VCSMethodologyVM0029MethodologyforAvoidedForestDegradationthroughFireManagementVersion1.08May2015SectoralScope14VM0029,Version1.0SectoralScope14Page2Methodologydevelopedby:MpingoConservationandDevelopmentInitiative(MCDI),TanzaniaInpartnershipwith:UniversityofEdinburghUniversityCollegeLondonValueforNatureConsultingMethodologyfundedby:TheRoyalNorwegianEmbassy,TanzaniaVM0029,Version1.0SectoralScope14Page3TableofContents1Sources.........................................................................................................................42SummaryDescriptionoftheMethodology......................................................................43Definitions......................................................................................................................64ApplicabilityConditions..................................................................................................85ProjectBoundary...........................................................................................................95.1Spatialextent....................................................................................................95.2Upperandlowercarbondensityboundariesoftheprojectareas......................95.3Delineationofprojectboundaries......................................................................95.4Greenhousegasesandcarbonpools.............................................................106BaselineScenario........................................................................................................127Additionality.................................................................................................................148QuantificationofGHGEmissionReductionsandRemovals........................................148.1BaselineEmissions.........................................................................................148.2ProjectEmissions...........................................................................................248.3Leakage..........................................................................................................298.4NetGHGEmissionReductionsandRemovals...............................................329Monitoring....................................................................................................................339.1DataandParametersAvailableatValidation..................................................339.2DataandParametersMonitored.....................................................................439.3DescriptionoftheMonitoringPlan..................................................................5010References..................................................................................................................56Appendix1:GapFireDescription.........................................................................................58Appendix2:TheRothermelModel......................................................................................81VM0029,Version1.0SectoralScope14Page41SOURCESThismethodologyusesthelatestversionsofthefollowingtoolsandmodules:VT0001ToolfortheDemonstrationandAssessmentofAdditionalityinVCSAFOLUProjectActivitiesVMD0013Estimationofgreenhousegasemissionsfrombiomassandpeatburning(E–BPB)2SUMMARYDESCRIPTIONOFTHEMETHODOLOGYAdditionalityandCreditingMethodAdditionalityProjectMethodCreditingBaselineProjectMethodThismethodologyappliestoprojectsthatimplementpreventativeearlyburningactivitiesinmiombowoodlandsintheEasternMiomboecoregionofAfrica.AsspecifiedbytheWorldWildlifeFund1,theEasternMiomboecoregionconsistsofarelativelyunbrokenareacoveringtheinteriorregionsofsoutheasternTanzaniaandthenorthernhalfofMozambique,withafewpatchesextendingintosoutheasternMalawi.TheCentralZambezianMiomboWoodlandecoregionliesbeyondLakeMalawitothewest,whilenorth,theecoregionisborderedbyAcacia-CommiphoraBushlandandThicketbelongingtotheSomali-Masaiphytochorion(White1983).TheEastAfricancoastalmosaicofWhite’s(1983)Zanzibar-InhambaneRegionalCenterofEndemismlinestheshore.TheZambezianandMopaneWoodlandecoregionliestothesouth.ThisecoregionisseparatedfromothermiomboecoregionsinthatitismostlyconfinedtolowerelevationsoftheEastAfricanPlateau,andisdominatedbythefloristicallyimpoverished‘drierZambezianmiombowoodland’outlinedbyWhite(1983).DominanttreespeciesincludeBrachystegiaspiciformis,B.boehmii,B.allenii,andJulbernardiaglobiflora.Inareasofhigherrainfall,atransitiontowettermiombooccurs(White1983).ThisecoregionisseparatedfromadjacentecoregionstothewestbyLakeMalawiandtheShireRiver,runningsouthfromMountMulanje.InthenorthitisseparatedfromtheCentralZambezianMiomboWoodlandecoregionbytheEasternArcandSouthernRiftMontaneareas.SeveralriverstraversetheEasternMiomboecoregioninapredominantlywest-eastdirection;theseincludetheRufijiRiverinTanzaniaandtheRioRuvumaandRioLurioinnorthernMozambique.TheZambeziRiverisfoundbeyondthesouthernborderoftheecoregion.Moderatelyundulatingridgesmixedwithshallowflat-bottomedvalleys,ordambos,thatareoftenseasonallywaterlogged,characterizethelandscape.Inselbergsarecommon,especiallyinnorthernMozambique,risingnoticeablyabovetheuniformwoodlands.TheunderlyinggeologyoftheEasternMiomboWoodlandconsistsmainlyofmetamorphosedupper-Precambrianschistsandgneisses,interspersedwithintrusivegranites(Bridges1990).Thecombinationofthecrystallinenatureoftheserocks,lowrelief,moistclimateandwarm1http://worldwildlife.org/ecoregions/at0706VM0029,Version1.0SectoralScope14Page5temperatureshasproducedhighlyweatheredsoilsthatarecommonlymorethan3mdeep(Frost1996).Thesoilsaretypicallywell-drained,highlyleached,nutrient-poor,andacidicwithloworganicmatter.Oxisolsandalfisolsaremostcommoninthesouthandcentralregionsoftheecoregion,whileahigherpercentageofultisolsarefoundtothenorth.TheecoregionexperiencesaseasonaltropicalclimatewithmostrainfallconcentratedinthehotsummermonthsfromNovemberthroughMarch.Thisisfollowedbyanintensewinterdroughtthatcanlastupto6months(WergerandCoetzee1978).Meanannualrainfallrangesbetween800and1,200mm,althoughpeaksupto1,400mmperannumarefoundalongthewesternmargins.Meanmaximumtemperaturesrangebetween21°Cand30°Cdependingonelevation,withthehottesttemperaturesexperiencedinthelowlandareas.Theecoregion’smeanminimumtemperaturesarebetween15°Cand21°C,andtheareaisvirtuallyfrost-free.Firesinthelatedryseasonburnwithahigherintensityduetodrier(mostlygrassy)fuelandhotterandoftenwindierclimaticconditions.Growingpopulationpressurehasincreasedthefrequencyofburningofthewoodlands,leadingtohighermortalityoflargetrees(thelargestcarbonpoolinthesystem)andresultinginbiomassdegradationandrelatedemissionsofgreenhousegases.Thistrendisexpectedtointensifyaspopulationanddevelopmentalpressuresbuild.Firesearlyinthedryseasonburnwithlowerintensityduetohigherfuelmoisturecontentandcoolerclimaticconditions,resultinginlowertreemortalityandnetbiomassgrowth.Earlydryseasonburnspreventlatedryseasonburnsbyremovingfuelload,andmaypreventpassageoflatedryseasonfirestootherforests.Targeted,preventativeearlyburningmaythereforereversethehistoricaltrendofbiomassdegradationintooneofbiomassregeneration.Earlyburningprojectsmaythereforeresultingreenhousegasemissionreductionsandremovals.2ThismethodologyusestheGapFiremodel(describedinAppendix1)tocalculateemissionreductionsandremovalsresultingfromtheproject’sfiremanagementactivities.GapFiremodelsthegrowthandmortalityofindividualtreesunderdifferentfireregimesbasedonanensembleofcanopy-tree-sizedwoodlandpatches.ThemodelwasdevelopedandcalibratedtotheEasternMiomboecoregionbyresearchersattheSchoolofGeoSciences,UniversityofEdinburgh.Modelinputsincludehistoricallyobserved(baseline)andmonitored(project)frequenciesofearlyburn,lateburnandnoburnatapatch,eachtriggeringdifferingtreemortalityprobabilityfunctions.Baselineburnprobabilitiesareestimatedfromburnscarobservationsonsatelliteimagesfromthe10-yearperiodbeforetheprojectstartdate.ItisanticipatedthatprojectsinotherdrylandforestecoregionscouldapplythismethodologywhereitisrevisedtoallowfortheuseofaversionofGapFirecalibratedforthatecoregion.Monitoringofprojectburnprobabilitiesisperformedbyassessingcheckpointsthroughouttheprojectareaaftertheendoftheearlyburningseason,andre-assessingthemaftertheendofthelateburningseason.Burnedcheckpointsarerecorded,andattheendoftheyeartherelativeproportionsofcheckpointsthatwereearlyburned,lateburnedandnotburnedarecalculated.ThisdataistranslatedintoburnprobabilitiesforthatyearwhichareinputintoGapFiretocalculateprojectcarbonstockchanges.2ForamoredetaileddiscussionseeGOFC-GOLD(2012,Section2.6.2),andStronach(2009).VM0029,Version1.0SectoralScope14Page6Selectiveharvestingoftreesispermittedinboththebaselineandprojectscenarios.3DEFINITIONSInadditiontothedefinitionssetoutinVCSdocumentProgramDefinitions,thefollowingdefinitionsapplytothismethodology:BaselineReferenceRegion(BRR)TheregionusedtoestablishbaselineburnprobabilitiesCountablePixelsThosepixelsinthecompositemapofburnscarobservationsforwhichthereisatleastoneconclusiveearly-seasonburnobservation,oroneconclusivelate-seasonorpost-late-seasonburnobservation,availableforatleastfiveyearsoutofthe10-yearperiodofanalysis.Aconclusiveobservationmeansnodataaremissinganditisnotobscuredbyclouds.Post-lateseasonobservationsmaybemadeupto3monthsaftertheendofburningseasondate.DBHDiameteratbreastheight,ameasurementoftreestemsizetakenat1.3mabovegroundheightEarliestPossibleBurnDateThefirstdayoftheearlyburningseasonEarlyBurning,EarlyBurnThefiremanagementpracticeofusingpre-emptivelow-intensityfiresintheearlyburningseasontoremovemostgrasses–theprinciplemeansbywhichuncontrolledbushfiresspread–whenfuelmoisturecontentishigh.Lowtemperaturesanddewfallusuallyextinguishthesefiresovernight.EarlyBurningProbabilityTheprobabilitythatanareawillburnintheearlyburningseasoninagivenyearEarlyBurningSeasonTheearlypartofthedryseason(usuallyaroundMayandJuneintheEasternMiomboecoregion)thatstartsattheendoftherainyseason(ie,whenfuelmoisturestartsdecreasing)andendsthedayaftertheearlyburningseasoncut-offdateEarlyBurningSeasonCut-offDateThelastdayoftheearlyburningseasonEasternMiomboEcoregionArelativelyunbrokenareacoveringtheinteriorregionsofsoutheasternTanzaniaandthenorthernhalfofMozambique,withafewpatchesextendingintosoutheasternMalawi.SeeSection2forfurtherinformationabouttheEasternMiomboecoregion.EndofBurningSeasonDateThedatethattypicallymarkstheendofthedryseasonandtheonsetofrainsVM0029,Version1.0SectoralScope14Page7FireDrivingActivitiesActivitiesthatdrivethestartingofuncontrolledfiresinmiombowoodlands(ie,nottheactuallightingofthematch,buttheunderlyingreasonwhythereisfire).Theseincludehunting(fireisusedtoimprovevisibility),walkingalongfootpaths(fireisusedtoimprovevisibilitytopreventsurpriseattacksfromwildanimals),agriculture(fireisusedforclearingnewfieldsorburningagriculturalresiduesfromthepreviousyear)andcharcoalproduction(firemayescapefromcharcoalmoundsiftreatedimproperly).FireintensityAmeasureoftheenergyreleasedbyafire,closelyrelatedtoitsexpectedimpactsuponawoodlandecosystemForestManagementUnit(FMU)Acontiguousareaofforestedlandcontrolledbyasinglelandmanagerinwhichtheprojectisimplemented.AsinglelandmanagermaycontrolmorethanoneFMUandasingleFMUmaycovermultiplebiomassstrata.GapFireThemodelusedtopredictratesofforestgrowthanddegradation,basedonthemodeloriginallypublishedinRyanandWilliams(2011)KilwaMiombowoodlandfieldsiteintheKilwadistrictofsoutheastTanzaniaLandManagerAlegalentitythatcontrolslandonwhichtheprojectisimplemented(eg,acommunity,village,districtorprivatelandowner).Landmanagersaremembersofaprojectthatareledandmanagedbytheprojectproponent.Landmanagershavethelegalauthoritytoenterintoanagreementwiththeprojectproponent(ie,notindividualsbelongingtoacommunity).Inthecaseofanon-groupedproject,thereisonlyonelandmanager.LateBurnTheresultwhenfiresarelitinthedryseasonwhenfuelmoisturecontentislow.Thesefiresfrequentlyburnthroughthenightandaredifficulttocontrol.LateBurningProbabilityTheprobabilitythatanareawillburninthelateburningseasoninagivenyearLateBurningSeasonThelaterpartofthedryseason(usuallyfromJulytoNovemberintheEasternMiomboecoregion)thatstartsthedayaftertheearlyburningseasoncut-offdateandendswhentherainyseasonstarts(ie,whenfuelmoistureincreasesagain)NDVINormalisedDifferenceVegetationIndex,ameasureofplantgreennesscalculatedusingmeasurementsofnearinfraredandredsurfacereflectanceVM0029,Version1.0SectoralScope14Page8NoBurningProbabilityTheprobabilitythatanareawillnotburninagivenyear4APPLICABILITYCONDITIONSThismethodologyappliestoprojectsthatimplementpreventativeearlyburningactivitiesinmiombowoodlandsintheEasternMiomboecoregionofAfrica.Thismethodologyisapplicableunderthefollowingconditions:1)ProjectsmustbelocatedwithintheEasternMiomboecoregion.2)Projectsmustimplementpreventativeearlyburningactivitiesinmiombowoodlands.3)Projectsmayincludetheselectiveharvestingoftrees,thoughtheprojectdescriptionmustspecifyhowharvestingismanaged,andspecifyhowitwillbemonitoredtoensuresustainability(ie,thatharvestedbiomassisnotgreaterthanregenerationcapacity)usingindustrystandardmeasuressuchasannualallowablecutandmeanannualincrement.4)Projectareasmustmeetaninternationallyaccepteddefinitionofforestandhavedonesoforatleast10yearspriortotheprojectstartdate.5)Thepre-projectlandusewithintheprojectareamusthavebeencontinuouslypresentforatleast10yearspriortotheprojectstartdate.6)Firemusthavebeenthepredominantagentofdegradationforatleast10yearspriortotheprojectstartdate.Thismethodologyisnotapplicableunderthefollowingconditions:1)Thebaselinescenarioincludesanthropogenicactivitiesthatincreasecarbonstocksorreducecarbonstockdegradationrelativetothepre-projectlanduse.2)Theprojectincludestheimplementationofanyactivitiesnotrelatedtofiremanagementorselectivesustainabletimberharvesting(eg,charcoalproduction,unsustainabletimberharvestingorgrazing)thatresultinemissions,unlesstheprojectproponentdemonstratesthattheseemissionswillbedeminimisfortheentiretyoftheprojectcreditingperiod.3)Morethan10percentofthebasalareaofforestsintheprojectareahasbeenremovedinthepast10yearspriortotheprojectstartdatefromlogging,charcoalmaking,grazing3orotheractivitiesthataffectthestemsizedistribution4.3Theimpactofgrazinganimalsontreebiomassisgenerallyminimal,evenfrombrowserssuchasgoatsandsheep.Themainimpactonmiomboforestsfromgrazingisfromhumansfellingtreesandburningtoopenupthewoodlandforthedevelopmentofmoregrassyvegetationfortheirlivestocktofeedon.4Visualassessmentwillsufficeundernormalcircumstances.Otherwise,comparethebasalareaofstumpstothebasalareaofstandingtrees.Thismethodologyisnotapplicablewherethebasalareaofstumpsexceeds10percentofthebasalareaofstandingtrees.VM0029,Version1.0SectoralScope14Page95PROJECTBOUNDARY5.1SpatialExtentThespatialextentoftheprojectboundaryencompassestheareasundercontroloftheparticipatinglandmanager(s).TheprojectareasmayconsistofonecontiguousormultiplediscreteFMUswithinthelandmanager’slands.5.2UpperandLowerCarbonDensityBoundariesoftheProjectAreasProjectareasmustbecoveredinmiombowoodlandswithanabovegroundtreecarbondensitynotlowerthan5tC/haandnotgreaterthan35tC/haattheprojectstartdate.Themiombowoodlandsincludedwithintheprojectareamustcomplywithaninternationallyacceptedforestdefinition.ThismethodologyusestheFAOdefinition5:Landwithtreecrowncover(orequivalentstockinglevel)ofmorethan10percentandanareaofmorethan0.5hectares(ha)Thetreesshouldbeabletoreachaminimumheightof5meters(m)atmaturityinsitu.Mayconsisteitherofclosedforestformationswheretreesofvariousstoreyandundergrowthcoverahighproportionofthegroundoropenforestformationswithacontinuousvegetationcoverinwhichtreecrowncoverexceeds10percent.Wheretheprojectproponentwishestouseadifferentforestdefinition,theprojectproponentmustusecredibleliteraturesourcestoestablishthelowercarbondensityboundary(equivalenttotheminimumvalueofcrowncoverofthatdefinition)andapplysuchboundaryinthismethodology.Sinceallinternationallyacceptedforestdefinitionsuseacrowncoverof10percentorhigher,thelowercarbondensitythresholdwillalwaysbeequaltoorabove5tC/ha,andthismethodologywillthereforebeapplicable.5.3DelineationofProjectAreaFornon-groupedprojects,theprojectareamustbedelineatedatthestartofprojectactivities.Forgroupedprojects,theprojectarea(distinctfromthegeographicareawithinwhichnewprojectactivityinstancesmaybeadded)mustbeexpandedeachtimeanewlandmanagerjoinsatprojectverification.TheareaforeachlandmanagermustbeestablishedindigitalformatbywalkingtheboundarieswithaGPS,orbyuploadinganexistingshapefileoftheareaintoaGIS,whereitmustbeoverlaidwiththecarbondensitymap.5Thelowercarbondensitythresholdforprojectareasof5tC/hausedinthismethodologycomplieswiththeFAOforestdefinition,assupportedbyIsango(2007)whoidentifiedtheaveragecarbondensityat10%canopycoverinmiomboataround2tC/ha.Theuppercarbondensitylimitforprojectareasof35tC/haisatthelowerendofatransitiontowardsforeststhatburnsignificantlylessfrequentlyduetoareducedfuelloadasaresultofcanopyclosure(basedonunpublishedfielddatabytheUniversityofEdinburghinmiombowoodlandsinTanzania).VM0029,Version1.0SectoralScope14Page10TheprojectareamustbeupdatedintheGISeachyear,excludingthoseareaswherelandusechangeeventshavebeenmonitored.5.4GreenhouseGasesandCarbonPoolsThegreenhousegasesandcarbonpoolsincludedinorexcludedfromtheprojectboundaryareshowninTable1andTable3below,respectively.Table1:GHGSourcesIncludedInorExcludedFromtheProjectBoundarySourceGasIncluded?Justification/ExplanationBaselineBiomassburningCO2YesReducingCO2emissionsfromforestfiresisthemainobjectiveofthismethodology.CH4OptionalEmissionsarerelatedtotheburningofbiomassinmiombofires.TheyarecalculatedusingVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).Sinceprojectsapplyingthismethodologyareexpectedtoreduceemissionsfrombiomasscombustion,theinclusionofthissourcewouldleadtonetGHGemissionreductions.N2OOptionalEmissionsarerelatedtotheburningofbiomassinmiombofires.TheyarecalculatedusingVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).Sinceprojectsapplyingthismethodologyareexpectedtoreduceemissionsfrombiomasscombustion,theinclusionofthissourcewouldleadtonetGHGemissionreductions.BaselineSelectiveharvestingCO2NoBaselineCO2emissionsfromselectiveharvestingareconservativelynotquantifiedinthismethodologyCH4NoBaselineCH4emissionsfromselectiveharvestingareconservativelynotquantifiedinthismethodologyN2ONoBaselineN2OemissionsfromselectiveharvestingareconservativelynotquantifiedinthismethodologyProjectBiomassburningCO2YesReducingCO2emissionsfromforestfiresisthemainobjectiveofthismethodology.CH4OptionalEmissionsarerelatedtotheburningofbiomassinmiombofires.TheyarecalculatedusingVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).SinceVM0029,Version1.0SectoralScope14Page11SourceGasIncluded?Justification/Explanationprojectsapplyingthismethodologyareexpectedtoreduceemissionsfrombiomasscombustion,theinclusionofthissourcewouldleadtonetGHGemissionreductions.N2OOptionalEmissionsarerelatedtotheburningofbiomassinmiombofires.TheyarecalculatedusingVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).Sinceprojectsapplyingthismethodologyareexpectedtoreduceemissionsfrombiomasscombustion,theinclusionofthissourcewouldleadtonetGHGemissionreductions.ProjectSelectiveharvestingCO2YesSelectiveharvestingoftreesleadstoprojectCO2emissions,whichneedtobeincludedintheoverallcalculationofemissionreductions.CH4NoCH4emissionsfromselectiveharvestingcanbeconsideredtobenegligible.N2ONoN2Oemissionsfromselectiveharvestingarenilorcanbeconsideredtobenegligible.Table2:CarbonpoolsIncludedInorExcludedfromtheProjectBoundaryCarbonPoolIncluded/ExcludedJustification/ExplanationofChoiceAbove-groundtreebiomassIncludedMajorcarbonpoolaffectedbyprojectactivitiesAbove-groundnon-treebiomassExcludedChangeisexpectedtobepositiveorinsignificantandthereforethispoolisconservativelyexcluded.Fireexperimentshaveshownthehighvulnerabilityofsmallstems(includingshrubs)tofire,particularlyintensefire(RyanandWilliams,2010).Grasses(aboveground)areknowntobecompletelycombustedinfire.Therefore,undertheprojectscenario,itisexpectedthatnon-treebiomasswillbelargercomparedtothebaselinescenario,asreducedfireintensitywillreducelossesinthispool.Below-groundbiomassExcludedChangeisexpectedtobepositiveorinsignificantandthereforethispoolisconservativelyexcluded.VM0029,Version1.0SectoralScope14Page12CarbonPoolIncluded/ExcludedJustification/ExplanationofChoiceDeadwoodExcludedChangeisexpectedtobepositiveorinsignificantandthereforethispoolisconservativelyexcluded.SeeSection5.3ofAppendix1forfurtherdiscussion.LitterExcludedChangeisexpectedtobepositiveorinsignificantandthereforethispoolisconservativelyexcluded.SoilorganiccarbonExcludedChangeisexpectedtobepositiveorinsignificantandthereforethispoolisconservativelyexcluded.WoodproductsIncludedCarbonpoolaffectedbyprojectactivities6BASELINESCENARIOTheprojectproponentmustapplyStep1ofthelatestversionofVT0001,ToolfortheDemonstrationandAssessmentofAdditionalityinVCSAFOLUProjectActivities,whichresultsinalistofrealisticandcrediblealternativelandusescenariostotheprojectactivityandidentifiesthemostplausiblebaselinescenario.FortheapplicationofStep1cofVT0001,theprojectproponentmustapplythefollowingprocedure:Step1.Barrieranalysis.TakingthelistofcrediblealternativelandusescenariosresultingfromStep1bofVT0001,abarrieranalysismustbeconductedtoidentifyrealisticandcrediblebarriersthatpreventimplementationoftheselandusescenariosfollowingtheproceduresdescribedinSub-step3aofVT0001,mutatismutandis.Theprojectproponentmustindicatewhichofthealternativelandusescenarioswouldfacewhichidentifiedbarrier,andprovideverifiableinformationtosupportthepresenceofeachparticularbarrierinrelationtoeachalternativelandusescenario.Step2.Eliminationofalternativeswithbarriers.UsingtheinformationinStep1,eliminatealternativelandusescenariosthatfaceabarriertoimplementation.Alllandusescenariosthatfaceabarriertoimplementationmustberemovedfromthelist.Atleastoneviablealternativelandusescenarioshallbeidentified.Step3.Selectionofmostplausiblebaselinescenario(ifallowedbybarrieranalysis).Wherethereisonlyonealternativelanduseremaininginthelist,thisalternativemustbeselectedasthemostplausiblebaselinescenario.Wherethereismorethanonealternativelanduseremaininginthelist,andoneofthesealternativesincludescontinuationofcurrentlanduse(ie,landuseimmediatelypriortocommencementoftheproject),continuationofcurrentlandusemustbechosenasthemostplausiblebaselinescenariowherethefollowingconditionsaremet:1)Thelandmanager(s)hasnotchangedinthefiveyearspriortotheprojectstartdate;VM0029,Version1.0SectoralScope14Page132)Thecurrentlanduseshavenotchangedinthefiveyearspriortotheprojectstartdate;and3)Therehavebeennochangesinmandatoryapplicablelegalorregulatoryrequirementsduringthefiveyearspriortotheprojectstartdateandnosuchchangesarecurrentlyunderlegalreviewbytherelevantauthorities.Wherethereismorethanonealternativelanduseremaininginthelistandthemostplausiblelandusehasstillnotbeenidentified,proceedtoSteps4and5.Step4.Assesstheprofitabilityofalternativelandusescenarios.TakingthelistofalternativelandusescenariosresultingfromStep2thatfacenobarrierstoimplementation,documentthecostsandrevenuesassociatedwitheachalternativelanduseandestimatetheprofitabilityofeachalternativelanduse.Theprofitabilityofalternativelandusesmustbeassessedintermsofthenetpresentvalue(NPV)ofnetincomesovertheprojectcreditingperiod.Thekeyeconomicparametersandassumptionsusedintheanalysismustbejustifiedinatransparentmanner.Step5.Selectionofmostplausiblebaselinescenario(fromprofitabilityanalysis).SelectthemostprofitablelandusescenariofromtheanalysisinStep4asthemostplausiblebaselinescenario.WheretheNPVofoneormoreofthealternativelandusescenarioscannotbeestablished,andwherealloftheremainingalternativelandusescenariosdescribealandusewherecarbonstocksaredegradedequallyormoreseverelyrelativetothecurrentlandusethismethodologyconservativelyassumesthebaselinescenariotobethecontinuationofthecurrentlanduseonly(ie,miombovegetationsubjectedtocarbonstockdegradationbyanthropogenicfires,withtheoccurrenceandintensityoffiresequaltothoseobservedinthe10yearsbeforethestartoftheprojectactivity).Thisassumptionisconservativebecause:Futurecarbonstockdegradationasaresultofanthropogenicfiresislikelytoincreaseastheoccurrenceoffiresislikelytoincreaseduetopopulationgrowth.Notconsideringplausiblealternativelandusescenariosthatincludeothercausesofforestdegradationordeforestationleadstoahigherestimateofbaselinecarbonstocksandthusalowerestimateofaproject’sGHGemissionreductionsandremovals.WheretheNPVofoneormoreofthealternativelandusescenarioscannotbeestablished,andwhereoneofthesealternativelandusescenariosisananthropogenicactivitywhichincreasescarbonstocksorreducescarbonstockdegradationrelativetothepre-projectlanduse(eg,reforestation,enrichmentplantingorareductioninfireoccurrenceand/orfireintensityduetofiremanagementorduetootheranthropogeniccausessuchasareductioninpopulationdensity),thismethodologyconservativelyassumesthistobethemostplausiblebaselinescenarioand,aspertheapplicabilityconditionsofthismethodology,themethodologyisnotapplicable.VM0029,Version1.0SectoralScope14Page147ADDITIONALITYTheprojectproponentmustapplythelatestversionofVT0001ToolfortheDemonstrationandAssessmentofAdditionalityinVCSAFOLUProjectActivities.8QUANTIFICATIONOFGHGEMISSIONREDUCTIONSANDREMOVALS8.1BaselineEmissionsBaselineemissionsarecalculatedasfollows:BEy=BEBM,y+BEBiomassburn,y+BEHarvest,y(1)Where:BEy=Baselineemissionsinyeary(tCO2e)BEBM,y=Baselineemissionsfromabovegroundbiomassdegradationinyeary(tCO2e)BEBiomassburn,y=BaselineemissionsofN2OandCH4frombiomassburninginyeary(tCO2e)BEHarvest,y=Baselineemissionsfromselectivetreeharvestinginyeary(tCO2e)ForthecalculationofBEBM,y,seeSection8.1.1.ForthecalculationofBEBiomassburn,y,seeSection8.1.2.ForthecalculationofBEHarvest,y,seeSection8.1.3.8.1.1BaselineEmissionsfromAbovegroundBiomassDegradationTheprojectproponentmustcalculateBEBM,yusingthefollowingsteps:8.1.1.1Step1:PrepareaCarbonDensityMapoftheRegionaroundtheProjectAreaTheprojectproponentmustprepareacarbondensitymaporuseanexistingone.Thismapmustmeetthefollowingrequirements:Itmustbebasedoninformationcollectedwithinfiveyearspriortotheprojectstartdate,andmustconsistofremotelysenseddatacalibratedbyfieldmeasurements.Thecarbondensitymapmustnotatanypointduringtheprojectcreditingperiodbeolderthan10years.Thus,thecarbondensitymapmustberevisedatleastevery10yearsanditmustbebasedoninformationcollectedwithinfiveyearspriortotherevisiondate.RefertoSection9.3.3forfurtherdetailsonmaprevisions.Itmusthaveaspatialresolutionnotcoarserthan200x200mpixelsize.ItmustgiveanabovegroundcarbonvalueintC/ha.SyntheticApertureRadarsatelliteimagerydatawithwavelengths>20cmmaybeusedasthebasisforthisanalysis.TerraincorrectionmustbeappliedtoradardatausingapublishedmethodandapublishedDigitalElevationModel(DEM).VM0029,Version1.0SectoralScope14Page15TheremotesensingdatashouldbecalibratedwithfielddataasdescribedinSection9.3.1below.Ifalternativefielddataisused,justificationmustbeprovidedtoshowhowitisequivalenttoorexceedstheserequirements.Miomboistypicallyahighlyheterogeneousenvironment,sothecarbondensitymapmayexhibitalotofspecklecomplicatingmanagement.Themapmaybedegradedinresolutionusinganyneutralalgorithmsolongasthepixelsizeisnomorethan25percentofthesampleplotsusedtoanchorit(50percentlimitineachdirection).Aconfidenceintervalmustbecalculatedforthemeanbiomassacrosstheentiremapusingmultiplerandomsamplesofcalibrationdatatobootstraptheclassificationwiththeremainderusedtoassessaccuracy.The95percentconfidenceintervalmustbenowiderthan30percentofthemeanestimate.Themapmaybeextendedbeyondthescenesthathavebeencalibratedbyfielddatabyequalizingoversceneoverlaps,providedtheneighboringscenesarebroadlysimilarinvegetationtypeandtopography,andtheconfidencelimitoftheextendedmapstillmeetstheabovecriteria.Themapmustincludetheproject’sbaselinereferenceregion(BRR-seeStep3,Section8.1.1.3below).SincetheestablishmentoftheBRRrequirestheuseofthecarbondensitymap,aniterativeprocessisnecessaryinwhichthecarbondensitymapisexpandeduntilminimumrequirementsfortheBRRaremet.8.1.1.2Step2:StratifytheCarbonDensityMapThefirstcarbondensitymapoftheprojectmustbestratifiedintosixstrata,eachcoveringarangeof5tC/haandaccordingtothestratificationschemepresentedinTable3below.Theloweststratum(Stratum1)maycoverarangelessthan5tC/ha,andmaystartatanyvaluehigherthan5tC/ha,dependingonthelowercarbondensitythresholdasspecifiedbytheforestdefinitiontheprojectproponentselects.Ifthethresholdishigherthan5tC/ha,thenStratum1willstartatthelowercarbondensitythresholdforthatforestdefinition,andextendtothenextmultipleof5.Themedianvalueforthisstratumisthemid-pointbetweenthelowerandhigherboundaryvaluesforthestratum.Thenextstrataarethennumberedconsecutively,witheachcoveringarangeof5tC/hauntiltheuppercarbondensitythresholdof35tC/haisreached.Subsequentrevisionsofthecarbondensitymapmayaddnewstrataabove35tC/ha,eachcoveringarangeof5tC/ha.Thesenewstratamayonlyincludeprojectareasthatwerepreviouslybelowthemaximumbiomassallowable,buthavesincerisenabovethatthreshold.Table3:RangeandMedianValue(tC/ha)oftheSixCarbonDensityStrataStratumRange(tC/ha)Mediancarbondensity(tC/ha)1≥5and<107.52≥10and<1512.53≥15and<2017.5VM0029,Version1.0SectoralScope14Page16StratumRange(tC/ha)Mediancarbondensity(tC/ha)4≥20and<2522.55≥25and<3027.56≥30and<3532.58.1.1.3Step3:DelineateProjectAreaandBRRontheCarbonDensityMapAllshapefilesoftheprojectareamustbeoverlaidontothecarbondensitymap.Theshapefilesmustbeupdatedeachyearalandmanagerisaddedtotheproject,andtheymustbeoverlaidontothemostrecentcarbondensitymap.TheprojectproponentmustthendeterminetheareaAi,yofeachstratumiwithintheprojectareainyeary.Thiswillbeusedinequation(3)(Section8.1.1.6).Thebaselinereferenceregion(BRR)istheregioninwhichhistoricalfireoccurrence(baselinefirehistory)dataarecollected.TheBRRmustbeselectedpriortotheprojectstartdateandremainsfixedforthevalidityofthefirstcarbondensitymap.AnewBRRmustbeselectedonthemostrecentcarbondensitymapateachrevisionofthebaselinefirehistory(seeproceduresspecifiedinSection9.3.4).TheBRRmayconsistofmultiplediscreetareas.Areasthatmeetthebelowcriteriathatareincloserproximitytotheprojectareamusttakepriorityoverareasfurtheraway(ie,theremustbeno“cherrypicking”ofareas).TheprojectproponentmaydelineatemorethanoneBRRtodeterminedifferentbaselinefirehistoriesfordifferentportionsoftheprojectarea.TheprojectproponentmustclearlyindicatewhichprojectareashavetheirbaselinefirehistorydeterminedbywhichBRR,andtheirsimilarityincarbondensityandfirehistorymustbedemonstratedpertherequirementsinthissectionandSection8.1.1.6.TheBRRmust(ataminimum)incorporate:1)Allprojectareaspresentedintheprojectdescription,aswellasanynewareasthatareaddedinthefuture;and2)Sufficientcountablepixelswithineachbiomassstratumtoestablishhistoricalburnprobabilities(seeStep5,Section8.1.1.5).Theprojectareamustnotbesignificantlydifferent(atthe95percentconfidencelevel)inthedistributionofbiomassfromtheBRR,asmeasuredbythechi-squaredgoodnessoffitstatistic,comparingthenumberofpixelsintheprojectareaineachbiomassstratumwiththatpredictedbyproportionsintheBRR.Theprojectproponentmustdemonstratethat,fortheareasinsidetheBRR,butoutsidetheprojectarea,thefollowingaretrue:1)TheyarelocatedwithintheEasternMiomboecoregion.VM0029,Version1.0SectoralScope14Page172)Theyarecoveredwithmiombowoodlandwithabovegroundcarbondensitiesbetween5tC/ha(unlessotherwisespecifiedbytheforestdefinitionselectedbytheprojectproponent)and35tC/ha.Allotherareas(eg,agriculturalareas,urbanareasandgrasslandsandwoodlands/forestwithcarbondensitieslowerthan5tC/haandhigherthan35tC/ha)mustbeexcluded.TheprojectproponentmustusethecarbondensitymappreparedunderStep1(Section8.1.1.1)todelineatetheBRR.3)ThemaincauseofcarbonstockdegradationintheBRRisanthropogenicfires.Theprojectproponentmayusepublishedorunpublisheddatasets,expertliteraturesourcesorcredibleexpertopiniontodemonstratethiscriterion.4)Thefiredrivingactivitiesaresimilarandofsimilarrelativeimportancetofireoccurrenceasthoseintheprojectareas.Theprojectproponentmayusepublishedorunpublisheddatasets,expertliteraturesourcesorcredibleexpertopiniontodemonstratethiscriterion.Step6belowaddressesthequantitativesimilarityofthefirehistory.8.1.1.4Step4:EstablishtheCut-OffDatebetweentheEarlyBurningSeasonandtheLateBurningSeason,andtheEarliestPossibleBurnDateandtheEndoftheBurningSeasonDateTheprojectproponentmustspecifyanearlyburningseasoncut-offdatebetweentheearlyburningseasonsandlateburningseasons,whichmustbeusedforboththebaselineandprojectscenarios.Thecut-offdateisthelastdayoftheearlyburningseason.Itmustbespecifiedbytheprojectproponentpriortotheprojectstartdateandremainfixedfortheprojectcreditingperiod.Theprojectproponentmustusethedefaultdateof30Juneorchooseanalternativedatesupportedbythecredibleopinion(s)ofoneormoreexpertswithdemonstrableinsightsinmiombofiremanagementand/orhands-onexperiencewithfiresintheprojectregion6.Theprojectproponentmustalsospecifytheearliestpossibleburndatethatmarksthebeginningoftheearlyburningseason.Aswiththecut-offdate,itmustbespecifiedpriortotheprojectstartdateandremainfixedfortheprojectcreditingperiod.Theprojectproponentshouldusethedayatthebeginningofthedryseasonatwhichthemonthlyrunningaveragerainfall(atleast10yearsofdata)firstfallsbelow33percentofpeakwetseasonrainfall.Alternatively,theprojectproponentmayusethecredibleopinion(s)ofoneormoreexpertswithdemonstrableinsightsinmiombofiremanagementand/orhands-onexperiencewithfiresintheprojectregiontojustifyanalternativeearliestpossibleburndate.Finally,theprojectproponentmustestablishtheendofburningseasondatethatmarkstheendofthedryseason.Aswiththecut-offdate,itmustbespecifiedbytheprojectproponentpriortotheprojectstartdateandremainfixedfortheprojectcreditingperiod.Theprojectproponentshouldusethedayattheendofthedryseasonatwhichthemonthlyrunningaveragerainfall(atleast10yearsofdata)firstrisesabove33percentofpeakwetseasonrainfall.Alternatively,theprojectproponentmayusethecredibleopinion(s)ofoneormore6Suitableexpertsinclude,butarenotlimitedto,localgovernmentofficialsandecologicalresearchers.Theprojectproponentmayadditionallychoosetoconsultinternationalexpertsinlargescaleearlyburning.VM0029,Version1.0SectoralScope14Page18expertswithdemonstrableinsightsinmiombofiremanagementand/orhands-onexperiencewithfiresintheprojectregiontojustifyanalternativeendofburningseasondate.8.1.1.5Step5:DetermineHistoricBurnProbabilitiesforEachStratumintheBRR(FireHistory)Sub-step5.1.Thebaselinefirefrequenciesaredeterminedfromafirehistorythatmustbederivedattheoutsetoftheproject,assetoutbelow.Every10yearsthefirehistorymustberevalidatedassetoutinSection9.3.4below.Toconstructtheinitialfirehistory,theprojectproponentmustgatherasmanyearly-andlateseasonburnobservationsaspossiblefortheentireBRRforthehistorical10-yearperiodbeforetheprojectstartdateor,foreachrevisionofthefirehistorybaseline(seeSection9.3.4),saidobservationsmustbegatheredforthehistorical10-yearperiodbeforetherevisiondate.Theprojectstartdatemaynotbedelayedbymorethan5yearsbeyondthelastyearofthehistorical10-yearperiod.Theprojectproponentmust:1)CollectcompositesatelliteobservationsfortheBRRfrommediumresolution(ie,10sofmeterpixels)images(suchasLandsatorSPOT)usingspectralindicesasaproxyforfireoccurrence,andgroupthemforeachofthe10years.Imagesfrombeforetheearliestpossibleburndate,orfrommorethanthreemonthsaftertheendofburningseasondate,mustbeexcluded.Theremustbenosystematicbiasintheselectionofimages.Wherefeasible,allavailableimagesinthattimeperiodfromtheselectedsourceshouldbeanalyzed.2)Applyaburnedareadetectionalgorithmtodeterminewhereafirehasoccurred(eg,thespectralindexmethodasdescribedinBastarrika,2011).Thismustbeperformedforallcloud-freepixelsontheavailableimages(ie,no“cherrypicking”).Createtheburnscardatasetaccordingtothefollowingprocedure:a)Satelliteimageryshouldundergoatmosphericcorrection,cloudmaskingandconversiontoanestimateofsurfacereflectance,whereappropriate.b)Whereappropriate,aseriesofspectralindicesshouldbecalculatedwhichhighlightthereflectivepropertiesofburnscars(eg,NDVI,GEMI,MIRBI).Inadditiontoaidingclassification,useoftheseindicesmitigatesagainsttheeffectsofvaryingillumination,acquisitiongeometriesandtopographiceffectsinsatelliteimagery.c)Atrainingdatasetofburnedandunburnedareasmustbemanuallyidentifiedinremotesensingimagerybyacompetentoperator.Theoperatormusthaveaccesstoallavailablesatelliteimagery.Thisdatasetmustincludeclassifiedpixelsacrossanumberofimages,spanningdifferentlocalities,landcovertypes,seasonsandyears.Thetrainingdatamustbedevelopedconservatively,withonlythecentralpartsofburnscars(wheretherecanbenodoubt)labeledasburnedinthetrainingdata.Thisdatasetmustbepreservedasevidenceofconservativeness.VM0029,Version1.0SectoralScope14Page19d)Thelikelihoodofaburnmustbecomputedforeachusablepixelineachimageusingasupervisedimageclassificationalgorithm.Theclassifierderivationshouldincorporateavariablesparsitypreferencetoavoidover-fitting.Theclassificationshouldbecarriedoutasacross-validation,andboot-strappedwithapartofthetrainingdatausedtoguidetheclassifier,andtheremainingtrainingdatausedtoassessaccuracy.Aconsistentaverageclassificationaccuracyof>95percentisrequiredforacceptanceoftheclassificationalgorithm.e)Wherethestatusofapixelisambiguous,theclassifiermustuseaconservativeminimumthresholdofburnlikelihood(atleast60percent7)whenregisteringapixelasburned.f)Theearliestburnscarobservationofapixelinagivenyeardisqualifiesanysubsequentobservationsofthesamepixelinthatyearfromfurtherconsiderationintheanalysis.3)Attributeeachobservedburnscartoaburnintheearlyburningseasonorthelateburningseasonusingthedateoftheobservation,accordingtotable5below.Table5:BurnScarAttributionPreviousObservationNextObservationTreatmentNoneNone(ie,noobservationsrecordedwithinthatyear)NodataNoneEarlyseasonburnEarlyseasonfireNoneEarlyseasonnoburnContinueprocessinglaterobservationsNoneLateseasonburnProbabilisticfireassignment(seebelow)NoneLateseasonnoburnContinueprocessinglaterobservationsNonePostlateseasonburnProbabilisticfireassignment(seebelow)NonePostlateseasonnoburnNodataEarlyseasonnoburnNone(ie,nomoreobservationsrecordedwithinthatyear)NodataEarlyseasonnoburnEarlyseasonburnEarlyseasonfireEarlyseasonnoburnEarlyseasonnoburnContinueprocessinglaterobservations7Giglioetal(2009)empiricallydeterminedthattheposteriorprobabilityofdetectionshouldbeatleast0.6tominimisethefrequencyofcommissionerrors.VM0029,Version1.0SectoralScope14Page20PreviousObservationNextObservationTreatmentEarlyseasonnoburnLateseasonburnProbabilisticfireassignment(seebelow)EarlyseasonnoburnLateseasonnoburnContinueprocessinglaterobservationsEarlyseasonnoburnPostlateseasonburnProbabilisticfireassignment(seebelow)EarlyseasonnoburnPostlateseasonnoburnNodataLateseasonnoburnNone(ie,nomoreobservationsrecordedwithinthatyear)NoburnLateseasonnoburnLateseasonburnLateseasonfireLateseasonnoburnLateseasonnoburnContinueprocessinglaterobservationsLateseasonnoburnPostlateseasonburnProbabilisticfireassignment(seebelow)LateseasonnoburnPostlateseasonnoburnNoburnDatapaucitywillmeanthatforsomeburnscarsdetecteditwillnotbepossibletodefinitivelystatethatthescarrelatestoanearly,lateorpost-lateseasonfire.Insuchcases,thefiremustbeattributedprobabilisticallyinauniformdistributionbetweenthedateoftheburnobservationandthedateofthepreviousno-burnobservation(subjecttoalimitofthreemonthsbackintime,beyondwhichburnscarsarenotconsidereddetectable)ortheearliestpossibleburndate,whicheverislater.Probabilitiesmustbeexpressedasfractions(eg,0.37)sothattheycanbecountedinSub-step5.2below.Wherethereisano-burnobservationappearinginthelateburningseason,andnosubsequentobservationsaremade,thepixelmustbeassumedtonothaveburnedthatyear.Wheretheonlyobservationisano-burnbeforetheearlyburningseasoncut-offdate,thedatapointmustbediscardedduetoinsufficientevidenceoffireactivityinthelateburningseason.Sub-step5.2.Theprojectproponentmustdeterminetherelativeproportions(mustaddupto1)ofoccurrencesofearlyburning,lateburningandnoburningineachbiomassstratumintheBRR.Theprojectproponentmust:1)Countthepixelsinthecompositemapofobservationsforwhichthereisatleastoneconclusiveearly-seasonburnobservationoroneconclusivelate-seasonorpost-late-seasonobservationavailableforatleastfiveyearsoutofthe10-yearperiodofVM0029,Version1.0SectoralScope14Page21analysis8.Aconclusiveobservationmeansnodataaremissinganditisnotobscuredbyclouds.Post-lateseasonobservationsmaybemadeupto3monthsaftertheendofburningseasondate.Thesearethecountablepixelsinthecompositemap.2)Overlaythecountablepixelsontothecarbondensitymapandattributeeachcountablepixeltoabiomassstratum.Countablepixelsmustcoveratleast50percentoftheareaofeachbiomassstratumwithintheBRR.3)Foreachyearinthe10-yearperiodineachbiomassstratum,countthecountablepixelsthatburnedintheearlyburningseason,thatburnedinthelateburningseasonandthatobservablydidnotburnineitherperiod.Includefractions(resultingfromtheprobabilisticattributiondescribedinSub-step5.1,step4cabove)inthecount.4)Foreachbiomassstratum,sumuptheyearlytotalsofcountablepixelsthatburnedintheearlyburningseasonforthe10-yearperiod.Theprojectproponentmustdothesameforthosecountablepixelswhichburnedinthelateburningseasonandthosewhichdidnotburnineitherperiod.5)Foreachbiomassstratum,calculatetheproportionofthetotalnumberofpixelsthatburnedintheearlyburningseasonoverthetotalnumberofpixelsforthe10-yearperiodforwhichthereisobservationalevidence,asfollows:BLPROBEarlyburn,i=∑CountEarlyburni,y/∑CountPixi,y(2)Where:BLPROBEarlyburn,i=Baselineprobabilityofearlyburningoccurringinstratumi(fraction)CountEarlyburni,y=NumberofcountablepixelsinstratumiinyearythatshowedaburnscarintheearlyburningseasonCountPixi,y=NumberofcountablepixelsinstratumiinyearyTheprojectproponentmustuseEquation(2)tocalculateBLPROBLateburn,iandBLPROBNoburn,i,mutatismutandis.BLPROBEarlyburn,I,BLPROBLateburn,iandBLPROBNoburn,imustbeinputintheGapFiremodelasthebaselinerelativeprobabilitiesofearlyburning,lateburningandnoburningforthatstratum.8.1.1.6Step6:DemonstratetheProjectAreaAndBRRHaveSimilarFireBaselineHistoriesStep3aboverequirestheprojectareaandBRRtobequalitativelysimilar,andquantitativelysimilarinrespecttobiomass.Toavoidanybiases,theymustalsoexhibitsimilarfirehistories.Manyvariablesaffectthefrequencyoffiresandtheirspread,includingaccessibility(infrastructure),distancefromhumansettlementsandvariousaspectsoftopography(hilliness8BasedonArchibaldetal.,2013itcanbeestablishedthatmostareasintheEasternMiomboecoregionburnatleastonceeverytwoyears.Therefore,iffiveyearsofobservationsareavailablethiswillprovidesufficientdatatoconclusivelyestablishbaselineburnoccurrences.FiveyearsisalsosuggestedasaperiodofsufficientobservationsbyESA,2011.VM0029,Version1.0SectoralScope14Page22andslope).Ratherthanassessingeachoftheseinturn,theprojectproponentmustdemonstratethattheresultingfirehistoriesasdetectedabovearenotstatisticallydissimilar.TheprimarysensitivityoftheGapFiremodelislateseasonburnrate(seeAppendix1forfurtherexplanation),soitisthisvariablethatmustbedemonstrablysimilar(ie,forthepurposesofthistest,anearlyburnistreatedthesameasanoburn).Toassessthissimilarity,theexpectedlateseasonburnratesmustbecomputedforeachstratumasshowninTable6below.Table6:ExpectedLateSeasonBurnRatesStratumExpectedBurnedPixelsExpectedUn-burnedPixels1NPPA,1×BLPROBLateburn,1NPPA,1×(1-BLPROBLateburn,1)2NPPA,2×BLPROBLateburn,2NPPA,2×(1-BLPROBLateburn,2)3……4……5……6……Where:NPPA,i=totalnumberofpixelsintheprojectareainstratumiBLPROBLateburn,i=Baselineprobabilityoflateburningoccurringinstratumi(asdeterminedinStep5above)Theseexpectedburnratesmustbecomparedwithactualburnratesintheprojectareausingthechi-squaredgoodnessoffitstatisticon11degreesoffreedom(+2foreachadditionalstratum)atthe90percentconfidencelevel.Thisanalysisofsimilaritymustbeperformedfortheentireprojectareaeachtimetheprojectareaisadjusted(eg,whennewprojectactivityinstancesareadded).WheremorethanoneBRRhasbeenspecified,eachdiscreetareaofthenewprojectactivityinstancemustbecoupledtooneBRR.Wherethechi-squaredtestdoesnotshowsignificantdifferencesbetweentheprojectareaandtheBRR,thebaselinefirehistoryasdeterminedinStep5maybeused.Ifthechi-squaredtestisfailed,projectproponentshavetwooptions:(1)modifythebaselinefirehistoryand/or(2)modifytheBRR.OptionA:modifythebaselinefirehistorytousethemostconservativehistoryThismustbeimplementedonaperstratumbasis.WhereBLPROBLateburn,ifortheBRRishigherthanintheprojectarea,thelowerfigurefromtheprojectareashouldbeused.WhereBLPROBLateburn,ifortheBRRislowerthanintheprojectarea,nomodificationisnecessarybecausethefirehistoryintheBRRismoreconservativethanintheprojectarea.Wherethisisthecaseforallstrata,noactualmodificationisnecessary,andthefirehistoryasgivenfortheBRRshouldbeused.OptionB:ModifyTheBRR,byeitheradjustingtheBoundariesand/orSplittingIt.TheBRRmayonlybemodifiedwhenthebaselinefirehistoryisrevised(asperSection9.3.4).AtsuchtimesprojectproponentsmayamendtheboundariesinordertoderiveanewBRRthatVM0029,Version1.0SectoralScope14Page23passesthechi-squaredtestoffirehistorysimilarity.ProjectproponentsmayalsochoosetosplittheBRRandprojectareaintopairedmultipleprojectactivityinstances(seeSection8.1.1.3).TheanalysisofsimilaritymustberepeatedandpassedforeachBRRandcorrespondingprojectareaseparately.Iftheprojectproponentisaddingoneormorenewprojectactivityinstancestoagroupedproject,andthischangecausesthesimilaritytesttofail,theprojectproponentmustestablishoneormorenewBRRspairedonlywiththenewactivityinstances.8.1.1.7Step7:RuntheGapfireModelandDetermineBaselineEmissionsfromAbovegroundBiomassDegradationDueToFireSub-step7.1.Thebaselineearlyburningprobability,lateburningprobabilityandnoburningprobabilityforeachstratummustbeenteredintotheGapFiremodel,therebycreatingaseparateGapFirefileforeachstratum.Thesebaselineinputsintothemodelremainfixedforthevalidityofthebaselinefirehistory,andmustbeupdatedeverytimethebaselinefirehistoryisrevised.Themediancarbondensityforthestratum(fromTable4)mustalsobeentered.Themodelmustberunwithatleast100,000patches9.SeeAnnex1foradetaileddescriptionoftheGapFiremodelincludingabriefusermanual(Annex1,Section5).Theprojectproponentmustrepeatthisstepforeachstratum.Notethemodelwillcalculatethebaselineandprojectscenariossimultaneously.Projectscenarioinputsaretheearlyburningprobability,lateburningprobabilityandnoburningprobabilityasmonitoredeachyearbytheprojectproponent(seeSection9.2).Sub-step7.2.CalculateannualbaselineemissionsfromabovegroundbiomassdegradationfortheprojectbyapplyingEquation(3)(ie,bymultiplyingthehectaresineachstratumwiththebaselineaverageannualcarbonstockchangesperhectareoverthefirst10yearsofthesimulation).BEBMy=∑((BBMCi,sy=0–BBMCi,sy=10)/10×Ai,y×44/12)(3)Where:BEBMy=Baselineemissionsfromabovegroundbiomassdegradationinyeary(tCO2e)BBMCi,sy=0=CarbonstoredinabovegroundbiomassperhectareinstratumiinGapFiresimulationyear0inthebaselinescenario(tC/ha)BBMCi,sy=10=CarbonstoredinabovegroundbiomassperhectareinstratumiinGapFiresimulationyear10inthebaselinescenario(tC/ha)Ai,y=Areaofstratumiinyearywithintheprojectarea(ha)44/12=ConversionfactorfromtCintotCO2e9ThisamountofpatchescomfortablydealswithoutputvariabilityofindividualpatchesintheGapFiremodel(seeAppendix1)andensuresasmoothoutputcurve.VM0029,Version1.0SectoralScope14Page248.1.2BaselineEmissionsfromBiomassBurningBaselineemissionsfrombiomassburningarecalculatedusingthelatestversionoftheVCSmoduleVMD0013Estimationofgreenhousegasemissionsfrombiomassandpeatburning(E–BPB)equation1ofVMD0013(Aburn,i,tBi,tCOMFi)mustbesubstitutedwiththeterm(Ai,y(AvCMortBLi,y/CF)),suchthattheequationreads:BEBiomassburn,y=∑(Ai,y×(AvCMortBLi,y/CF)×Gg×10-3×GWPg)(4)Where:BEBiomassburn,y=BaselineemissionsofN2OandCH4frombiomassburninginyeary(tCO2e)Ai,y=Areaofstratumiinyearywithintheprojectarea(ha)AvCmortBLi,y=Averageannualcarbonstoredinbiomassdyingasaresultoftreemortalityoverthefirst10years’outputoftheannualGapFiremodelbaselinesimulationforstratumi(seeSection8.1.1.7)(tC/ha)CF=Carbonfractionofwoodybiomass(dimensionless)Gg=Emissionfactorforgasg;kgt-1drymatterburntGWPg=Globalwarmingpotentialforgasg;tCO2/tgasgg=1,2,3…GreenhousegasesTheadaptationofequation1ofVMD0013calculatesbaselineemissionsfrombiomassburningforeachstratumbasedonthebaselinebiomassmortalityinyeary(theaverageovera10-yearperiodistaken),ascalculatedbytheGapFiremodel.Thisapproachassumesdirectemissionsofnon-CO2gasesfrommortality,ratherthanfromthedeadwoodpool.8.1.3BaselineemissionsfromselectivetreeharvestingBaselineemissionsfromselectivetreeharvestingareconservativelynotquantified.Therefore:BEHarvest,y=0(5)8.2ProjectEmissionsProjectemissionsarecalculatedasfollows:PRy=PRBM,y+PEBiomassburn,y+PEHarvest,y(6)Where:PRy=Projectemissionsinyeary(tCO2e)PRBM,y=Projectemissionsfromabovegroundbiomassinyeary(tCO2e)PEBiomassburn,y=ProjectemissionsofN2OandCH4frombiomassburninginyeary(tCO2e)PEHarvest,y=Projectemissionsfromselectivetreeharvestinginyeary(tCO2e)ForthecalculationofPRBM,y,seeSection8.2.1.VM0029,Version1.0SectoralScope14Page25ForthecalculationofPEBiomassburn,y,seegotoSection0.ForthecalculationofPEHarvest,y,seeSection8.2.3.8.2.1ProjectEmissionsfromAbovegroundBiomassTheprojectproponentmustcalculatePRBM,yusingthefollowingsteps:8.2.1.1Step1:CalculatetheProbabilitiesofEarlyBurning,LateBurningandNoBurningTheprojectproponentmustcalculatetheprobabilitiesofearlyburning,lateburningandnoburningbyapplyingEquation7below.PRPROBEarlyburn,y=∑(FFEarlyburn,z,y×Areaz,y)/∑(Areaz,y)(7)Where:PRPROBEarlyburn,y=Projectprobabilityofearlyburninginyeary(fraction)FFEarlyburn,z,y=EarlyburningfirefrequencyinFMUzinyeary(%)Areaz,y=AreaofFMUzinyeary(ha)TheprojectproponentmustuseEquation7tocalculatePRPROBLateburn,yandPRPROBNoburn,y,mutatismutandis.NotethatPRPROBEarlyburn,y,PRPROBLateburn,yandPRPROBNoburn,yare(forpracticalandcost-savingreasons)notbrokendownperbiomassstratum.Thisispermittedsolongasthemonitoringdataisrepresentativeoftheforestswithintheprojectarea(asisrequiredbySection9.3below)10.8.2.1.2Step2:RuntheGapfireModelandDetermineProjectEmissionsSub-step1.1.Theprojectproponentmustenterthemonitoredearlyburningprobability,lateburningprobabilityandnoburningprobability(theyareidenticalforeachstratum)intotheentrytableineachGapFiremodelfilepreparedforeachstratainSection8.1.1.6.Themodelmuststartitssimulationintheyearofthesatelliteimagesutilizedforthepreparationofthemostrecentcarbondensitymap.Wherethisdateisbeforetheprojectstartdate,theprojectproponentmustenterbaselineburnprobabilitiesfortheyearsbeforetheprojectstartdate.Eachfilemustberunwithatleast100,000patches.Repeatthisstepforeachstratum.Sub-step1.2.Basedonthemostrecentcarbondensitymap(seeSection8.1.1.1),theprojectproponentmustdeterminethenumberofhectareswithineachstratum.Thismustbeupdatedeachtimetheprojectareaisexpanded.10ThisapproachisconservativebecauseanalysisoftheGapFiremodelshowedthatemissionreductionsandremovalsfromfiremanagementincreaseinalinearmannerwithincreasingbiomass,whileburnprobabilitiesdecreasewithincreasingbiomass.Therefore,bymonitoringtheaverageburnprobabilitiesacrossthestrataandinputtingtheseintothemodel,theburnprobabilitiesinthehigherbiomassstrataareover-estimated;thesestrataarewherealowerburnprobabilitywouldhaveyieldedthehighestemissionreductionsandremovals.VM0029,Version1.0SectoralScope14Page26Sub-step1.3.TheprojectproponentmustcalculatethecurrenttotalprojectcarbonstockbyapplyingEquation8below.CCSBM,y=∑((PBMCi,y×Ai,y)–(Adeg–Areg)×5(8)Where:CCSBM,y=Currenttotalcarbonstocksintheabovegroundbiomasspoolinyeary(tC)PBMCi,y=CarbonstoredinabovegroundbiomassperhectareinstratumiinGapFiresimulationyearyintheprojectscenario(tC/ha)Ai,y=Areaofstratumiinyearywithintheprojectarea(ha)Adeg=Projectareasthatdegradebelow5tC/haduetocarbonmaprevision(ha)Areg=Projectareasthatregenerateabove5tC/haduetocarbonmaprevision(ha)Sub-step1.4.TheprojectproponentmustcalculatetheannualprojectcarbonstockchangesfortheprojectbyapplyingEquation9below.PRBM,y=(CCSBM,y–CCSBM,y-1)×44/12(9)Where:PRBM,y=Projectremovalsoremissionsfromforestbiomassregenerationordegradationintheabovegroundbiomasspoolinyeary(tCO2e)CCSBM,y=Currenttotalcarbonstocksintheabovegroundbiomasspoolinyeary(tC).44/12=ConversionfactorfromtCintotCO2e8.2.2ProjectEmissionsfromBiomassBurningTheprojectproponentmustcalculateprojectemissionsfrombiomassburningusingthelatestversionoftheVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).Equation1ofVMD0013(Aburn,i,t×Bi,t×COMFi)mustbesubstitutedwiththetermAi,y(AvCMortPRi,y/CF),suchthatEquation1reads:PEBiomassburn,y=∑(Ai,y×(AvCMortPRi,y/CF)×Gg,i×10-3×GWPg)(10)Where:PEBiomassburn,y=ProjectemissionsofN2OandCH4frombiomassburninginyeary(tCO2e)Ai,y=Areaofstratumiinyearywithintheprojectarea(ha)AvCmortPRi,y=Averageyearlycarbonstoredinbiomassdyingasaresultoftreemortalityoverthefirst10years’outputoftheyearlyGapFiremodelprojectsimulationforstratumi(seeSection8.1.1.7)(tC/ha)CF=Carbonfractionofwoodybiomass(dimensionless)Gg,i=Emissionfactorforstratumiforgasg;kgt-1drymatterburntGWPg=Globalwarmingpotentialforgasg;tCO2/tgasgg=1,2,3...GreenhousegasesVM0029,Version1.0SectoralScope14Page27ThisadaptationofEquation1ofVMD0013calculatesprojectemissionsfrombiomassburningforeachstratumbasedontheprojectbiomassmortalityinyeary(theaverageovera10-yearperiodistaken),ascalculatedbytheGapFiremodel.Thisapproachassumesdirectemissionsofnon-CO2gasesfrommortality,ratherthanfromthedeadwoodpool.8.2.3ProjectEmissionsfromSelectiveTreeHarvestingWhereselectiveharvestingoccursintheproject,theassociatedemissionsmustbequantified.Theprojectdescriptionmustdescribehowharvestingismanaged,andspecifyhowitwillbemonitoredtoensuresustainability(ie,thatharvestedbiomassisnotgreaterthanregenerationcapacity)usingindustrystandardmeasuressuchasannualallowablecutandmeanannualincrement.PEharvest,y=PEwoodproc,y+PEMT,y-21toy-1(11)Where:PEharvest,y=Projectemissionsfromselectivetreeharvestsinyeary(tCO2e)PEwoodproc,y=Projectemissionsfromwoodprocessinginyeary(tCO2e).Assumesallwoodisprocessedinyearofharvesting(mustbecalculatedusingthemethodineitherSection8.2.3.1orSection8.2.3.2).PEMT,y-21toy-1=Projectemissionsfromthemedium-termwoodproductspool(remaininginthispoolbetween3and100years)8.2.3.1CalculationofProjectEmissionsfromWoodProcessingBasedOnBoleVolumePEwoodproc,y=∑(HVj,y×(WasteFactor+BEF–1)×WDj×CF×44/12)(12)Where:PEwoodproc,y=Projectemissionsfromwoodprocessinginyeary(tCO2e)HVj,y=Harvestedbolevolumeofspeciesjintheprojectareainyeary(m3)–monitoredWasteFactorj=Fractionofharvestedvolumeeffectivelyemittedtotheatmosphereduringwoodprocessing.Defaultis0.2411.BEF=Biomassexpansionfactor(dimensionless).Ratioofabovegroundbiomasstobiomassofbolevolume.WDj=Wooddensityofharvestedspeciesj(t/m3)CF=Carbonfractionofwoodybiomass(default=0.47)12(dimensionless)44/12=ConversionfactorfromtCintotCO2e11Winjumetal.,1998.12IPCC,2006VM0029,Version1.0SectoralScope14Page288.2.3.2CalculationofProjectEmissionsfromWoodProcessingBasedOnTreeVolumePEwoodproc,y=∑(TTVj(Dj,y)×((WasteFactorj/BEF)+BEF–1)×WDj×CF×44/12)(13)Where:PEwoodproc,y=Projectemissionsfromselectivetreeharvestsinyeary(tCO2e)TTVj(D,j,y)=Totalvolumeofharvestedtreesofspeciesj,calculatedusinganallometricequationortablederivedfromrecognizedindependentsource(eg,governmentvolumetables).Theinputvariableofsuchequationmustbethediameterofeachtreeharvested(Dj,y)ofspeciesjintheprojectareainyeary(cm).WasteFactorj=Fractionofharvestedvolumeeffectivelyemittedtotheatmosphereduringwoodprocessing.Defaultis0.2413.BEF=Biomassexpansionfactor(dimensionless).Ratioofabovegroundbiomasstobiomassofbolevolume.WDj=Wooddensityofharvestedspeciesj(t/m3)CF=Carbonfractionofwoodybiomass(default=0.47)14(dimensionless)44/12=ConversionfactorfromtCintotCO2e8.2.3.3CalculationofProjectEmissionsfromtheMedium-TermWoodProductsPoolPEMT,y=VEMT,j,y×WDj×CF×44/12(14)VEMT,j,y=(VMT,j,y-21+VMT,j,y-20+VMT,j,y-19+….+VMT,j,y-1)/20(15)Where:PEMT,y-21toy-1=Projectemissionsfromthemedium-termwoodproductspool(remaininginthispoolbetween3and100years)VEMT,j,y=Volumeofspeciesjinyearyinthemedium-termwoodproductspool(remaininginthispoolbetween3and100years)thatisemittedeachyearVMT,j,y=Volumeofspeciesjinyearythatentersthemedium-termwoodproductspool(remaininginthispoolbetween3and100years)WDj=Wooddensityofharvestedspeciesj(t/m3)CF=Carbonfractionofwoodybiomass(default=0.47)15(dimensionless)44/12=ConversionfactorfromtCintotCO2e13Winjumetal.,199814IPCC(2006)15IPCC(2006)VM0029,Version1.0SectoralScope14Page298.3LeakageTheprojectdescriptionmustpresentanestimateofactivityshiftingleakagethattakesintoaccountallfiredrivingactivities.Theestimatemustbebasedoninterviews,ruralappraisalsand/orotherlocalexpertknowledge.ThematrixcomputationoutlinedinSection8.3.1belowmustbeusedforthepurposeofestimatingtheproportionoftheprojectareathatisburnedoutsideoftheprojectareafromtheshiftingoffiredrivingactivities.Theestimatemustberevalidatedevery10years.TheprocedureinSection8.3.2calculatesestimatedleakageemissions.Thisleakagecalculationmustbedoneeveryyear.Totalestimatedleakagethatisequaltoorhigherthan5percentoftheproject’soverallemissionreductionsandremovalsmustbesubtractedinthefinalcalculationofnetemissionreductionsandremovalsinSection8.4.8.3.1MatrixComputationofFireDrivingActivityLeakageTheextentoftheareaburnedoutsidetheprojectareaasaresultofshiftingfiredrivingactivities(ABADm,fc)mustbequantifiedbycompletingthematrixbelow(Table7).Cellsthatarecalculatedareshadedinlightgrey;cellsthatrequireinputhavenoshading.Everycellisapercentagevalue.Thematrixmustbecompletedusingthestepsbelow:1)Listthemonthsofthedryseasonacrossthetop.2)EnterthefirefrequencyforeachmonthfromthebaselinefirehistoryinthecellslabeledBBRm%wheremisthemonthnumber.3)Computeforeachmonththecumulativebaselineburnrate(thesumofallburnratesuptoandincludingthatmonth),andthefutureburnrate(thelikelihoodthatareanotburnedalreadywillbeburnedlaterintheyear).4)Listthedifferentcausesoffires,andforeachmonthinserttheestimatedproportionoffiresstartedbythatcause(FSm,fc,)weightedbyareaaffected.Thesemustsumto100percentforeachmonth,andmustbeobtainedfrominterviews,ruralappraisalsand/orotherlocalexpertknowledge.5)Listthedifferentratesoffiredisplaceability(Dfc,)accordingtocauseoffireasdeterminedfrominterviews,ruralappraisalsandotherlocalexpertknowledge(notexpectedtovarybymonth).6)Calculate,foreachmonthandcauseoffire,theresultingexpectedproportionoffiresthatwillbedisplaced(FDm,fc).7)Calculatethetotaldisplacementofburnedareaforeachmonthandcauseoffire(BADm,fc)8)ReduceBADm,fcbytheproportionFBBRmthatwouldburninthefutureintheabsenceoftheproject.9)Theresultingproportion(ABADm,fc,)istheadditionalareathatwillbeburnedinmonthmasaresultofcauseoffirefc,asaproportionoftheprojectarea.Summingacrossallmonthsandcausesoffiresproducesafinalestimateoftheproportionofthetotalprojectareaaffectedbyfiredrivingactivityleakageinanygivenyear.VM0029,Version1.0SectoralScope14Page30Table7:MatrixtableforcalculatingtheextentoftheareaburnedoutsidetheprojectareaasaresultofshiftingfiredrivingactivitiesMonthMayJune…OctNovOverallBaselineBurnRateBBR1%BBR2%…BBR6%BBR7%∑BBRCumulativeBBRCBBRm=∑BBR𝑖m1……FutureBBRFBBRm=∑BBR−CBBRm……FireCausesFC1FS1,1%FS2,1%…FS6,1%FS7,1%FC2FS1,2%FS2,2%…FS6,2%FS7,2%………………FCnFS1,n%FS2,n%…FS6,n%FS7,n%Total∑FS1,fcn1=100%………DisplaceabilityFC1D1%FC2D2%……FCnDn%FiresDisplacedFC1FD1,1%=FS1,1%xD1%…FD6,1%=FS6,1%xD1%FC2FD1,2%=FS1,2%xD2%…FD6,2%=FS6,2%xD2%………………FCnFD1,n%=FS1,n%xDn%…FD6,n%=FS6,n%xDn%BurnedAreaDisplacedFC1BAD1,1%=FD1,1%xBRR1%…BAD6,1%=FD6,1%xBRR6%∑BADm,1mVM0029,Version1.0SectoralScope14Page31MonthMayJune…OctNovOverallFC2BAD1,2%=FD1,2%xBRR1%…BAD6,2%=FD6,2%xBRR6%∑BADm,2m…………………FCnBAD1,n%=FD1,n%xBRR1%…BAD6,n%=FD6,n%xBRR6%∑BADm,nmTotal∑BAD1,fcfc∑BAD2,fcfc…∑BAD6,fcfc∑BAD7,fcfc∑BADm,fcm,fcAdjustedBurnedAreaDisplaced(lesswhatwouldhaveburnedanyway)FC1ABAD1,1%=BAD1,1%x(1-CBRR1%)……∑ABADm,1mFC2…ABAD2,m%=BAD2,m%x(1-CBRR2%)…∑ABADm,2m…………………FCn……ABADn,7%=BADn,7%x(1-CBRR7%)∑ABADm,nmFinalTotal∑ABAD1,fcfc∑ABAD2,fcfc…∑ABAD6,fcfc∑ABAD7,fcfc∑ABADm,fcm,fc8.3.2CalculationofLeakageEmissionsLFDA,y=∑Ai,y×∑ABADm,fc×LEFnoburn-to-lateburn(16)Where:LFDA,y=Leakageemissionsfromshiftingoffiredrivingactivitiesinyeary(tCO2e)Ai,y=Areaofstratumiinyearywithintheprojectarea(ha)ABADm,fc=Areaadditionallyburned(thatwouldnototherwisehaveburned)outsidetheprojectareaasaresultoftheshiftingoffiredrivingactivitiesinmonthmasaresultofcauseoffirefc,expressedasaproportionoftheprojectarea(fromSection8.3.1)(%)LEFnoburn-to-lateburn=Leakageemissionfactorofareasadditionallyburned(thatwouldnototherwisehaveburned)asaresultoftheshiftingoffiredrivingactivities(tCO2e/ha)LEFnoburn-to-lateburnmustbecalculatedusingthefollowingsteps:1)Basedonthecarbondensitymap(Section8.1.1.1),determinetheaveragecarbondensity(tC/ha)withintheBRR(theassumptionismadethatfiredrivingactivitieswouldbeshiftedintotheBRR).VM0029,Version1.0SectoralScope14Page322)PrepareaseparateGapFirefileforthecalculationofLEFnoburn-to-lateburn,thatstartsitssimulationattheaveragecarbondensitydeterminedinStep1.Enterasthebaselineburnprobabilities:100percentnoburn,0percentearlyburnand0percentlateburn.Enterastheprojectburnprobabilities:0percentNoburn,0percentearlyburnand100percentlateburn.Runthemodelwithatleast100,000patches.3)BasedontheGapFireoutput,calculateLEFnoburn-to-lateburnasfollows:LEFnoburn-to-lateburn=((BBMCi,y-1–BBMCi,y)–(PBMCi,y-1–PBMCi,y))×44/12(17)Where:LEFnoburn-to-lateburn=Leakageemissionfactorofareasadditionallyburned(thatwouldnototherwisehaveburned)asaresultoftheshiftingoffiredrivingactivities(tCO2e/ha)BBMCi,y=CarbonstoredinabovegroundbiomassperhectareinGapFiresimulationyearyinstratumiinthebaselinescenario(tC/ha)PBMCi,y=CarbonstoredinabovegroundbiomassperhectareinGapFiresimulationyearyinstratumiintheprojectscenario(tC/ha)44/12=ConversionfactorfromtCintotCO2e8.4NetGHGEmissionReductionsandRemovalsNetGHGemissionreductionsandremovalsarecalculatedasfollows:NERRy=PRy–BEy–LFDA,y(18)Where:NERRY=NetGHGemissionsreductionsandremovalsinyeary(tCO2e)PRy=Projectnetremovalsoremissionsinyeary(tCO2e)BEY=Baselineemissionsinyeary(tCO2e)LFDA,y=Leakageemissionsinyeary(tCO2e)ThenumberofVCUstobeissuedtoaprojectinyearyiscalculatedasfollows:VCUy=NERRy–EDMUy–BufferDiscounty(19)BufferDiscounty=(BEBM,y–PRBM,y)×RFy(20)Where:VCUy=NumberofVCUstobecreditedtoaprojectinyearyNERRY=NetGHGemissionsreductionsandremovalsinyeary(tCO2e)EDMUy=EmissionsDetectedatMapUpdateinyeary(tCO2e)asspecifiedinSection9.3.4(expectedtobezero)BEBM,y=Baselineemissionsfromabovegroundbiomassdegradationinyeary(tCO2e)VM0029,Version1.0SectoralScope14Page33PRBM,y=Projectremovalsoremissionsfromforestabovegroundbiomassregenerationordegradationinyeary(tCO2e)RFy=Riskrating(determinedusingthelatestversionoftheVCSAFOLUNon-PermanenceRiskTool).9MONITORING9.1DataandParametersAvailableatValidationData/ParameterAi,yDataunithaDescriptionTheareaofstratumiinyearywithintheprojectareaEquations3,4,8,10SourceofdataCarbondensitymapandprojectareashapefilesValueAppliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedRecordprojectboundarieswithaGPSinthefieldoruploadexistingshapefilesintoaGIS.Overlayontothecarbondensitymap.PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsandremovalsCalculationofleakageemissionsCommentsN/AData/ParameterBEyDataunittCO2eDescriptionBaselineemissionsinyeary(tCO2e)Equations1,18SourceofdataCalculatedinEquation1(Section8.1)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedSumofbaselineemissionsfromabovegroundbiomass,biomassburningandtreeharvestingPurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBEBM,yDataunittCO2eDescriptionBaselineemissionsfromabovegroundbiomassdegradationinVM0029,Version1.0SectoralScope14Page34yeary(tCO2e)Equations1,3,20SourceofdataCalculatedinEquation3(Section8.1.1.7)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedSummationoftheemissionsfromabovegroundbiomassfortheareaofeachstratumwithintheprojectareaPurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBEBiomassburn,yDataunittCO2eDescriptionBaselineemissionsfrombiomassburninginyeary(tCO2e)Equations1,4SourceofdataCalculatedinEquation4(Section8.1.2)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedSummationoftheemissionsfrombiomassburningfortheareaofeachstratumwithintheprojectareaPurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBEHarvest,yDataunittCO2eDescriptionBaselineemissionsfromselectivetreeharvestinginyeary(tCO2e)Equations1,5SourceofdataCalculatedinEquation5(Section8.1.3)Valueapplied0JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedThebaselineemissionsfromselectivetreeharvestingwillconservativelynotbequantifiedPurposeofdataCalculationofbaselineemissionsCommentsN/AVM0029,Version1.0SectoralScope14Page35Data/ParameterBLPROBEarlyburn,iDataunittCO2eDescriptionBaselineearlyburningprobabilityoccurringinstratumi(fraction)Equations2SourceofdataCalculatedinEquation2(Section8.1.1.5)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedBaselinerelativeprobabilityofearlyburningperstratum,asaGapFiremodelinputvaluePurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBLPROBLateburn,iDataunittCO2eDescriptionBaselinelateburningprobabilityoccurringinstratumi(fraction)Equations2SourceofdataCalculatedinEquation2(Section8.1.1.5)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedBaselinerelativeprobabilityofearlyburningperstratum,asaGapFiremodelinputvaluePurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBLPROBNoburn,iDataunittCO2eDescriptionBaselineNoBurningProbabilityoccurringinstratumi(fraction)Equations2SourceofdataCalculatedinEquation2(Section8.1.1.5)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedBaselinerelativeprobabilityofearlyburningperstratum,asaGapFiremodelinputvaluePurposeofdataCalculationofbaselineemissionsCommentsN/AVM0029,Version1.0SectoralScope14Page36Data/ParameterEarliestpossibleburndateDataunitDay/MonthDescriptionThefirstdayoftheearlyburningseason.EquationsN/ASourceofdataDefaultdate,expertopinionValueappliedEitherthedayatthebeginningofthedryseasonatwhichthemonthlyrunningaveragerainfall(atleast5yearsofdata)firstfallsbelow33percentofpeakwetseasonrainfall;oranotherlocally-appropriatedatesupportedbycredibleexpertopinion.JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedAtresearchsiteKilwa,the33percentofpeakrainfallthresholdscoincidewiththedatesgenerallyusedforthestartandendofthedryseason,aswellaswithfirstdetectionoffiresatthestartofthedryseason(aroundDayofYear125,seeFig.12inAnnex1).Amonthly(28day)runningaverageisappropriatetosmoothoutspikesfromonlyafewyears’worthofdata.Theaveragemustbecomputedfromthe14daysrunninguptoandincludingthedateconcerned,andthe14daysthatfollowit.Duetotherecentrains,onlyafewfirescanexpecttoburnclosetothisdate,andthusallowinglocalflexibilityinitsdeterminationshouldnothaveasignificantimpactonemissionreductions.PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterEarlyburningseasoncut-offdateDataunitDayandmonthDescriptionThelastdayoftheearlyburningseason.Thecut-offdateischosenandfixedatprojectvalidationforthedurationofthecreditingperiod.EquationsN/ASourceofdataDefaultdate,expertopinionValueappliedDefaultdateisthe30thdayofJune(day181oftheyearor182inleapyears),butexpertopinionmayestablishaproject-specificcut-offdateasanalternative.JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedProject-specificcut-offdatesshouldapproximatetheaveragedateonwhichconditionsallowfirestoburnthroughthenight.Ontheonehand,itisimportantthatthedateisnotsettoolate,whichcouldincreasetheriskthattheproject’sfiremanagementactivitiesclosetothecut-offdatecauseunintentionaldamage.Ontheotherhand,itshouldnotbesettooearlyasthiswillgivetheprojecttoosmallawindowtocarryoutitsfiremanagementVM0029,Version1.0SectoralScope14Page37activities,aswellashampertheseactivitiesdueincreasedvegetationmoisture.Itisinevitablethatinindividualyearsvariationsintheeasingofheavyrainstogetherwithsubsequentweatherconditionswillshifttheactualdatewhenfireswillself-extinguish.However,therangeofvariablesistoogreattobeeasilytractabletoanalysis,anduseofafixedcut-offdategreatlyincreasesprojectmanageability.Expertestimationofthisdateisbestpractice.Suitableexpertsinclude,butarenotlimitedto,localgovernmentofficialsandecologicalresearchers.Theprojectproponentmayadditionallychoosetoconsultinternationalexpertsinlargescaleearlyburning.PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCommentsN/AData/ParameterEndofburningseasondateDataunitDay/MonthDescriptionThelastdayofthedryseason,markingthetypicalonsetoftherains.Thisdateischosenandfixedatprojectvalidationforthedurationoftheprojectcreditingperiod.EquationsN/ASourceofdataHistoricalmeteorologicaldata,expertopinionValueappliedEitherthedayattheendofthedryseasonatwhichthemonthlyrunningaveragerainfall(atleast5yearsofdata)firstrisesabove33percentofpeakwetseasonrainfall;oranotherlocally-appropriatedatesupportedbycredibleexpertopinion.JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedAtresearchsiteKilwa,the33percentofpeakrainfallthresholdscoincidewiththedatesgenerallyusedforthestartandendofthedryseason.Amonthly(28day)runningaverageisappropriatetosmoothoutspikesfromonlyafewyears’worthofdata.Theaveragemustbecomputedfromthe14daysrunninguptoandincludingthedateconcerned,andthe14daysthatfollowit.Atthispointinthedryseason,evenpoorlymanagedlandscapesarelikelytobesubstantiallyfragmentedbyfiresoccurringearlierinthedryseason,sothenumberandextentofverylateseasonfiresislikelytobelow.Instead,practicabilityofmonitoringisexpectedtobeagreaterconcernfortheprojectproponent(aftersignificantrainfallitbecomesmuchhardertowalkthroughtheforestanddetectpreviousburns),henceallowinglocalflexibilityinitsdeterminationshouldnotsignificantlyimpactemissionreductions,butcouldincreaseuptakeofthemethodology.VM0029,Version1.0SectoralScope14Page38PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCommentsN/AData/ParameterCountPixi,yDataunitCount(integervalue)DescriptionNumberofcountablepixelsinstratumiinyeary.Observationsareinvalidwhentheyareinconclusiveduetocloudcoverormissingdata.Equations2SourceofdataCompositemapofsatelliteimages(eg,LandsatorSPOT)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedThecompositemapismadebycompilingasmanyimagesaspossibleforthe10-yearperiodofanalysisandapplyingaburnedareasdetectionalgorithm(eg,byapplyingthespectralindexmethodasdescribedinBastarrikaetal,2013).PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterCountEarlyburni,yDataunitCount(integervalue)DescriptionNumberofcountablepixelsinstratumiinyearythatshowedburningintheearlyburningseason.Equations2SourceofdataCompositemapofsatelliteimages(eg,LandsatorSPOT)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedThecompositemapismadebycompilingasmanyimagesaspossibleforthe10-yearperiodofanalysisandapplyingaburnedareasdetectionalgorithm(eg,byapplyingthespectralindexmethodasdescribedinBastarrikaetal,2013).PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterCountLateburni,yDataunitCount(integervalue)DescriptionNumberofcountablepixelsinstratumiinyearythatshowedburninginthelateburningseason.Equations2VM0029,Version1.0SectoralScope14Page39SourceofdataCompositemapofsatelliteimages(eg,LandsatorSPOT)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedThecompositemapismadebycompilingasmanyimagesaspossibleforthe10-yearperiodofanalysisandapplyingaburnedareasdetectionalgorithm(eg,byapplyingthespectralindexmethodasdescribedinBastarrikaetal,2013).PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterCountNoburni,yDataunitCount(integervalue)DescriptionNumberofcountablepixelsinstratumiinyearythatshowednoburninginboththeearlyandthelateburningseason.Equations2SourceofdataCompositemapofsatelliteimages(eg,LandsatorSPOT)ValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedThecompositemapismadebycompilingasmanyimagesaspossibleforthe10-yearperiodofanalysisandapplyingaburnedareasdetectionalgorithm(eg,byapplyingthespectralindexmethodasdescribedinBastarrikaetal,2013).PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterBBMCi,syDataunittC/haDescriptionCarbonstoredinabovegroundbiomassperhectareinstratumiinGapFiresimulationyearyinthebaselinescenarioEquations3,17SourceofdataGapFiremodeloutputValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedTheGapFiremodelsimulatesthebiomassresponseofanensembleoftree-sizedforestpatchestodifferentfireregimes,inputasprobabilitiesofearlyburn,lateburnandnoburn,forboththebaselineandtheprojectscenarios.Asimulationisrunforeachstratum,whichisspecifiedbyinputtingthemediancarbondensityofthestratum.Theoutputisthesimulatedcarbonstockchangesovertime.PurposeofdataCalculationofbaselineemissionsCalculationofleakageemissionsVM0029,Version1.0SectoralScope14Page40CommentsN/AData/ParameterAvCMortBLi,yDataunittC/haDescriptionAverageyearlycarbonstoredinbiomassdyingasaresultoftreemortalityoverthefirst10years’outputoftheyearlyGapFiremodelbaselinesimulationforstratumiinyeary(seeSection8.1.1.7)Equations4SourceofdataGapFiremodeloutputValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedTheGapFiremodelsimulatesthebiomassresponse,includingmortality(bothbackgroundandfire-induced)ofanensembleoftree-sizedforestpatchestodifferentfireregimes.AnnualcarbonlossesareavailableasaGapFireoutput.PurposeofdataCalculationofbaselineemissionsCommentsN/AData/ParameterGgDataunitkgt-1drymatterburntDescriptionEmissionfactorforgasgEquations4,10SourceofdataSectionIIIandannex2ofVCSmoduleVMD0013,Estimationofgreenhousegasemissionsfrombiomassburning(E-BB).ValueappliedUsevaluesfor‘Tropicalforest’JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedMiombowoodlandisatypeoftropicalforest.Thevaluesfor‘Savannahandgrassland’relatetotheburningofgrasses.PurposeofdataCalculationofbaselineandprojectemissionsfrombiomassburningCommentsData/ParameterGWPgDataunittCO2/tgasgDescriptionGlobalwarmingpotentialforgasgEquations4,10SourceofdataDefaultvaluesfromIPCCSARVM0029,Version1.0SectoralScope14Page41ValueappliedCO2=1;CH4=21;N2O=310JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedN/APurposeofdataCalculationofbaselineandprojectemissionsfrombiomassburningCommentsData/ParameterNPPA,iDataunitCount(IntegerValue)DescriptionTotalnumberofpixelsintheprojectareainstratumiEquationsSection8.1.1.6SourceofdataCarbondensitymapandprojectareashapefilesValueappliedN/AJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedOverlaythecarbondensitymapwiththeprojectareashapefiles.Countthepixelsonthecarbondensitymappertainingtostratumithatfallwithintheprojectareashapefiles.PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCommentsData/ParameterWasteFactorDataunitDimensionlessconstantDescriptionFractionofextractedbiomasseffectivelyemittedtotheatmosphereduringproduction.Equations12,13SourceofdataWinjumetal.1998indicatethattheproportionofextractedbiomassthatisoxidized(burningordecaying)fromtheproductionofcommoditiesisequalto19percentfordevelopedcountries,24percentfordevelopingcountries.WWisthereforeequaltoCXB,tymultipliedby0.19fordevelopedcountriesand0.24fordevelopingcountries.Valueapplied0.24JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedDerivedfromVCSmoduleVMD0005Estimationofcarbonstocksinthelong-termwoodproductspool(CP-W)VM0029,Version1.0SectoralScope14Page42PurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterBEFDataunitDimensionlessconstantDescriptionBiomassexpansionfactor.Equations12,13SourceofdataThesourceofdatamustbechosenwithpriorityfromhighertolowerpreferenceasfollows:(a)Existinglocalforesttype-specific;(b)Nationalforesttype-specificoreco-region-specific(egfromnationalGHGinventory);(c)Foresttype-specificoreco-region-specificfromneighboringcountrieswithsimilarconditions.