ebook img

Network analysis of wildfire transmission and implications for risk governance PDF

28 Pages·2017·3.34 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Network analysis of wildfire transmission and implications for risk governance

RESEARCHARTICLE Network analysis of wildfire transmission and implications for risk governance AlanA.Ager1*,CodyR.Evers2,MichelleA.Day3,HaiganoushK.Preisler4,AnaM. G.Barros5,MaxNielsen-Pincus2 1 MissoulaFireSciencesLaboratory,RockyMountainResearchStation,ForestService,UnitedStates DepartmentofAgriculture,Missoula,Montana,UnitedStatesofAmerica,2 DepartmentofEnvironmental SciencesandManagement,PortlandStateUniversity,Portland,Oregon,UnitedStatesofAmerica, 3 DepartmentofForestEcosystems&Society,CollegeofForestry,OregonStateUniversity,Corvallis, Oregon,UnitedStatesofAmerica,4 PacificSouthwestResearchStation,ForestService,UnitedStates a1111111111 DepartmentofAgriculture,Albany,California,UnitedStatesofAmerica,5 DepartmentofForestEngineering, a1111111111 Resources&Management,CollegeofForestry,OregonStateUniversity,Corvallis,Oregon,UnitedStatesof a1111111111 America a1111111111 *[email protected] a1111111111 Abstract OPENACCESS Wecharacterizedwildfiretransmissionandexposurewithinamatrixoflargelandtenures (federal,state,andprivate)surrounding56communitieswithina3.3millionhafireprone Citation:AgerAA,EversCR,DayMA,PreislerHK, BarrosAMG,Nielsen-PincusM(2017)Network regionofcentralOregonUS.Wildfiresimulationandnetworkanalysiswereusedtoquantify analysisofwildfiretransmissionandimplications theexchangeoffireamonglandtenuresandcommunitiesandanalyzetherelativecontribu- forriskgovernance.PLoSONE12(3):e0172867. tionsofhumanversusnaturalignitionstowildfireexposure.Amongthelandtenuresexam- doi:10.1371/journal.pone.0172867 ined,theareaburnedbyincomingfiresaveraged57%ofthetotalburnedarea.Community Editor:AsimZia,UniversityofVermont,UNITED exposurefromincomingfiresignitedonsurroundinglandtenuresaccountedfor67%ofthe STATES totalareaburned.Thenumberoflandtenurescontributingwildfiretoindividualcommunities Received:September9,2016 andsurroundingwildlandurbaninterface(WUI)variedfrom3to20.Communityfiresheds, Accepted:February11,2017 i.e.theareawhereignitionscanspawnfiresthatcanburnintotheWUI,covered40%ofthe Published:March3,2017 landscape,andwere5.5timeslargerthanthecombinedareaofthecommunitycoreand WUI.Forthemajorlandtenureswithinthestudyarea,theamountofincomingversusoutgo- Copyright:Thisisanopenaccessarticle,freeofall copyright,andmaybefreelyreproduced, ingfirewasrelativelyconstant,withsomeexceptions.Thestudyprovidesamulti-scale distributed,transmitted,modified,builtupon,or characterizationofwildfirenetworkswithinalarge,mixedtenureandfirepronelandscape, otherwiseusedbyanyoneforanylawfulpurpose. andillustratestheconnectivityofriskbetweencommunitiesandthesurroundingwildlands. TheworkismadeavailableundertheCreative Weusethefindingstodiscusshowscalemismatchesinlocalwildfiregovernanceresult CommonsCC0publicdomaindedication. fromdisconnectedplanningsystemsanddisparatefiremanagementobjectivesamongthe DataAvailabilityStatement:Allwildfiresimulation largelandowners(federal,state,private)andlocalcommunities.Localandregionalrisk andtransmissiondataaresharedat ScholarsArchive@OSU,OregonStateUniversity’s planningprocessescanadoptourconceptsandmethodstobetterdefineandmapthescale repositoryforscholarlymaterials.Datamaybe ofwildfireriskfromlargefireeventsandincorporatewildfirenetworkandconnectivitycon- accessedat:https://doi.org/10.7267/N9G15XS6. ceptsintoriskassessments. Allotherdataarepubliclyavailablewithsources listedinthemanuscriptorsupplementalmaterial. Funding:Thisworkwassupportedbyfunding fromtheNationalScienceFoundation(http://www. nsf.gov),CoupledHumanandNaturalSystems