisestimates. Futureworkcouldalsotakeadvantageofthefullsuiteof H*WindsdatasincetheonlyH*Windsdatautilizedinthisstudy werethedatafromtheentirestormsswath.Modelsthatutilize the3-hourlyH*WindwouldprovideaseriesofsnapshotsŽcharacterizingthestormasitmovedoverthestudyregion,thereby accountingmoreaccuratelyforthesizeoftheeye,theradiusof maximumwinds,andtheradiusofhurricane-forcewinds.Doppler radarcouldalsobeutilizedtodeterminethesizeoftheeyeasthe stormmovesinlandandlocationsreceivingthemaximumwinds tofacilitatecomparisonwiththemodel-derivedH*Winddata. Inaddition,modelingahurricanespotentialtoalterurban foreststhroughtreecanopycoverortreedensitychangesmaybe worthpursuingassuchindicatorshavebeenshowntohavesigni“cantrelationshipswithhurricanedamageatlandscapescales ( EverhamandBrokaw,1996;Stanturfetal.,2007;Escobedoetal., 2009 ).Managementcanresultinreducingtheriskofhurricane damagetotrees,asstandandstructureareimportantvariablesin predictingwinddamage( Merryetal.,2009 ).Resultsofthisproject provideurbanforestmanagerswithadebrisestimationmodelthat canbeimplementedandevaluatedagainstfuturehurricaneevents. Itisourhopethatthisstudyprovidesfutureresearchersandurban forestmanagerswithgreaterinsightsintohurricaneresponsesof urbanforestsandrelatedfactorswhichdrivehurricane-causedtree debrisgenerationinurbanecosystems. Acknowledgments TheauthorswouldliketothanktheUSDAForestServiceand SouthernGroupofStateForestersforfundingforthisproject.We arealsogratefultoMickieMerrittandPeteSmithoftheTexasForestServiceforuseoftheUFOREdataand“eldwork;MaryDuryea, DudleyHartel,EdMacie,andDavidNowakforhelpfulsuggestions;SebastianVarela,ErickSmith,EricKuehlerforpost-storm “eldwork;andVictorAyresatCityofHoustonforsupplyingFEMA projectworksheetdebrisdata. ReferencesAnderson,J.R.,1977.Land-useandlandcoverchanges:aframeworkformonitoring. J.Res.U.S.Geol.Surv.5,143…153. Bailey,R.G.,1995.DescriptionoftheEcoregionsoftheUnitedStates,2nded.(map). MiscellaneousPublicationNo.1391.U.S.DepartmentofAgriculture,U.S.Forest Service,Washington,DC. Balsillie,J.H.,2002.Expedientassessmentsofcoastalstormandhurricanedamage potential.Environ.Geosci.9(3),102…108. Blackard,J.A.,Finco,M.V.,Helmer,E.H.,etal.,2008.MappingU.S.ForestBiomass usingNationwideForestInventoryDataandModerateResolutionInformation. RemoteSens.Environ.112,1658…1677. Burban,L.L.,Andresen,J.W.,1994.StormsovertheUrbanForest:Planning,Responding,andRegreening„ACommunityGuidetoNaturalDisasterRelief,2nded. U.S.D.A.ForestService,NortheasternArea,152pp. Burley,S.,Robinson,S.L.,Lundholm,J.T.,2008.Post-hurricanevegetationrecovery inanurbanforest.LandscapeUrbanPlan.85,111…122. Buyantuyev,A.,Wu,J.,Gries,C.,2007.EstimatingvegetationcoverinanurbanenvironmentbasedonLandsatETM+imagery:acasestudyinPhoenix,USA.Int.J. RemoteSens.28,269…291. Cadenasso,M.L.,Pickett,S.T.A.,Schwarz,K.,2007.Spatialheterogeneityinurban ecosystems:reconceptualizinglandcoverandaframeworkforclassi“cation. Front.Ecol.Environ.5,80…88. Campbell,P.,1997.CurrentPopulationReports,PopulationProjections1995…2025. U.S.CensusBureau,P25-1131,6pp. COES(CaliforniaOf“ceofEmergencyServices),2005.DisasterDebrisManagement TrainingManual.GovernorsOf“ceofEmergencyServices,StateofCalifornia,p. 118. Cutter,S.L.,Smith,M.M.,2009.Fleeingfromthehurricaneswrath:evacuationand