(d)Globalforesttypeoreco-region-specific(eg,IPCC2006InventoryGuidelines,AFOLU,Chapter4,Table4.5)ValueappliedVariableJustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedN/APurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterCFDataunittC/tDryMatterDescriptionCarbonfractionofwoodybiomassEquations4,10,12,13,14,15SourceofdataIPCC2006,AFOLU,Chapter4,Table4.3Valueapplied0.47JustificationofchoiceofdataordescriptionofmeasurementmethodsandproceduresappliedTheIPCCisconsideredawidelyacceptablesourceofdefaultvalues.PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCommentsN/AVM0029,Version1.0SectoralScope14Page439.2DataandParametersMonitoredData/ParameterAbovegroundtreebiomassDataunittC/haDescriptionAbovegroundtreebiomassisdeterminedinsampleplotsforthepurposeofvalidationofthecarbondensitymapEquationsN/ASourceofdataFieldmeasurementsDescriptionofmeasurementmethodsandprocedurestobeappliedMeasurementsmustbetakeninsampleplots,eachnotsmallerthan0.5hainsize,andshouldbedistributedacrossthelandscapeinanunbiasedmanner.Abovegroundtreebiomasswithdiameter>5cmmustbemeasuredusingscientificbestpractice,mustcomplywithVCSrequirementsforaccuracyandprecisionandmustbecalculatedusinganallometricequationthatis1)publishedinpeer-reviewedscientificliteratureand2)justifiablyappropriatefortheecologicalconditionsintheprojectarea.Ifalternativefielddataisused,justificationmustbeprovidedtoshowhowitisequivalenttoorexceedstheserequirements.Frequencyofmonitoring/recordingWhenthecarbondensitymapisfirstmadeandeverytimeitisrevisedQA/QCprocedurestobeappliedScientificcommonpracticemustbeusedtocheckforimplausibledataentries(eg,identificationandelimination,correctionorre-measurementofoutliers).Atleast5percentoftreesmustbere-measuredanddatacompared(re-measurementmayhavetakenplaceatanearlierorlaterdate,solongasbothmeasurementsareavailablewhenthecarbondensitymapismade).Theaverageplot-scalemeasurementplausibilityrate(afterallowingforexpectedtreegrowth)mustbeatleast85%.Wherethemeanmeasurementplausibilityrateislessthan90%inoneormoreplotstheseplotsshouldbeincludedinsubsequentanalysisifitisconservativetodoso(ie,lowergrowthisdetected)andexcludedwheretheywouldleadtogreatercarbongainsbeingdetected.PurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCommentsN/AData/ParameterAi,yDataunithaDescriptionTheareaofstratumiinyearywithintheprojectareaEquations3,4,8,10VM0029,Version1.0SectoralScope14Page44SourceofdataCarbondensitymapandprojectareashapefilesDescriptionofmeasurementmethodsandprocedurestobeappliedRecordprojectboundarieswithaGPSinthefieldoruploadexistingshapefilesintoaGIS.Overlayontothecarbondensitymap.Frequencyofmonitoring/recordingEverytimetheprojectareaandthecarbondensitymaparerevisedQA/QCprocedurestobeappliedActionsbytheoperatormustbedouble-checkedandsignedoffbyasupervisorforcorrectnessPurposeofdataCalculationofbaselineemissionsCalculationofprojectemissionsCalculationofleakageemissionsCommentsN/AData/ParameterAreaz,yDataunithaDescriptionTheareaofFMUzinyearywithintheprojectareaEquations7SourceofdataProjectareashapefilesDescriptionofmeasurementmethodsandprocedurestobeappliedRecordFMUboundarieswithaGPSinthefieldoruploadexistingshapefilesintoaGISFrequencyofmonitoring/recordingEverytimetheprojectareaisrevisedQA/QCprocedurestobeappliedActionsbytheoperatormustbedouble-checkedandsignedoffbyasupervisorforcorrectnessPurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterFFEarlyBurn,z,yDataunitFraction(dimensionless)DescriptionEarlyburningfirefrequencyinFMUzinyearyEquations7SourceofdataFieldobservationsDescriptionofmeasurementmethodsandprocedurestobeappliedSeeSection9.3.2VM0029,Version1.0SectoralScope14Page45Frequencyofmonitoring/recordingFirstmonitoringeventofeachyearQA/QCprocedurestobeappliedGeo-locateddigitalphotosmustbetakenateachcheckpoint,andarandomselectionofatleast5percentmustbecheckedafterwardsforaccuracy(burn/no-burn).PurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterFFLateBurn,z,yDataunitFraction(dimensionless)DescriptionLateburningfirefrequencyinFMUzinyearyEquations7SourceofdataFieldobservationsDescriptionofmeasurementmethodsandprocedurestobeappliedSeeSection9.3.2Frequencyofmonitoring/recordingSecondmonitoringeventofeachyearQA/QCprocedurestobeappliedGeo-locateddigitalphotosmustbetakenateachcheckpoint,andarandomselectionofatleast5percentmustbecheckedafterwardsforaccuracy(burn/no-burn).PurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterFFNoBurn,z,yDataunitFraction(dimensionless)DescriptionNoburningfirefrequencyinFMUzinyearyEquations7SourceofdataFieldobservationsDescriptionofmeasurementmethodsandprocedurestobeappliedSeeSection9.3.2Frequencyofmonitoring/recordingSecondmonitoringeventofeachyearQA/QCprocedurestobeappliedGeo-locateddigitalphotosmustbetakenateachcheckpoint,andarandomselectionofatleast5percentmustbecheckedafterwardsforaccuracy(burn/no-burn).VM0029,Version1.0SectoralScope14Page46PurposeofdataCalculationofprojectemissionsCommentsN/AData/Parameter:PBMCi,syDataunittC/haDescriptionCarbonstoredinabovegroundbiomassperhectareinstratumiinGapFiresimulationyearyintheprojectscenarioEquations8,17SourceofdataGapFiremodeloutputDescriptionofmeasurementmethodsandprocedurestobeappliedObtainedfromtheGapFiremodelrunwithfirefrequencyparametersPRPROBEarlyBurn,y,PRPROBLateBurn,yandPRPROBNoBurn,yascalculatedinEquation7.Frequencyofmonitoring/recordingAnnuallyQA/QCprocedurestobeappliedN/APurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterAdegDataunithaDescriptionAreathathasdegradedbelowtheminimum5tC/hathresholdEquations8SourceofdataCarbondensitymapDescriptionofmeasurementmethodsandprocedurestobeappliedAtcarbondensitymaprevision,countallpixelsinsidetheprojectareathatpreviouslyhadabiomassover5tC/habutarenowbelowthatminimum.Frequencyofmonitoring/recordingOnceevery10years,ormorefrequentwherecarbondensitymapisupdatedsoonerthanevery10yearsQA/QCprocedurestobeappliedN/APurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterAvCmortPRi,yDataunittC/haVM0029,Version1.0SectoralScope14Page47DescriptionAverageannualcarbonstoredinbiomassdyingasaresultoftreemortalityoverthefirst10years’outputoftheannualGapFiremodelprojectsimulationforstratumi(seeSection8.1.1.7)Equations10SourceofdataGapFiremodeloutputDescriptionofmeasurementmethodsandprocedurestobeappliedObtainedfromtheGapFiremodelrunwithfirefrequencyparametersPRPROBEarlyBurn,y,PRPROBLateBurn,yandPRPROBNoBurn,yascalculatedinEquation7.Frequencyofmonitoring/recordingAnnuallyQA/QCprocedurestobeappliedN/APurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterAreg,yDataunithaDescriptionAreasthatpreviouslydegradedbelowthe5tC/hathreshold,butthatinyearyregenerateabovethisthresholdEquations8SourceofdataCarbondensitymapDescriptionofmeasurementmethodsandprocedurestobeappliedAtcarbondensitymaprevision,countallpixelsinsidetheprojectareathathadoncebeenoverthe5tC/hathreshold,thenonapreviousrevisionhadbeenfoundtohavedegradedbelowthe5tC/habiomassthreshold,buthavesinceregeneratedsuchthatbiomassisonceagainover5tC/ha.Frequencyofmonitoring/recordingOnceevery10years,ormorefrequentwherecarbondensitymapisupdatedsoonerthanevery10yearsQA/QCprocedurestobeappliedN/APurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterHVj,yDataunitm3DescriptionHarvestedbolevolumeofspeciesjintheprojectareainyearyEquations11SourceofdataFieldmeasurementsandvolumecalculationsVM0029,Version1.0SectoralScope14Page48DescriptionofmeasurementmethodsandprocedurestobeappliedOfeachharvestedtree,recordthespecies.Takediametermeasurementsatthebaseandthetopoftheboleandcalculatetheaveragediameter(Daverage).Measurethelength(l)ofthebole.HVj,y=(Daverage/2)2×π×lFrequencyofmonitoring/recordingEachtimeatreeisharvestedQA/QCprocedurestobeappliedAtleast5percentoftreesmustbere-measuredanddatacomparedPurposeofdataCalculationofprojectemissionsCommentsN/AData/ParameterDt,j,yDataunitcmDescriptionDiameteratBreastHeightofharvestedtreetofspeciesjinyearyEquations13SourceofdataFieldmeasurementsDescriptionofmeasurementmethodsandprocedurestobeappliedDiametermeasuredatBreastHeightbeforethetreeisharvestedusingameasuringtapeFrequencyofmonitoring/recordingEachtimeatreeisharvestedQA/QCprocedurestobeappliedAtleast5percentoflogmeasurements(randomlychosen)shouldbecheckedagainstrecordedtreediameterforconsistency.Sincelogvolumeisthebasisforsaleprice,ForestManagersareincentivizedtomaximizelogdiameter,andthusonecanbeconfidentthatdataisnotbiasedonthelowside(ie,isconservativewithrespecttoemissionsreductionsachievedafterdeductingprojectemissionsandremovals).PurposeofdataCalculationofprojectemissionsandremovalsCommentsN/AData/ParameterWDjDataunittdrymatter/m3DescriptionWooddensityofharvestedspeciesjEquations12,13,14SourceofdataFieldmeasurementsoravailableliteratureDescriptionofWhereliteraturesourcescanbefoundforWDj,theprojectVM0029,Version1.0SectoralScope14Page49measurementmethodsandprocedurestobeappliedproponentmaychoosetousethepublishedvaluesorestablishtheirownvalues(seebelow).Thesourceofliterature-deriveddatamustbechosenwithpriorityfromhighertolowerpreferenceasfollows:(a)Nationalspecies-specificorgroupofspecies-specific(eg,fromnationalGHGinventory);(b)Species-specificorgroupofspecies-specificfromneighboringcountrieswithsimilarconditions.(c)Globalspecies-specificorgroupofspecies-specific(eg,IPCC2006INVGLs,AFOLU,Chapter4,Tables4.13and4.14).Species-specificwooddensitiesmaynotalwaysbeavailable,andmaybedifficulttoapplywithcertaintyinthetypicallyspeciesrichforestsofthetropics.Therefore,itisacceptablepracticetousewooddensitiesdevelopedforforesttypesorplantfamiliesorspeciesgroups.Whereliteraturesourcescannotbefoundforspecies-specificWDj,theprojectproponentmustestablishWDjofharvestedspeciesjasrecordeduponharvestingbythewaterdisplacementmethod,asfollows:1)Takeapieceofwoodfromthemiddleoftheharvestedbole.2)Oven-drythepieceofwoodusingstandardprocedurestoachievewoodmoisturecontentbelow15percent.3)Establishweight(W)ofthedrypieceofwood.4)Putwaterinacontainerthatallowsforaccuratewatervolumemeasurement.Measurethevolume(V1)ofwaterinthecontainer.5)Submergethedrypieceofwoodfullyinthewater.Measurethevolumeofwater(V2)inthecontaineragain.6)ThevolumeofthepieceofwoodisV=V2–V17)WDj=W/V8)Convertthemeasurementunitstot/m3Frequencyofmonitoring/recordingOnce,whenthefirsttreeofspeciesjisharvestedQA/QCprocedurestobeappliedRepeatmeasurementofWDjseveraltimestoestablishanaveragevaluewithinthe90percentconfidenceinterval.PurposeofdataCalculationofprojectemissionsandremovalsCommentsN/AData/ParameterVMT,j,yDataunitm3DescriptionVolumeofspeciesjinyearythatentersthemedium-termwoodVM0029,Version1.0SectoralScope14Page50productspool(remaininginthispoolbetween3and100years)Equations15SourceofdataTimbersalesrecords,interviewswithtimberbuyersandprocessors,expertopinionandcrediblesourcesDescriptionofmeasurementmethodsandprocedurestobeappliedOfeachharvestedtreespeciesrecordthemostlikelyendusesanddeterminetheirrelativeproportionsoftheharvestedvolume.Includethevolumethatgoesintoenduseswithalife-timebetween3and100yearsintoVMT,j,yFrequencyofmonitoring/recordingEachtimeanyvolumeofspeciesjisprocessedand/orsoldQA/QCprocedurestobeappliedContactdetailsofintervieweesmustbestoredsothatlikelyendusesandtheirrelativeproportionscanbeindependentlyverified.PurposeofdataCalculationofprojectemissionsandremovalsCommentsN/AData/ParameterFSm,fcDataunit%DescriptionProportionoffiresstartedbyfirecausefcinmonthm,weightedbyareaaffected.EquationsMatrixcomputationofleakageemissionsSourceofdataInterviews,ruralappraisalsandotherlocalexpertknowledge.DescriptionofmeasurementmethodsandprocedurestobeappliedTheproportionsofallidentifiedfirecausesfcmustsumto100percentforeachmonth.Frequencyofmonitoring/recordingAnnuallyQA/QCprocedurestobeappliedN/APurposeofdataCalculationofleakageemissionsCommentsN/AData/ParameterDfcDataunit%DescriptionRateoffiredisplaceabilityoffirecausefcEquationsMatrixcomputationofleakageemissionsSourceofdataInterviews,ruralappraisalsandotherlocalexpertknowledge.DescriptionofmeasurementmethodsTheratesarenotexpectedtovarybymonthoryearVM0029,Version1.0SectoralScope14Page51andprocedurestobeappliedFrequencyofmonitoring/recordingAnnuallyQA/QCprocedurestobeappliedN/APurposeofdataCalculationofleakageemissionsCommentsN/A9.3DescriptionoftheMonitoringPlan9.3.1CarbonDensityMapValidationThecarbondensitymapmustbeground-truthedusingfielddataappropriateforthispurpose.Datamustbecollectedfromsampleplotsthatcanbeeasilyrelatedtopixelsintheremotelysenseddata.Sampleplotsshouldthereforebelocatedusingsomebasicstratificationtoensurebroadcoverageoftherangeofforestdensities.Sufficientsampleplotsmustbesurveyedtobootstraptheclassificationfunction,andotherstoassessuncertaintywhilstgeneratinganarrowenoughconfidencelimitthatmeetstherequirementsofSection8.1.1.1.Duetotheheterogeneityinmiombovegetationcover,mainsampleplotsmustcoveratleast0.5ha.However,thesurveydesignmayincorporatesmallersub-plotsinwhichagreaterrangeofinformationiscollected(eg,onsmallerstemsizes),andwhichcanbeusedtoimprovetheclassificationofremotelysenseddata.Thesizeofthelargestplotineachclustermustbelargerthanthedesiredmapresolution(atleasttwiceaslargeineachdirection).AbovegroundbiomassintreeswithDBH>5cm,andstemdensityforsaplings1-5cmDBH,mustbemeasuredusingscientificbestpracticeandmustcomplywithVCSrequirementsforaccuracyandprecision.Biomassvaluesmustbecalculatedusinganallometricequationthatis1)publishedinpeer-reviewedscientificliteratureand2)justifiablyappropriatefortheecologicalconditionsintheprojectarea.169.3.2AnnualMonitoringofFireFrequenciesMonitoringtheprojectscenariomustbeundertakenbydirectfieldobservations,sinceestimatesderivedfromremotesensingdatagenerallyhavetoowideanerrormarginforspecificprojectsites.Duetotheunpredictabilityoffirespread,eachforestmanagementunit(FMU)mustbeseparatelymonitored.16NotethattheGapFiremodelitselfusesthetreeallometryderivedbyRyanetal.