Program(CHH-1013296),theUnitedStates DepartmentofAgriculture,ForestService,Pacific PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 1/28 Networkanalysisofwildfiretransmission NorthwestResearchStation(http://www.fs.fed.us/ Introduction pnw/)andtheJointFireScienceProgram(http:// AsfireeventsinthewesternUSandelsewherehavebecomeincreasinglylargeanddestructive, www.firescience.gov/)grant(14-1-01-22toAA). Thefundershadnoroleinstudydesign,data developinglong-termsocialandecologicalstrategiestomitigateimpactsremainsachallenge collectionandanalysis,decisiontopublish,or [1].Awiderangeofsolutionshavebeenadvocatedinrecentliterature,includingchanging preparationofthemanuscript. existingsuppressionpoliciestoallowmorenaturalignitionssothatwildfireseventually Competinginterests:Theauthorshavedeclared becomeself-limiting[2],improvingtheefficiencyofprogramstotreatandreducewildland thatnocompetinginterestsexist. fuels[3],changingmanagementcultureandattitudestowardsfire[4],andincorporatingnew analyticaltoolsforwildfirescenarioplanningtoimprovethecouplingofhumanandbiophysi- calsubsystems[1,5].Thewiderangingdiscussionsreinforcethefactthatmanaginglong-term riskfromlarge,highlyuncertainwildfireeventsisacomplexsocioecologicalproblemthatwill requirerapidadjustmentsinexistingriskgovernancesystemsinachangingclimate.Byrisk governance,wemeantheprocessbywhichauthorityisexercisedanddecisionsaretaken withinsocialandinstitutionalenvironmentstoidentify,assess,manage,andcommunicate risk[6,7].Existingriskgovernanceischallengedbythefine-graineddiversityoflandtenures andsocioecologicalsettings(e.g.fireregime,attitudetowardsfire)relativetothescaleoflarge fireevents[8,9]. Perspectivesonwildfiremanagementareoftendeeplydivisiveamongpublicandprivate entities,andaconsistentperceptionofriskimpedesthedevelopmentofsocialorganizations thatareneededtoaligngoalswithrespecttowildfiresandtheirimpacts[9,10].Forinstance, inthewesternUS,statewildfiremanagementagencieshaveapublicmandatetoprotectpri- vatelandsfromwildfire,whilefederalpolicyonadjacentfire-adaptedforestsencourages increaseduseofprescribedandmanagedwildfiretoreducefuelsandwildfirerisk[11].The increaseduseofwildandprescribedfireasafuelmanagementstrategyonpublicforests[12] inparticularhasandwillcontinuetogeneratefiremanagementconflictsamongstate,federal, andprivatelandjurisdictions.Anotherexampleisthemanagementofprivateindustrialtim- berlands,wherethelackoffinancialincentivestoremovesurfacefuelsgeneratedfromharvest- ingcanexacerbatewildfirerisktosurroundingparcels[13–15].Atthecommunityscale, residentslivingnearfire-adaptedpublicforestsmayrecognizethebenefitsfrompoliciesfor increaseduseofprescribedandnaturalfiretorestorefireresiliency,butobjecttosmokeand otherbyproductsthathavedeleteriouseffectsonairquality[14].Severalprescribedfireshave escapedcontrolmeasuresand,insomecases,resultedincatastrophiclossesandsubsequent tighteningoffiremanagementregulations[16–18].Riskgovernanceconflictsamonglarge landowners,publiclandagencies,andcommunitieswillbecomemoreacuteintheAmerican Westwhereurbanizationandamenityproductionincreasinglycompetewithtraditionalland uses.Atthesametime,individualwildfireeventsincreasinglyintersectparcelswithlandown- ershavingdiverseattitudes,firemanagementpoliciesandriskgovernancesystems,and changingcurrentfiremanagementpoliciesisslowedbysubstantial“institutionalmomentum” [14,15]. Theinterdependenceofriskbetweendifferentlandownersandparcelswithinpublicland isrecognizedinnewerUS“alllands”wildfirepolicy(e.g.USCohesiveWildfireManagement Strategy[11]),andmanagementinitiativesincreasinglyemphasizecollaborative,cross- boundarysolutionstofiremanagement[19,20].Herewedefinerelevantboundariesasthose