thetwoAmericas.Environment51,26…36. Dunion,J.P.,Landsea,C.W.,Houston,S.H.,Powell,M.D.,2003.Areanalysisofthe surfacewindsforHurricaneDonnaof1960.Mon.WeatherRev.131,1992…2011. Duryea,M.L.,Kampf,E.,Littell,R.C.,2007a.Hurricanesandtheurbanforest1:effects onsoutheasternUnitedStatescoastalplaintreespecies.J.Arboriculture33, 83…97. Duryea,M.L.,Kampf,E.,Littell,R.C.,Rodriguez-Pedraza,C.D.,2007b.Hurricanesand theurbanforestII:effectsontropicalandsub-tropicaltreespecies.J.Arboriculture33,98…112. Escobedo,F.J.,Luley,C.J.,Bond,J.,etal.,2009.HurricanedebrisanddamageassessmentforFloridaurbanforests.ArboricultureUrbanForestry35,100…106. Everham,E.M.,Brokaw,N.V.L.,1996.Forestdamageandrecoveryfromcatastrophic wind.Bot.Rev.62,113…185. FEMA(FederalEmergencyManagementAgency),2006.HAZUSMHMR3Hurricane ModelTechnicalManual.DepartmentofHomelandSecurityMitigationDivision, Washington,DC. www.fema.gov/library/viewRecord.do?id=3034 (accessed 25.5.08). FEMA,2007.PublicAssistanceDebrisManagementGuide.FEMA-325,Washington, DC,260pp. Foster,D.R.,Boose,E.R.,1992.Patternsofforestdamageresultingfromcatastrophic windincentralNewEngland,USA.J.Ecol.80,79…98. Francis,J.K.,Gillespie,A.J.R.,1993.RelatinggustspeedtotreedamageinHurricane Hugo1989.J.Arboriculture19,368…373. Homer,C.,Dewitz,J.,Fry,J.,etal.,2008.Changesinforestbiomasscarbonstoragein theSouthCarolinaPiedmontbetween1936and2005.ForestEcol.Manag.225, 1400…1408. Hu,H.,Wang,G.G.,2008.ChangesinforestbiomasscarbonstorageintheSouth CarolinaPiedmontbetween1936and2005.ForestEcol.Manag.225,1400…1408. Kimball,S.K.,Mulekar,M.S.,2004.A15-yearclimatologyofNorthAtlantictropical cyclones.PartI.Sizeparameters.J.Climate17,3555…3575. Kupfer,J.A.,Myers,A.T.,McLane,etal.,2008.Patternsofforestdamageinasouthern MississippilandscapecausedbyHurricaneKatrina.Ecosystems11,45…60. Merry,K.,Bettinger,P.,Hepinstall,J.,2009.Physicalandbiologicalresponsesof foreststotropicalcyclonesaffectingtheUnitedStatesAtlanticOceanandGulf ofMexicocoasts.Am.J.Environ.Sci.5(1),784…800. Myeong,S.,Nowak,D.J.,Duggin,M.J.,2006.Atemporalanalysisofurbanforest carbonstorageusingremotesensing.RemoteSens.Environ.101,277…282. NationalHurricaneCenter,2008.TropicalCycloneReport:HurricaneIke.National OceanicandAtmosphericAdministration,p.55. NationalHurricaneCenter,2009.Of“cialRecordofHurricanesandTropicalStormsintheAtlanticBasin.Availableat http://www.aoml.noaa.gov/ hrd/hurdat/easyread-2010.html (accessed18.1.11). Neter,J.,Kutner,M.H.,Wasserman,W.,Nachtsheim,C.J.,1996.AppliedLinearStatisticalModels,4thed.WCB/McGraw-Hill,Boston,MA,USA. Nowak,D.J.,Crane,D.E.,Stevens,J.C.,Ibarra,M.,2002.BrooklynsUrbanForest [GeneralTechnicalReportNE-290],USDAForestService.NortheasternForest ExperimentStation,Radnor,PA,USA. Nowak,D.J.,Rowntree,R.A.,McPherson,etal.,1996.Measuringandanalyzingurban treecover.LandscapeUrbanPlan.36,49…57. Nowak,D.J.,Smith,P.D.,Merritt,M.,etal.,2005.HoustonsRegionalForest.Texas ForestServiceCommunication/UrbanandCommunityForestry9/05-5000,24 pp. Oswalt,S.N.,Oswalt,C.M.,2008.Relationshipsbetweencommonforestmetricsand realizedimpactsofHurricaneKatrinaonforestresourcesinMississippi.Forest Ecol.Manag.225,1692…1700. Pielke,R.A.,Landsea,C.W.,1998.NormalizedhurricanedamagesintheUnited States:1925…95.WeatherForecast.13,621…631.