(2011)fromN’HambitainMozambique.SubstitutingalternativeallometriesintheGapFiremodelisonlyfeasibleiftheyincludeleafarea(whichmanydonot).However,usingdifferentallometriesinthecarbondensitymapcomputationandGapFiremodelisnotafundamentalproblem,andoflessimportancethanusingthemostappropriateallometricrelationshipinvalidatingthecarbondensitymap.VM0029,Version1.0SectoralScope14Page52Monitoringmusttakeplacetwiceayear,attheendoftheearlyburningseason,andattheendofthelateburningseason.Monitoringmusttakeplacewithinonecalendarmontheithersideoftheearlyburningseasoncut-offdateandendofburningseasondate,respectively.Ifeithermonitoringrequirementismissedorincomplete,thennoemissionsreductionsmaybeclaimedforthatFMUthatyear,andforfutureyearsthebaselinefireregimeshouldbeassumedtohavetakenplace.WheremonitoringiscompleteforsomebutnotallFMUs,onlythemonitoredFMUsmayclaimemissionreductions.FMUswheremonitoringisnotcompletedtwoyears’runningmustbevisuallyassessedfordeforestationorsignificantdisturbanceattheendofthesecondyearsincetheprojectstartdateintheFMU,orsincethepreviousmonitoringevent,whicheverislater.Failuretodosowillleadtotheautomaticassumptionthatallabovegroundcarbonstocksintheareahavebeenemitted.ThelosseventmustbereportedtoVCSinaccordancewithVCSrequirements.Thesameoccursifthevisualassessmentdetectsdeforestationorsignificantdisturbance.MonitoringfirefrequencymustbeundertakenusingasetofcheckpointsdistributedthroughouttheforestintheFMU.Thesurveyormustassessthegroundimmediatelybeneaththeirfeetateachcheckpoint.Thesurveyorrecordsafireifsomeorallofthatsmallpatchofgroundhasblackenedduetopassageoffire.Ifnosuchsignsareevident,thennofireisdetected.Thefirefrequencyisthenspecifiedasthepercentageofcheckpointsatwhichafirewasrecorded(thedetectionrate).Inthecaseoftheearlyburningseasonmonitoringperiod,allfiresareassumedtobenewfires,andnoadjustmentsarerequired.Inthecaseofthelateburningseasonmonitoringperiod,thefirefrequencyisdefinedasthedifferencebetweenthetwopercentages:FFlateseasononly=FFlateseasondetectionrate–FFearlyseasondetectionrate(21)Notethatthemonitoringofburnprobabilitiesdoesnotconsidercarbondensitystrata,asexplainedinSection8.2.1.1.9.3.2.1DistributionofFireMonitoringCheckpointsAtleast86checkpointsarerequiredacrosstheentireprojectareasoastoensurethe95percentconfidenceintervalisnowiderthan15percent17.Inpracticeitisrecommendedatleast100checkpointsbesurveyedineachFMUeachmonitoringperiod,asacost-effectivewaytominimizeerrorsbeingpassedthroughtothemodel.Thesecheckpointsmaybeeitherpermanentlyestablishedornewlygeneratedineachmonitoringperiod,orsomemiddleoption(eg,checkpointsmaybeunmarkedbutonpermanenttransects,suchthatroughlythesamesetofpointsismonitoredeachyear).Theprojectproponentwhochoosestousepermanentorsemi-permanentlylocatedcheckpointsmustjustifythischoice,andshowhowtheyarenotsubjecttoabiasedorunrepresentativefiremanagementorothertreatmentcomparedtotherestoftheFMU.17Fromthebinomialdistribution,assumingthattheprobabilityofaburnisnotcloseto0or1,theminimumis43forasinglemonitoringexercise.SinceFFlateseasononlydependsontwomonitoringexercisesthen86checkpointsisaconservativeminimumtoensuretheconfidenceintervalisnowiderthan15percent.VM0029,Version1.0SectoralScope14Page53Checkpointsmaybelocatedeitherthroughapurelyrandomprocess,orthroughapartlyrandomprocesswhichtakesintoaccountlogisticalconstraintstosomedegree.Wheretheprocessisnotfullyrandom,theprojectproponentmustclearlydescribethestepstakentoavoidaccidentalbias(eg,transectorplotlocationsshouldbegeneratedwithoutreferencetobiophysicaldatabeyondtheminimumrequiredfortheprocess(suchasFMUboundaries)).Exceptinthecasewhereanewsetofcheckpointsisgenerateddirectlyandpurelyrandomlyeachmonitoringperiod,checkpointsandtheirspatialunderpinning(eg,transectlines)shouldbeevenlydistributedthroughouttheforest.Checkpointlocationshouldalsonotbebiasedwithrespecttobiomassortootherrelevantvariablessuchasaltitudeandsteepnessofslope(whichaffectfirespread).Theprojectdescriptionmustsetoutspecificproceduresandcriteriaforensuringthatthiscriterionismet.Wherearandomcheckpointgenerationprocessfailsthesetests,theprojectproponentmayeitheraddadditionalcheckpoints,ormoveexistingonessoastocomewithintherepresentativenesscriteria(solongastheprocesstodosoisitselfclearlyrandomandnotinformedbydetailedknowledgeoftheFMU).Indefiningsuchprocedures,itisimportanttonotetheGapFiremodelproducesestimatesofemissionsreductionsthatscaleroughlylinearlywithstartingbiomass.Inthissituationitisnotnecessarytoensureminimumrepresentationfromeachstratum,solongasthedistributionofbiomassatcheckpointsisstatisticallysimilartotheFMUpopulationinbothmeanandvariance.9.3.2.2CheckingforForestLossThebaselinecarbondensitymapmustberevisedatleastevery10years(seesection9.3.3below),butcatastrophicfiresandotherforestlosseventsfromunplannedlandusechangecouldhappenatanytime.Ifthetwice-yearlymonitoringoffirefrequenciesdetectssuchoccurrences,thentheareaofanyaffectedareamustberecordedinthefieldwithaGPS.Theareaofeachcarbondensitystratumpresentintheaffectedareamustbedeterminedfromthecarbondensitymap.Itwillbeassumedthatallabovegroundcarbonstocksintheareawillhavebeenemitted.ThelosseventmustbereportedtoVCSinatimelymanner.Theprojectproponentmaycomplementthiswithadditionalmonitoringthatmaybelessregularorrigorousindesign.9.3.3RevisionofCarbonDensityMapThismethodologyrequiresabiomasscarbondensitymapthatisnevermorethan10yearsold.Thus,themapmustbeupdatedatleastevery10years,thoughtheprojectproponentmaychoosetoupdatemorefrequentlyiftheysowish.Toupdatethecarbondensitymap,thesameproceduresasgiveninsections8.1.1.1to8.1.1.3abovemustbefollowed.Duringrevisionofthecarbondensitymap,projectareasthatareidentifiedashavingdegradedbelowthelowerthresholdvalueintheprojectscenariomustbeexcludedfromsubsequentmodelingofcarbonfluxes(ie,nofurtherclaimsmaybemadeonthoseareas).Itmustbeassumedthatallabovegroundtreecarbondensityisemittedinstantly(Equation8,Section8.2.1.1).Inasubsequentrevisionofthecarbondensitymap,wheresuchexcludedVM0029,Version1.0SectoralScope14Page54areasaredeterminedtohaveregeneratedagainabovethe5tC/hathreshold,theymaybere-includedinthemodelingofcarbonfluxes,atwhichpointtheirabovegroundtreecarbondensityistreatedasaninstantaneoussequestrationandisaddedtotheproject’semissionreductions(Equation8).Forprojectareasthatregenerateabovethehigherthresholdvalueintheprojectscenarioduringtheprojectcreditingperiod,thesespecificareasmayremainwithintheproject.EligibleareasmustbeidentifiedaspixelsonacarbondensitymapasperSection8.1.1.1.Beforestratifyingandusingthenewcarbondensitymap,thenewvaluesmustbecomparedtopredictedmodeloutputintheProjectArea.Todothis,therevisedcarbondensitymapmustbeoverlaidwiththeexistingprojectarea.Then,calculatethefollowingfortheprojectarea:TPBC=TotalPredictedabovegroundBiomassCarbon(tonnes)derivedfromGapFiremodelpredictionstodateTABC=TotalActualabovegroundBiomassCarbon(tonnes)derivedfromtherevisedcarbondensitymapλTABC=95percentupperconfidencelimitofTABCComparingthesefiguresleadstooneofthreepossibleoutcomes:TABC>TPBCThetotalnewbiomassintheProjectAreaisgreaterthanthetotalbiomasspredictedbytheGapFiremodel.Inthiscasethemodelpredictionswereconservative.Noactionisrequired,andthenewcarbondensitymapmaybeadopted.TABC≤TPBC≤λTABCThetotalbiomasspredictedbytheGapFiremodeliswithinthemarginoferroronthecarbondensitymap.Noactionisrequired,andthenewcarbondensitymapmaybeadopted.TPBC>λTABCThenewtotalbiomassissubstantiallylessthanpredicted,andoutsidethemarginoferror.Theprojectmustaccountfortheloss.Inthethirdcase,theprojectmustregisteralosseventintheyearthatthenewcarbondensitymapwasconstructedasperEquation22.EmissionsDetectedatMapUpdate,EDMU=(TABC–TPBC)×44/12(22)IfEDMUislessthanthetotalclaimofavoidedemissions,NERRy,ascalculatedinstep8.4,thendeductEDMUfromthetotal.IfEDMUexceedsNERRy,reportthelosseventinaccordancewithVCSrequirements.AlldeviationsfromthemodelshouldbereportedtotheUniversityofEdinburghsothattheycanbeincorporatedintomodelrevisions.VM0029,Version1.0SectoralScope14Page55StratifythenewcarbondensitymapasperSection8.1.1.2,andusethemapasthebasisforGapFiremodelpredictionsforthefollowing10years(ie,thesearelikelytodifferfromthepreviousstrata).Areasofthemapthatwerepreviouslyabovetheminimumcarbondensityallowableinthemethodology,buthavesincedroppedbelowit,mustbeexcludedfromfuturecrediting(asperSection5.2).However,whereitcannotbedemonstratedfrommonitoringthatthecauseofthereductionincarbondensitywasalandusechangeeventoracatastrophicevent,suchareasmuststillbeincludedintheanalysisofthenextcarbondensitymapupdateandthecalculationofEDMU.Where,atthatpoint,thebiomasshasrisenabovetheminimumbiomassallowableinthemethodology,suchareasmayonceagainbeincorporatedintotheprojectforcreditingofemissionsreductions.Areasofthemapthatwerepreviouslybelowthemaximumbiomassallowableinthemethodology,buthavesincerisenabovethatthresholdmayremain,sincealthoughfuturefirefrequenciesmaybeassumedtobemuchlower,evenwithoutsustainedfiremanagement,allfuturegainsmadearetheresultofpreviousprojectaction,andthusclaimable.AdditionalstratashouldbeaddedtothoselistedinSection8.1.1.2,Table4,at5tonintervalstoaccountforsuchincreases.9.3.4DecadalRevisionofFireHistoryBaselineRevalidatethebaselineevery10yearsusingoneofthefollowingoptions:1)Provideaqualitativejustificationshowinghowlocalfirepracticeshavenotchangedsignificantly(eg,frequencyofburning,andlateseasonfiresinparticularhavenotsignificantlydecreased).Adeclineinearlyseasonfiresisallowablesolongasitisoffsetbyanequalorlargerincreaseinlateseasonfires.Thisjustificationmustbesupportedwithcredibleevidenceandbylocalexperts.Wherethereisreasontobelievethatlocalfirepracticesmighthaveimprovedsoastoreducetheseverityofthebaselinescenario,thisoptionforrevisingthefirehistoryisnotpermitted.2)RedothefirehistoryderivedfromburnscaridentificationonsatelliteimagesasspecifiedinSection8.1.1.5.Duetotheprojectactivitiesimplementedovertheprevious10years,thismayinvolvedeterminationofanewBRRthatexcludesbothprojectareas,andotherareasthatcouldbeprotectedfromburningbyprojectactivities(ie,a“fireshadow”effectinwhichprojectareasblockfireroutes,thusreducingfirefrequencyinareasdownwindoftheprevailingwinddirection).ThisnewBRRmustsatisfyalltherelevantcriterialistedinSection8.1.1.3,andmustbecoveredinthebiomassmap.Notethatitisnotnecessarytoupdatethebiomassmapandthefirehistorybaselineatthesametime.However,iftheBRRhasbeenexpanded,thismaynecessitatealargerbiomassmap.VM0029,Version1.0SectoralScope14Page5610REFERENCESArchibald,S.,Lehmann,C.E.,Gómez-Dans,J.L.,andBradstock,R.A.2013.Definingpyromesandglobalsyndromesoffireregimes.ProceedingsoftheNationalAcademyofSciences,110(16),6442-6447.Bastarrika,A.,Chuvieco,E.,andMartín.M.P.2013"MappingburnedareasfromLandsatTM/ETM+datawithatwo-phasealgorithm:Balancingomissionandcommissionerrors."RemoteSensingofEnvironment115.4:1003-1012.http://www.ehu.es/aitor.bastarrika/AitorDocs/Bastarrika_2011a.pdfBridges,E.M.1990.WorldGeomorphology.CambridgeUniversityPress,UK.ESAClimateChangeInitiative,2011:Fire,Userrequirementsdocument,http://www.esa-fire-cci.org/webfm_send/264Frost.P.1996.TheecologyofMiombowoodlands.In:B.Campbell(ed)TheMiomboinTransition:WoodlandsandWelfareinAfrica.CFIOR,Bogor.Giglio,L.,Loboda,T.,Roy,D.P.,Quayle,B.andJustice,C.O.2009.Anactive-firebasedburnedareamappingalgorithmfortheMODISsensor.RemoteSensingofEnvironment113:408–420GOFC-GOLD2012,Asourcebookofmethodsandproceduresformonitoringandreportinganthropogenicgreenhousegasemissionsandremovalsassociatedwithdeforestation,gainsandlossesofcarbonstocksinforestsremainingforests,andforestation.GOFC-GOLDReportversionCOP18-1,(GOFC-GOLDLandCoverProjectOffice,WageningenUniversity,TheNetherlands).Hauf,H.2012.MitigatingforestfiresinKilwaDistrict,Tanzania:Aninvestigationofanthropogenicdrivers.Studentdissertation,UniversityofOxfordDepartmentofContinuingEducation.49pp.IPCC.2006,2006IPCCGuidelinesforNationalGreenhouseGasInventories,Volume4:Agriculture,Forestry,andOtherLandUse.PreparedbytheNationalGreenhouseGasInventoriesProgramme,EgglestonH.S.,BuendiaL.,MiwaK.,NgaraT.andTanabeK.