delineatingnotonlylandownership,butalsothepotentialtomanageforestsandfuelsby mechanicalmeans(henceforthlandtenures).However,mappingthescaleofrisk,quantify- ingrisktransmissionandmeasuringwildfireconnectivityandinterdependenceacrossland tenuresarelargelyunexploreddomains.Thestudyofrisktransmissioniswidelydiscussedin thepropagationofinfectiousdiseaseinhumans,plants,andanimalpopulations[21]where, forinstance,oneorganismtransmitsinfectiousdiseasetoanother.Inthecaseofwildfire,the PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 2/28 Networkanalysisofwildfiretransmission originsandextentofthedisturbancearewelldefinedspatially(i.e.mappedwithinafew meters),comparedtoothernaturaldisturbances(e.g.earthquakesandhurricanes),andits propagationisaffectedbyfuelloadingonindividualparcels[22].Afewstudieshaveexam- inedwildfirerisktransmissionfrompubliclandstothewildlandurbaninterface[23–26],but havenotexaminedcommunityspecificwildfiretransmissiononlarge,mixed-tenureland- scapes.Forinstance,howiswildfirerisksharedandaretherehotpotsofrisktransmission associatedwithspecificlandtenuresandlocations? Tounderstandtheimportanceofrisktransmissionandhelpcreateassessmentmethods thatcanadvanceconceptsforcross-boundaryriskgovernance,westudiedtheexchangeof wildfireexposureamonglargelandtenures(federalmanaged,federalwilderness,private industrialforest,andstatelands),and56adjacentcommunitiesona3.3millionhafireprone areaofthewesternUS.Weusethetermexposureratherthanrisksincewelimitouranalyses toareaburnedratherthanpredictingexpectedloss[27].Sincemostareaburnedbylargewild- fireeventsisunderextremeburningconditions,wecanassumethatmeasuresofburnedarea translatetohighseveritywildfireandlossofvaluesonburnedparcels.Weusedsimulation modelingandnetworkanalysistoexaminefivespecificquestions:(1)towhatextentdolarge tractsoffederal,state,andprivatelandstransmitwildfireexposuretoeachother;(2)dosome landtenurestransmitmorewildfireexposurethantheyreceive;(3)howmanyuniqueland tenurescontributetothewildfireexposureoftypicalwesternUScommunities;(4)whatisthe relativeinfluenceofhumanversusnaturalignitionsinthetransmissionofwildfireexposure; and(5)aretheresubstantialdifferencesinfireexposureamongthedifferentlandtenuresas describedbyfireintensityandlikelihood? Ourstudysetsthestageforimprovingcross-boundarycommunity,collaborative,and nationalforestplanning[19,20,28,29],withanalysesthatcanhelpidentifyconflictsand opportunitiestoachievefederalwildlandfirepolicy,includingfireadaptedcommunities,fire resilientlandscapes,andwildfireresponse.Specificallywefillagapinlocal,regional,and state-scaleriskassessmentmethods[27,30,31]usedforcollaborativeplanningbyprovidinga frameworktoquantifycross-boundaryexposureanddefinethegeographicscaleofriskto communities.Thisinformationcanbeusedtominimizescalemismatchesbetweenlandten- uresandriskplanningatthecommunityandlandscapescales. Methods Studyarea Thestudyareawaspartofthe“Forests,People,Fire”project[5]andspans3.3millionhain centralOregon,alongtheeastslopeoftheCascaderange(Fig1A).Thephysiographicgradi- ents,forestvegetation,climate,andpublicmanagementpracticesregardingwildfiremanage- mentresemblethenationalforeststhroughoutthewesternUS,andaredescribedindetail elsewhere[5].ThestudyareaincludesthetownsofRedmond,BendandKlamathFallsand manyothersmallruralcommunities.Elevationrangesbetween500and3240mandthetopo- graphic-moisturegradientfromwesttoeasthasastronginfluenceonthedistributionof majorecologicalzones.Higherelevationregionstothewestaredominatedbymountainhem- lock(Tsuga mertensiana)andfirandsprucesubalpineforest,whilemidelevationcoolerslopes arecharacterizedbymoistmixedconiferpatchesofDouglas-fir(Pseudotsugamenziesii),grand fir(Abiesgrandis),whitefir(Abiesconcolor)sometimesintermingledwithlodgepolepine (Pinuscontorta),andponderosapine(Pinusponderosa)atlowerelevations.Highelevationpla- teausontheeastsidearedominatedbyponderosapinemixedwithsemi-aridjuniper(Juni- perusoccidentalis)woodlandsandaridshrublands.About30%ofthestudyareaisdrymixed conifer,whilearidareasandjuniperwoodlandsrepresent19%.Standsofmoistmixedconifer PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 3/28 Networkanalysisofwildfiretransmission Fig1.Studyareamapsoflandtenuresandcommunities.