PAGE 12

Pleasecitethisarticleinpressas:Thompson,B.K.,etal.,Modelinghurricane-causedurbanforestdebrisinHouston,Texas.LandscapeUrban Plan.(2011),doi: 10.1016/j.landurbplan.2011.02.034 ARTICLE IN PRESSGModel LAND-2006;No.ofPages12 12 B.K.Thompsonetal./LandscapeandUrbanPlanning xxx (2011) xxx…xxx Powell,M.D.,Bowman,D.,Gilhousen,D.,etal.,2004.Tropicalcyclonewindsatlandfall:theASOS…C-MANWindExposureDocumentationProject.BAMS,B.Am. Meteorol.Soc.85(6),845…851. Powell,M.D.,Houston,S.H.,1996.HurricaneAndrewslandfallinsouthFlorida.2. Surfacewind“eldsandpotentialreal-timeapplications.WeatherForecast.11 (3),329…349. Ridd,M.K.,1995.ExploringaV-I-S(vegetation-impervioussurface-soil)modelfor urbanecosystemanalysisthroughremotesensing:comparativeanatomyfor cities.Int.J.RemoteSens.16,2165…2185. Sales,M.H.,Souza,C.M.,Kyriakidis,etal.,2007.Improvingspatialdistributionestimationofforestbiomasswithgeostatistics:acasestudyforRondonia,Brazil. Ecol.Model.205,221…230. Stanturf,J.A.,Goodrick,S.L.,Outcalt,K.W.,2007.Disturbanceandcoastalforests:a strategicapproachtoforestmanagementinhurricaneimpactzones.ForestEcol. Manag.250,119…135. Staudhammer,C.,Escobedo,F.J.,Luley,etal.,2009.Patternsofurbanforestdebris from2004and2005Floridahurricaneseasons.South.J.Appl.For.33,193…196. Staudhammer,C.,Escobedo,F.,Lawrence,A.,Duryea,M.,Smith,P.,Merritt,M.,2011. RapidassessmentofchangeandhurricaneimpactstoHoustonsurbanforest structure.ArboricultureUrbanForestry37(2),60…66. Tanner,E.V.J.,Kapos,V.,Healey,J.R.,1991.Hurricaneeffectsonforestecosystemsin theCaribbean.Biotropica23,513…521. Umpierre,D.,Margoles,G.,2005.BrowardCountysWeb-BasedHurricaneDebris EstimationTool(HurDET).BrowardCountyEmergencyManagementAgency. http://proceedings.esri.com/library/userconf/proc05/papers/pap2200.pdf (accessed25.1.11). Xiao,Q.,Ustin,S.L.,McPherson,E.G.,2004.UsingAVIRISdataandmultiplemaskingtechniquestomapurbanforesttreespecies.Int.J.RemoteSens.25, 5637…5654. Zeng,H.C.,Talkkari,A.,Peltola,H.,etal.,2007.AGIS-baseddecisionsupportsystemforriskassessmentofwinddamageinforestmanagement.Environ.Model. Softw.22,1240…1249.