(A)Landtenures(definedbyownershipandmanagementpractice),towns,andmajor highways.SeeTable1forlandtenuredescriptions.(B)CommunitiesasdefinedbyUSCensusdataoncorecommunityareaandsurroundingSILVIS wildlandurbaninterface(WUI)polygons.Communityboundaries(lightgraypolygons)representthecombinationofthecorecommunityareaand associatedWUI. doi:10.1371/journal.pone.0172867.g001 forests,ponderosapineandlodgepolepineaccountfor9%,14%and5%ofthestudyarea, respectively.Alpineandhighelevationnon-forestedareasandnon-vegetatedareasrepresent 10%and13%oflanduse,respectively.Theareaisnotedforextensivecontiguousstandsof lowdensityponderosapineoldgrowththatweremaintainedhistoricallywithperiodicnatural fire[32].Manyoftheseforestshavedevelopeddenseunderstoryconditionsandladderfuels largelyduetopoliciesthatprioritizesuppressionofnaturalfires[28]andlimitharvestingto preserveoldforesthabitatforwildlifeandrecreationalamenities.Recentlargefiresincludethe B&BComplexin2003(36,733ha),PoleCreekin2012(10,844ha)andSunnySidein2013 (21,448ha).Thestudyareahasalsoexperiencedrapidpopulationgrowthandexpansionof urbaninterface/intermixaroundseveralcoreareasincludingBend,Redmond,Sisters,and KlamathFalls,contributingtoagrowingconflictbetweenlandmanagementobjectivesofpub- licforestsandamenityinterestsofnewerresidents. Landtenureandcommunitydata Themajorlandownershipsinthestudyareaarethefederalgovernment(66%,28%protected), ConfederatedTribesofWarmSprings(native,8%),andprivateindustrial(7%)(Fig1A, Table1).Privateindustriallandsaregenerallymanagedforwoodproductsandareharvested PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 4/28 Networkanalysisofwildfiretransmission Table1. Majorlandtenuresandtransmittedwildfireestimatedwithwildfiresimulation. Landtenure Description Area(100 Total Totalfire Annualareaburned(hayr-1) ha) (%) (hayr-1) Human Natural NonTF TF-IN TF-OUT NonTF TF-IN TF-OUT BLM BureauofLandManagement 2050 6 1377 233 406 410 283 456 465 Community Corecommunities+wildlandurbaninterface 4338 13 2580 496 905 826 352 828 698 NF-M Nationalforest,managed(e.g.generalforest, 9350 29 4559 1139 679 764 1791 950 1094 timberproduction) NF-P Nationalforest,protected(e.g.biodiversity 9280 28 3996 1056 709 777 1429 801 1085 reserves,wilderness) NPS NationalParkService 390 1 72 4 9 2 37 23 24 Native NativeAmerican(ConfederatedTribesof 2460 8 6706 3738 525 487 2000 443 351 WarmSprings) ODF OregonDepartmentofForestry 180 1 66 5 36 21 4 21 16 OtherFED Otherfederalland(e.g.DepartmentofEnergy) 290 1 423 59 138 124 53 172 93 Private Non-industrialprivate 1450 4 829 70 337 304 50 371 223 Private Corporateforestsmanagedforwoodproducts, 2140 7 882 299 269 314 106 208 175 Industrial harvestedonregularrotation USFWS USFishandWildlifeService 300 1 129 25 37 34 19 47 57 Totalfire=nonTF+TF-IN.NonTF=theannualareaofnon-transmittedfire(self-burning).TF-IN=theannualareaburnedfromincomingfiresignitedon otherlandtenures.TF-OUT=theannualareaburnedonotherlandtenuresbyfiresignitedlocally.Additionaldescriptionsofthecalculationsfortransmitted fireareinthemethodssection. doi:10.1371/journal.pone.0172867.t001 onaregularrotationtomaximizerevenue.TheBureauofLandManagement(BLM)andnon- industrialprivateownersownroughly6%and4%ofthearearespectively.Thelatterarerela- tivelysmalltracksoflandrangingfrom5to100ha.Oregonstateforestedlandsaccountfor 1%ofthestudyarea.FederallandincludestheDeschutesandFremont-WinemaNationalFor- ests,CraterLakeNationalPark(NPS),BLMandareasmanagedbytheUSFishandWildlife Service(USFWS).Asmallportion(1%)ofthelandscapeincludesminorfederalownerssuch astheDepartmentofEnergyandDepartmentofDefense.Theownershipdatawerecombined withfederallandmanagementdatatocreatealandtenuredatasetusingdatasourcesfromthe “Forest,People,Fire”study[5].Nationalforestlandswerepartitionedbasedonlandand resourcemanagementplans(e.g.[33])intoprotectedareas(variousconservationandbiodi- versityreserves)versusthoseavailableformechanicalfueltreatments.Allotherfederalland withtheexceptionofnationalparklandwasconsideredavailableformechanicalfueltreat- ment.Theseparationofmanagedversusnon-managedwasthenincorporatedintotheland tenureclassification. Inadditiontothelandtenureslistedabove,wecreatedacommunityclassbasedoncore communitiesandsurroundingsparselypopulatedwildlandurbaninterface(WUI)areas.We definedcorecommunitieswithinthestudyareaaccordingtotheUSCensusdataoncommu- nityboundaries[34].Fifty-sixcommunitieswereidentifiedinthesedatarepresenting233,000 peopleand118,000structures.WethenattachedsurroundingSILVISwildlandurbaninterface (WUI)[35]polygonsoutsidethecorecommunityboundarybasedondistancetotheclosest communitycore.WeremovedSILVISpolygonsthatwere:(1)classifiedasuninhabited,(2) classifiedaswater,or(3)<0.1hainsize.Thecriteriaforremovingpolygonsconservedeven thelowestdensityWUIareasadjacenttonationalforests.AllotherWUIpolygonswere assignedtoacommunityandeachWUIpolygonwasattributedwithhousingunitdensity PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 5/28 Networkanalysisofwildfiretransmission (hereafterreferredtoasstructures)andarea(ha).Theresultingcommunitylayerconsistedof boththecoreareadefinedintheUSCensusandtheadjacentWUI(Fig1B). Wildfiresimulation Analyzingcross-boundaryexposurefromwildfiresrequiresspatialinformationonignition locationsandfireperimeters.Broadgeographicpatternsofcross-boundaryexposurehave beenpreviouslydescribedwithhistoricalfiredatabasesforlargeareas(e.g.westernUS,Fig8 inAgeretal.[36]).However,dataonhistoricallargefires,includingperimeters,areonlyavail- ableatbestforthepast20–30yearsformuchofUS,andthusareinsufficienttomapandpre- dictfuturepatternsofpotentialwildfiretransmissionatthescaleconsideredinthisstudy.A solutioniswildfiresimulation,whichisnowawidelyusedpracticeforbothtacticalandstrate- gicwildfiremanagementintheUSandelsewhere[37–40].MonteCarlosimulationsofmany fires(>100,000)usinghistoricalweatherandcalibratedmodelscanbeusedtogeneratesuffi- cientfiresamplestomapfinescaletransmissionamongparcelson>2millionhawithdesktop computers.MostpractitionersintheUSusetheminimumtraveltime(MTT)algorithm[41] thatcanreplicatelargefireperimetersandfireintensitypatterns[37,42–46].Weusedaver- sionofMTTencapsulatedinFConstMTT,acommandlineversionofFlamMap[41].Fire simulationsusedaspatiotemporalfirepredictionsystemusingthemethodsofPreisleretal. [47]anddescribedindetailelsewhere(S1Appendix).Thespatiotemporalignitionprediction modelwasbuiltusing18yearsofignitiondata[48]tocalculate:(1)theprobabilityofafireina givenpixelasafunctionofdailyenergyreleasecomponent(ERC),location(X,Ycoordinate) andday-of-year;and(2)theexpectedfiresizeforlocationswheretheabove-mentionedproba- bilityequalsoneasafunctionofdailyERCandlocation.Weseparatedhumancausedigni- tionsfromnaturalcausedignitions(S1Appendix). Weexecutedthespatiotemporalignitionmodeltogenerate3000fireseasonsforatotalof 63,736simulatedfires,sufficienttosaturatethelandscapewithwildfireandburneverypixel atleast10times.Thefirelistsgeneratedbythespatiotemporalignitionmodelusedby FConstMTTspecifyignitionlocation,day-of-year,year,cause,ERC,windspeed,winddirec- tion,fuelmoisture,burnperiodandexpectedfiresizeofeachfiresimulated. Burningconditionsassociatedwitheachignitionweresampledfromhistoricaldaysbased ondailyERC.Fuelmoistureforeachfuelsizeclassandfuelmodelwereaveragedoverhistori- calconditions(1987–2011)foreachvalueofERC,whilewindspeedanddirectionwerebased ondailyERCbutrestrictedtogustobservationsindaysofthehistoricalrecord(1994–2011) wherefiresexceeded500hainsize.AllweatherdatawereobtainedfromtheLavaButte RemoteAutomatedWeatherStation(RAWS,[49]).Weestimatedtheburnperiod(required byMTT)toachieveaspecificfiresizeusingarelationshipbetweenburntimeandfiresize builtfromcontrolledsimulations(S1Appendix).Simulatedwildfireperimetersboreaclose resemblancetohistoricalfireswithinthestudyarea[26].Examplelargefireperimetersfrom thesimulationsshowthepotentialimpactsintermsofareaandstructuresaffected(Fig2). WeobtainedsurfaceandcanopyfuelforwildfiresimulationsfromthenationalLANDFIRE dataset[50],aswellaselevation,slope,aspect,fuelmodel[51],canopycover,canopybase height,canopyheight,andcanopybulkdensity[52].LANDFIREisastandardizedfueldataset availablefortheconterminousUSandwidelyusedforwildfiremodelingandresearchinthe US[50,53]. FConstMTTsimulationswereperformedat90mundertheconditionsspecifiedinthefire- lists,andgenerated:(1)fireperimeters,(2)annualburnprobabilities,and(3)conditional burnprobabilitiesfor200.5mflamelengthcategories.Theburnprobability(BP)foragiven intensityclassistheratioofthenumberoftimesapixelburnedtothetotalnumberoffires PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 6/28 Networkanalysisofwildfiretransmission Fig2.Simulatedfireperimetersandignitionlocationsandassociatedlandtenuremap.Perimetersandignitionlocationsof(A)sixofthelargest firesbasedonareaburned,and(B)twoofthemostdestructivefiresbasedonexposuretothemostnumberofstructures.Perimeterswereextractedfrom over63,736simulatedfiressampledfromdistributionsmodeledfromhistoricalignitionlocationandfiresize.SeeTable1forlandtenuredescriptions. doi:10.1371/journal.pone.0172867.g002 simulatedandestimatestheannualburnprobabilityateachpixel.Theconditionalburnprob- abilitiesareoutputasagriddedpointfileandestimatetheprobabilityofafireatagiveninten- sity.Aprobabilityweightedaverageconditionalflamelength(CFL)wascalculatedfromthe conditionalburnprobabilitiestomeasuretheexpectedflamelengthgivenapixelburns. Wildfiretransmissionbetweenmajorlandtenures Wecalculatedtransmittedwildfireexposureusingmethodsdevelopedinourpreviousstudies [54,55].Specifically,fireperimeteroutputsandignitionlocationswereintersectedwiththe landtenureandcommunitymap(Fig1).Wethencross-tabulatedtotalareaburnedineach landclassandcommunitybyignitionsourcetoderivetheamountofincoming(TF-IN),out- going(TF-OUT),andnon-transmittedfire(NonTF).Theresultingdatawerethenusedto PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 7/28 Networkanalysisofwildfiretransmission buildtransmissionnetworkstoanalyzetheconnectivityamongspecificlandtenuresandinto communities.NetworkconstructionandanalysisweredoneinR[56]usingtheigraphpackage [57]. Networksarecomprisedofnodesandedges.Inthisstudynodescorrespondedtolandten- uresandcommunitieswhiletheedgesrepresenteddirectionaltransferoffire(i.e.directed weightednetwork).Nodesrepresentedmultiplepolygonsforagivenlanddesignation;hence theedgesmeasuretheaggregatetransmissionofallthepolygonsofonelandtenuretothe polygonsofotherlandtenures.Networkswereanalyzedattwo‘scales’:thescaleoflargeland tenuresand,thescaleofindividualcommunities.Weseparatedtransmissionoriginatingfrom humanandnaturalignitionswithineachnetwork.Abriefoverviewofnetworkterminology usedinthisanalysiscanbefoundinTable2. Wecalculatedwholenetworkandnode-specificmeasurespertainingtothefrequencyand strengthoflinkages.Thenode-degreemeasuresthenumberoflinkagesforeachnodeand indicateshowcentralanodeiswithinthenetworkandisofteninterpretedasanindicatorof connectivityandinfluence[58].Inthecaseofwildfire,nodedegreerepresentsthenumberof landownersthatcontributewildfiretoaspecificparcel,andtheoverallpotentialcomplexityof managingandresolvingcross-boundaryexposureissuesforparticularcommunitiesorparcels ofland.Networkdensitydescribesthenumberofnodespresentinthenetworkcomparedto themaximumpossible,whichrepresentstheoverallconnectednessinthenetwork.Wealso calculatedthenumberoflinkagestransmittingfireintothelandtenure(in-degree),andthe numberoflinkagesthelandtenurewastransmittingfiretoothernodes(out-degree).Node degreeisthetotalnumberoflinkagesforanode(landtenure),whilein-degreeandout-degree arethenumberofincomingoroutgoinglinkages.Networkswereconstructedandanalyzedat twoscalesofaggregation.Thecoarser-scalegroupingdescribedtransmissionbetweenlarger landownerswithcommunitiesaggregatedintoasinglelandtenure.Thefiner-scalegrouping includedthesamelargelandtenuresandindividualcommunities. Table2. Organizationofriskgovernanceandwildfiretransmissionintonetworks. Function Network Networkscale Riskgovernanceimplications representation Spatialdefinitionoflandscapeparcelin Node Node Defineslandscapeorganizationandthescaleformeasuringwildfire termsoflandtenure transmissionandsharedrisk Numberofwildfirelinkagesamongland Nodedegree Node Potentialcomplexityformanagingfireexposuretransmittedtoand tenures fromsurroundinglandtenures Wildfireexchangebetweenapairofland Directededge Node Magnitudeofrisksharingamongtwolandtenures tenures Numberofincomingwildfirelinkagesfora Nodein-degree Node Numberofsurroundinglandtenuresthatcontributetolocal landtenure exposure Numberofoutgoingwildfirelinkagesfora Nodeout-degree Node Numberofsurroundinglandtenuresthatreceivefirefromthelocal landtenure landtenure Overallwildfireconnectivityamongland Networkdensity Network Measurestheproportionofobservedwildfireconnectionsamong tenures landtenurestothetotalpossible Theimportanceofaparticularlandtenure Centrality Node/network Prioritylandtenureforriskmitigation inthenetwork Networkwidestatisticsdescribelandscape-scalepropertiesrelatedtowildfireconnectivityandsharingofwildfireexposureamonglandtenures. Communitiesareconsideredaspecificlandtenurecategoryinthetableforsimplicity.Nodestatisticsaredescriptorsofindividuallandtenureexposureto wildfireintermsofamount,numberofsources,andoverallconnectivitytootherlandtenures.Thecombinedanalysesofnodeandnetwork-widestatistics canfacilitatecommunityandlandscapeplanningbyrevealingthescaleandconnectivityofwildfiredisturbances. doi:10.1371/journal.pone.0172867.t002 PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 8/28 Networkanalysisofwildfiretransmission Housingexposureandcommunityfireshedcharacterization Communitywildfireexposureoccurredwheresimulatedfirescoincidedwithdeveloped areas(i.e.thecorecommunityandthesurroundingwildlandurbaninterface).Weinter- sectedfireperimetersgeneratedfromthefiresimulationswithcommunitypolygonsand thencalculatedthenumberofstructuresineachcommunitypotentiallyaffected.Thestruc- tureestimateswerecalculatedastheproductoftheproportionofareaofeachcommunity polygonburnedandthestructurecountforthatpolygon.Structuresweregroupedbycom- munity(n=56)allowingasingleignitionpointtocreatewildfireexposureinmultiple communities. Weusefireshedstoexplorethespatialscaleofcommunityfireexposurewithinthestudy area.Wedefinecommunityfireshedsastheareasthatarelikelytotransmitwildfiretocom- munities[36].Usingthesimulationoutputsdescribedabovewecreatedacontinuous smoothedsurfaceofstructureexposureforeachcommunityusingthelogistickrigingalgo- rithmintheRgstatpackageinR3.1.1[56,59].Krigingisbasedonasemi-variogramthatcap- turesobservedspatialdependencebetweenpoints[60,61].Sincethekrigingalgorithmis generalizableweusedittocaptureoccurrencedataasabinarylogitfunction. Thekrigingsurfacesdescribethepredictedlikelihoodthatanignitionoccurringatagiven pointwouldexposestructurestowildfirewithinthatcommunity(i.e.valuesvariedbetween0 and1).Toconvertthekrigingsurfacetoadiscreteboundary(i.e.thefireshed)weuseda thresholdof0.1meanttocaptureignitionlocationswheretherewasatleastanominal(10%) chanceofexposure.Toaccountforareasinthefireshedwhereignitionswererareorabsent weremovedareaswheretheignitionpointdensitywasverylow(i.e.lessthan1.5e-4km-2). Finally,communityfireshedswereintersectedwiththelandtenuremaptocalculatethe fireshedareabylandtenure,andtheproportionofthestudyareathattransmitsfiretothe WUI.Ourmethodfordelineatingfiresheds,ascomparedtoconvexhullsusedbyothers(e.g. [62,63]),eliminatesoutlierignitionsthatrarelythreatencommunitiesandcontributetolarger fireshedsizes. Results Historicalfireactivityandignitionpatterns Analysisofhistoricalfiredataprovidedinsightsintotheseasonalityandspatialdistributionof wildfiresandtherelativefrequenciesofhumanversusnaturalignitionsandresultingburned area.Theresultsconfirmedthatbothhumanandnaturalignitionscontributedsubstantiallyto areaburned,andthateachignitionsourcerequiredadifferentspatiotemporalmodelforthe wildfiresimulations.About16,061haofwildfireburnedwithinthestudyareaeachyearannu- allybasedon11,618ignitions(645ignitionsyr-1)occurringbetweenJanuary1992andDecem- ber2009(Fig3).Lightningresultedinroughlythesameburnedareaashumancausedfires (7,877haversus7,184hayr-1)althoughwithmoreignitions(6,379or354ignitionsyr-1versus 5,239,or291ignitionsyr-1).Lightningcausedignitions(hereafternatural)hadaslightly smalleraveragefiresize(22ha)comparedtohuman(26ha).Ignitionlocationsfromhumans andlightningdifferedsubstantially(Fig3).Humanignitionswereclusteredneartownsand citieswithinthestudyarea,especiallyWarmSprings,LaPineandBend.Naturalignitions werefrequentinthesouth-centralandsoutheastportionofthestudyareawherehumanigni- tionswererelativelysparse.Temporalvariationinignitionswasalsopronounced(Fig4).Nat- uralignitionspeakedinthemiddleofthesummermonths(Fig4C)whereashumanignitions wererelativelymorefrequentinthespringandfall. PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 9/28 Networkanalysisofwildfiretransmission Fig3.Spatialdistributionofhistoricalignitionsbyfirecause.(A)Humanand(B)naturalcausedignitionsinthestudyareabasedonhistoricalignition databetween1992and2009[64],totaling11,618ignitions. doi:10.1371/journal.pone.0172867.g003 Simulatedfireactivity Wesummarizedsimulatedfireactivitytoexaminebroadtrendsinrelationtohistoricalfires. Simulatedfiresburned21,650hayr-1:11,200hayr-1weretheresultofhumancausedignitions; 10,460hayr-1burnedduetonaturallycausedignitions.Firewasmostcommononnational forests(8,554ha)andnativelands(6,710ha),whichtogetheraccountedfor72%ofallland burned.Morethanone-thirdoftheareaburned(8,402ha,39%)resultedfromwildfireevents thatspannedatleastoneboundary.Humancausedfiresburnedcomparativelymoreareaof native,community,privateandprivateindustriallandtenures(Fig5A;Table1).Thisdiffer- encewasmostpronouncedfornativelandsandleastfortheprivatelandtenures.Firesfrom PLOSONE|DOI:10.1371/journal.pone.0172867 March3,2017 10/28

Description:
Barros AMG, Nielsen-Pincus M (2017) Network analytical tools for wildfire scenario planning to improve the coupling of human and biophysi- .. [54, 55]. Specifically, fire perimeter outputs and ignition locations were intersected with the land tenure and community map (Fig 1). We then cross-tabulat
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.