(eds).Published:IGES,Japan.http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.htmlIsango,J.A.,2007.StandStructureandTreeSpeciesCompositionofTanzaniaMiomboWoodlands:ACaseStudyfromMiomboWoodlandsofCommunityBasedForestManagementinIringaDistrict.WorkingPapersoftheFinnishForestResearchInstitute50:43–56Stronach,N.2009.Areviewoftheimpactoffireonthecarboncontent,dynamicsandbiodiversityvalueofmiombowoodlands.ReportforFauna&FloraInternational.Werger,M.J.A.andB.J.Coetzee.1978.TheSudano-ZambezianRegion.M.J.A.Werger,editor.BiogeographyandEcologyofSouthernAfrica.W.Junk,TheHague.VM0029,Version1.0SectoralScope14Page57White,F.1983.ThevegetationofAfrica,adescriptivememoirtoaccompanytheUNESCO/AETFAT/UNSOVegetationMapofAfrica(3Plates,NorthwesternAfrica,NortheasternAfrica,andSouthernAfrica,1:5,000,000.UNESCO.Paris.Winjum,J.K.,Brown,S.andSchlamadinger,B.1998.Forestharvestsandwoodproducts:sourcesandsinksofatmosphericcarbondioxide.ForestScience44:272-284VM0029,Version1.0SectoralScope14Page58APPENDIX1:THEGAPFIREMODELThisappendixwaspreparedbySamBowersandProf.MathewWilliamsoftheUniversityofEdinburgh.A1.1DefinitionsAGBAbovegroundbiomass,thebiomassoftreestems(incl.crowns)AllometricequationsTreeallometrydescribesthequantitativerelationsbetweenkey,easytomeasurecharacteristicsoftreesandothermoredifficulttoassessproperties.HerewerelateDBHtotreemorphology.ChronosequenceAseriesofforestsiteswithsimilarattributesthathavebeenregrownfollowingclearanceandabandonment,allowingtheanalysisofforestprocessesthatoccuroverlongtimescalesECMWFEuropeanCentreforMediumRangeWeatherForecasts,sourceoftheERA-interimglobalclimatereanalysisdatasetFireintensityAmeasureoftheenergyreleasedbyafire,closelyrelatedtoitsexpectedimpactsuponawoodlandecosystemFRIFirereturninterval,themeantimeperiodbetweenfiresinyearsGPPGrossPrimaryProductivity,therateatwhichorganismscaptureandstoreenergythroughphotosynthesisMODISModerate-ResolutionImagingSpectroradiometer,aNASAopticalremotesensinginstrumentwithadataarchivecoveringtheperiod1999-presentNDVINormalisedDifferenceVegetationIndex,ameasureofplantgreennesscalculatedusingmeasurementsofnearinfraredandredsurfacereflectanceNhambitaMiombowoodlandfieldsiteincentralMozambiqueNPPNetPrimaryProductivity,GPPminustheenergylosttorespirationPSPPermanentsampleplot,awoodlandplotusedtomeasureandmonitorvegetationstatusVM0029,Version1.0SectoralScope14Page59Top-killAbovegroundstemmortality,whichisdecoupledfromrootstockmortalityduetotheimportanceofresproutinginfireproneecosystemsA1.2IntroductionInordertounderstandtheimpactsofforestdisturbanceoverlongtimeperiods,modellingtechniquesmaybeusedinplaceofdirectmeasurementsofdifficulttodetectforestdegradation.AdynamicmodelofforestbiomassnamedGapFirehasbeendevelopedtosimulatetheimpactsofavarietyoffireregimesonthestructureofmiombowoodlands.Themodelforecaststhegrowthandmortalityofindividualtreesunderfuturefireregimesbasedonanensembleoftree-sizedwoodlandpatches.Whenrununderarangeoffirereturnintervals(FRIs)andfireintensities,themodelprovidesapredictionofwoodlandresponsetoanalteredfireregime.Althougharelativelysimplemodel,GapFirehasbeenshowntobeproficientatcapturingthecomplexdynamicsofmiombowoodlanddisturbance(RyanandWilliams,2011).Thisreportprovidesadetaileddescriptionofhowthemodelisstructuredandoperated,includingdiscussionofprincipalassumptionsmadeandtheirimpactonmodelpredictions.Throughoutthemodellingprocessanemphasishasbeenplacedonutilizingallavailablefielddatafrommiombowoodlands.Asdataonthedynamicsinsomeareasofmiombowoodlandgrowth,mortalityandfirearelimited,themodelensuresthatpredictionsleadtoconservativeestimatesofbiomasschangesinplaceofabest-guessestimateofmiombowoodlandresponse.A1.3ModelOverviewThedeathoftreesinnaturalwoodlandsresultsinsmall‘gaps’openinginthetreecanopy.Theseopeningsinotherwiseshadedforestallowforlighttopenetratethroughtheforestcanopy,stimulatingthegrowthofsmalltreestore-populatetheopening.Byexplicitlymodellingthegrowth,mortalityandregenerationoftreesattheseforestgaps,thebalancebetweencarbonsequestrationthroughforestregrowthandcarbonemissionfromwoodlanddisturbancemaybepredictedasFigure1).Gapdynamicshavebeenwidelyutilizedbyforestlandmanagersfortheircloserelationtomanypracticalforestapplications.Gapmodelsareparticularlyappropriatefordescribingmiombowoodlandswherewoodlandstructureisprofoundlymodifiedbyfrequentdisturbanceevents.Themodelissimilarinconceptiontomanyothergapmodels(ShugartandSmith,1996,Williams,1996)thathavebeenusedpreviouslyinmiombowoodlands(Desanker,1996,DesankerandPrentice,1994)andalsotosimulatetheeffectsoffire(MillerandUrban,2000).Agap-modellingapproachhasadvantagesfromitsexplicitrepresentationofpopulationstructureandvariability,itsabilitytoexplorethestochasticnatureofdisturbanceeventsthroughlargemodellingensembles,andbecauseitallowsdensity-dependentfeedbacksongrowththroughlightcompetition.ThemodelpresentedhereinamodifiedarrangementofthepublishedmodelofRyanandWilliams(2011)forthepredictionofbiomasschange.VM0029,Version1.0SectoralScope14Page60Figure1:DiagrammaticRepresentationofGapfire,ShowingPatchInitialization,GrowthinResponsetoPhotosyntheticallyActiveRadiation(PAR),FireInducedMortalityandRegenerationGapFirehasbeendevelopedtomodelstems,theirlightenvironment,phenology,mortalityandresultantcarbonfluxesasdescribedinfigure2below.Themodelsimulatesindividual‘patches’withanarearoughlyequivalenttothecanopyofalargetree(0.02ha).Lightisinterceptedbytheleavesofeachstemataraterelatingtotheheightofleavesinthewoodlandcanopy.Leafareaiscalculatedasafunctionofstemdiameter,andvariesthroughtheyearwithobservationsofleafphenology,restrictinggrowthtothewetseasonwhensoilmoistureisplentiful.Interceptedlightisconvertedtocarbonusingamiombo-specificlightresponsecurve.Assimilatedcarbonisallocatedtorespiration,leafandfine-rootformation,withremainingcarbonallocatedtoincreasingstemandlargerootbiomass.Asstemsincreaseinbiomasstheirmorphology(DBH,leafarea,canopystructure)isalteredaccordingtoaseriesofallometricmodelsdevelopedformiombowoodlands(see2.4below).Followingannualgrowth,stemsareexposedtoachanceoftop-killfromintrinsicsourcesorasaresultoffire,whereprobabilityoftop-killcanberelatedtostemsize,frequencyandintensityoffire.Eachyearwoodlandpatchesregenerate,withtheresproutingoftop-killedrootstocksandtherecruitmentofnewseedlings.Aspatchesaresmall,andmortalityandregenerationprocessesarestochastic,thebiomasschangeinasinglepatchwillnotrepresentthetrendsofmiombowoodlandasawhole,hencethemodelisrunasanensembleofmanypatches,withmeantrendsrepresentativeofthenetchangeinmiombowoodlandcarbonstorage.Thefollowingtextgivesadetailedoverviewofinitialization,growth,mortalityandregenerationinthemodel.Themodelissummarizedinaschematicdiagramasdescribedinfigure2below.PARPatchInitialisationGrowthMortalityRegenerationNextyearVM0029,Version1.0SectoralScope14Page61Figure2:GapFiremodelschematicParametersareshowninred,allometriccalculationsareshowningreen,andmodeldrivingdataisshowninblue.Stochasticprocessesareindicatedwithadashedline.Themodelproducesannualoutput,butthelightandcarbonassimilationfunctionsrunonhourlytimesteps.Aspatchesareverysmallandmortalityandregenerationprocessesarestochastic,themodelrunsasanensembleofmanypatches.A1.3.1InitializationExtensivestemsizedatagatheredfrompermanentsampleplots(PSPs)inKilwaareusedtosetupmodelpatcheswitharepresentativestemsizedistribution(SSD).Ofthe25monitoredPSPs,asubsetof17wereselectedfittingthemiombowoodlandcriteria(5-35tC/ha,noclosedcanopy).Thesedatacomprisearangeofbiomassmeasurements(5.2-33.9tC/ha),covering17haofdataforNextyearInitializationStemsizedistributionCalculateverticalleafareadistributionSeasonalcycleofLAIPARabsorbed(k)andCassimilated(Pmax,kp)Photosyntheticallyactiveradiation(PAR)Callocatedtoleaves(LCA),fineroots(Afr)GrowthCallocatedtowoodybiomassMortalityIntrinsic(Mi)orfire-induced(kearly,λearly,klate,λlate)RegenerationResprouting(Sresprout,Snew,Smort)andrecruitment(Precruit)Firefrequency(Early/Late)OutputCalculateabovegroundCstocksCrespired(Ra)loopoverstemsVM0029,Version1.0SectoralScope14Page62stemsizes5-40cmDBHand170haofstems>40cmDBH.Thedatawerefittedwithtwoexponentialtrendlines,withabreakpointat15cmDBHtooptimallyfitobservationsofstemsizesasdescribedinFigure3below.Fromthesetrends,twoprobabilitydensityfunctionsweregeneratedfromwhichstemsizesmayberandomlyallocatedtotreesinmodelpatches.Aminimumstemsizeof1cmDBHwasimposedtomatchthesizethataseedlingmayreachfollowingoneyearofgrowth.Anupperlimitof100cmDBHwassettopreventthegenerationofimprobablylargestems.ItisassumedthattherelativefrequenciesofstemsizeclassesrecordedinKilwaarerepresentativeofwoodlandsacrosstheEasternmiomboecoregion.Whereasthemodeldoesallowforinitializationusinglocallygeneratedstemsizedistributions,therelativeinsensitivityofthemodeltosmallsteminputsandanalteredsizedistributionoflargestemsmeanthatthiswillbeunlikelytoimprovemodelperformance.Onthisbasis,largeandsmallstemsizedistributiondatafromKilwashouldremainfixedinthemodel.ThePSPsweregenerallylocatedinmiombowoodlandsthatwererelativelyundisturbedbylogging,charcoalmakingorotheractivitiesthatcouldsignificantlyaffectthedistributionofstems.Forthisreasonthemodelisnotapplicableincomparativelydisturbedwoodlands,aspertheapplicabilityconditionsofthemethodology.Figure3:CharacteristicStemSizeDistributionofaMiomboWoodland,FittedwithTrendsforStems<15CmDBHand≥15CmDBHDatainlight-greywerenotincludedintrend-fittingasdataforthesestemsizesarelimited.Dataforstemsizes<5cmDBHwerenotroutinelycollectedatthePSPsandtrendsareextrapolatedbackward.The5-6cmsizeclassisalsoexcludedasstemsnearthelowerboundofmeasurementareoftenundercountedinsampleplots.Stems>60cmDBHarerare,sotrendsareextrapolatedforward.Abovegroundbiomass(>5cmDBH)iswellrelatedtolargestemdensity(Dlarge)(linearregression:R2=0.67,P<0.001);accordinglythemodelvarieslargestemdensitytoattainarangeofinitialabovegroundbiomassvalues.Largestemsarerelativelyrare(87±36stems/ha),andoccurinfrequentlyatpatch-scale,thereforestemsareallocatedtopatchesbasedonaPoissondistribution.Littleevidencewasfoundthatsmallstemdensity(Dsmall)atthePSPsvariessystematicallytoalargeVM0029,Version1.0SectoralScope14Page63degreewithbiomass(linearregression:R2=0.26,P=0.0564).Smallstemdensitywasthereforesettothemeandensityof835±176stems/ha,basedonanormaldistributionofstemdensities.Theallocationofrandomnumbersoflargestemstoeachpatchleadstoawiderangeofvegetationstructures,encompassingtherangeofbiomassvariabilityinmiombowoodlandand‘gaps’.Thisentirerangeofvariationismodeledforeachstratumwithnumbersoflargestemsscaledappropriately.Thismechanismassumesthatinthe5-35tC/habiomassrangeinmiombowoodlandsbiomassdifferencesarepredominantlyafunctionoflargestemdensity.However,aslargestemsoccurrarely,itisnecessarytorunlargenumbersofmodelensemblestoensurearepresentativesampleofinitialpatchbiomasses.A1.3.2GrowthGrowthinGapFireisregulatedbythelightenvironmentofeachstem,whichisdeterminedfromthecanopystructureasdeterminedbyallometricequations(see2.4below).Stemgrowthisfurthermoderatedbyleafphenology,respirationandmaintenancecostsoftreesineachpatch.A1.3.2.1CanopyStructureAteachpatch,allometricequationsareappliedtorelatestemDBHtotree-topandcanopybaseheights.Treecanopydepth(Tdepth)iscalculatedasthedifferenceofthesetwovalues,givingtheverticalspaceamongstwhicheachtree’sleavesaredistributed.Afurtherallometricequationrelatesbasalareaofeachstemtoitsleafarea.Thecanopyofeachpatchisrepresentedas25onemetredeeplayers,wherelayersarepopulatedwiththeleafareaassociatedwitheachstem,assumingthatleavesareuniformlydistributedinthecanopylayersoverTdepth.Thetotalleafareaofeachlayer(LAlayer)issummedtocalculateleafareaindex(LAI):𝐿𝐴𝐼=∑𝐿𝐴𝑙𝑎𝑦𝑒𝑟𝑔𝑎𝑝_𝑎𝑟𝑒𝑎25𝑙𝑎𝑦𝑒𝑟=1(1)wheregap_area=0.02ha(200m2),approximatelytheareacoveredbythecanopyofasinglematuretree.A1.3.2.2LightenvironmentLightavailabilityateachcanopylayerisestimatedusinganapplicationoftheBeer-Lambertlaw(Jones,1992).TheBeer-Lambertlawdescribestheattenuationoflightthroughatreecanopy,wherethetopcanopylayerexposedtoallincomingPhotosyntheticallyActiveRespiration(PAR)anddeeperlayersareshadedbythoseabove.PARisabsorbedaccordingtotheLAlayer,assumingasphericalleafangledistribution(k=0.5)andthatallradiationisdiffuseandleaveshavenotransmittanceoralbedo.HourlyestimatesofPARfrommeasurementsinNhambitaareusedtodrivephotosynthesisinthemodel.GrowthisrestrictedtothewetseasonbyphenologyinputsfrommonthlymeasurementsofLAIinNhambita,expressedasafractionofpeakLAI(LAI_frac).PARabsorbedbyleavesatthetopcanopylayeroverayear(PARmaxlayer)iscalculatedfromthetotalincomingPAR(PARin):𝑃𝐴𝑅𝑚𝑎𝑥𝑙𝑎𝑦𝑒𝑟=𝑃𝐴𝑅𝑖𝑛∙(1−𝑒−𝑘∙𝐿𝐴𝑙𝑎𝑦𝑒𝑟∙𝐿𝐴𝐼_𝑓𝑟𝑎𝑐)(2)PARtransmittedthroughtolowerlayers(PARthru)iscalculatedbydifferencingPARinandPARmaxlayer.VM0029,Version1.0SectoralScope14Page64𝑃𝐴𝑅𝑡ℎ𝑟𝑢=𝑃𝐴𝑅𝑖𝑛−𝑃𝐴𝑅𝑚𝑎𝑥𝑙𝑎𝑦𝑒𝑟(3)Lightabsorptionatlowerlayers(PARlayer)iscalculateddownthroughthefurther24canopylayers:𝑃𝐴𝑅𝑙𝑎𝑦𝑒𝑟=𝑃𝐴𝑅𝑡ℎ𝑟𝑢∙(1−𝑒−𝑘∙𝐿𝐴𝑙𝑎𝑦𝑒𝑟∙𝐿𝐴𝐼𝑓𝑟𝑎𝑐)(4)𝑃𝐴𝑅𝑡ℎ𝑟𝑢=𝑃𝐴𝑅𝑙𝑎𝑦𝑒𝑟+1−𝑃𝐴𝑅𝑙𝑎𝑦𝑒𝑟(5)Thisproducesaverticalprofileoflightabsorptionineachpatchthrougheachyear,whichdrivesthephotosynthesisandgrowthofeachstem.A1.3.2.3CarbonAssimilationGrowthisdeterminedseparatelyforeachstem,withabsorbedPARconvertedtoassimilatedcarbonusingphotosyntheticlightresponsecurves.Twoparametersdescribethelightresponsecurve:themaximumrateofassimilation(Pmax)andtheintensityoflightneededtoachievehalfthisrate(kp).Foreachstem,massofphotosynthateissummedforeachcanopylayerforeachhourofthe12diurnalcyclesrepresentativeofeachmonth,andscaleduptoayearlytotal.Thegrossprimaryproductivityofeachstemateachlayerforeachhourofdaylight(GPPi)iscalculatedasaproportionofmaximumphotosyntheticrate(Pmax)andthetotalleafareaofeachtree(LAtree):𝐺𝑃𝑃𝑖=[𝑃𝑚𝑎𝑥∙𝑃𝐴𝑅𝑙𝑎𝑦𝑒𝑟𝑃𝐴𝑅𝑙𝑎𝑦𝑒𝑟+𝑘𝑝∙𝐿𝐴𝐼_𝑓𝑟𝑎𝑐]∙𝐿𝐴𝑡𝑟𝑒𝑒(6)ThesumofGPPioveralllayersovertheentireyeargivesthetotalcarbonfixationbyeachtreefromphotosynthesis(GPP).Netprimaryproductivityofeachstem(NPP)iscalculatedasGPPminusthefractionofcarbonthatisrespiredbytheplant(Ra):𝑁𝑃𝑃=𝐺𝑃𝑃∙(1−𝑅𝑎)(7)Carbonrequiredfortheyearlyreplacementofleaves(Cleaf)iscalculatedasafunctionofleafcarbonperunitarea(LCA)andthetotaltreeleafarea(LAtree):𝐶𝑙𝑒𝑎𝑓=𝐿𝐴𝑡𝑟𝑒𝑒∙𝐿𝐶𝐴(8)Cleafandfinerootcarbon(Cfr)aredeductedfromNPP,leavingcarbonallocatedtowoodybiomass(Cwood).𝐶𝑤𝑜𝑜𝑑=𝑁𝑃𝑃−𝐶𝑙𝑒𝑎𝑓−𝐶𝑓𝑟(9)VM0029,Version1.0SectoralScope14Page65Carbonallocationforeachtreeispartitionedtoaboveandbelowgroundpoolsinproportionshoot_frac,followinganallometricrelationship.Stemcarbonallocation(Cstem)isthereforecalculatedas:𝐶𝑠𝑡𝑒𝑚=𝑠ℎ𝑜𝑜𝑡_𝑓𝑟𝑎𝑐∙𝐶𝑤𝑜𝑜𝑑(10)StemgrowthiscalculatedastheannualincreaseinCstem,whichisrelatedtoanequivalentincreaseinDBHbyanallometricmodel.Thisseriesofcalculationsaccountingforgrowtharerepeatedforeachstem,producingayearlyestimateofgrowthineachwoodlandgap.A1.3.3MortalityGapFiremodelsstemmortalityinresponsetofireoccurrenceandintrinsic(non-fire)sources.Becauseoftheimportanceofresproutinginfire-proneecosystems(BondandMidgley,2001,Chidumayo,2004,MlamboandMapaure,2006),abovegroundstemmortality(top-kill)isdecoupledfrombelowgroundrootstockmortality.Themortalitymoduleaccountsforstemtop-kill,whereasrootstockmortalityiscalculatedwithplotregeneration(seeSection3.4).Firesarestratifiedintoearlyburnsandlateburns,whichoccurwithprobabilityP(early)andP(late).A1.3.3.1Fire-inducedTop-killStemsaretop-killedeachyearbyeitherintrinsicsources,suchassenescence,herbivory,elephantdamageandtimberextraction,orbyfire.Inayearthatfiredoesnotoccur,intrinsicmortalityprobability(Mi)issetataconstantrateforallsizeclasses.Inthecaseoffire,stemtop-killratesarederivedfromfieldexperimentsinMozambique,whichshowedstemtop-killtobeafunctionofstemdiameterandthermalanomaly(RyanandWilliams,2011).Asstemdiameterincreases,thicknessofprotectivebarkincreasesproportionally(Jacksonetal.,1999,SutherlandandSmith,2000,JohnsonandMiyanishi,2001),offeringgreaterresilienceastreesincreaseinsize.Inlargerstems(>10cmDBH),thiseffectsaturates,andfurtherincreasesinDBHprovidenoadditionalprotectionfromfire.Thisrelationshipismodelledwithasigmoidalfunction,withsaturationforlargerstems:𝑙𝑜𝑔𝑜𝑑𝑑𝑠𝑜𝑓𝑡𝑜𝑝-𝑘𝑖𝑙𝑙=−𝑎𝑥𝐷𝐵𝐻+𝑏𝑥whereDBH<10cm(11)𝑙𝑜𝑔𝑜𝑑𝑑𝑠𝑜𝑓𝑡𝑜𝑝-𝑘𝑖𝑙𝑙=𝑏𝑥_𝑠𝑎𝑡𝐷𝐵𝐻whereDBH>10cm(12)Variablesax,bx,andbx_sataredeterminedbyfireintensity.FirelineIntensity(FLI)isthereleaseofheatenergyperunittimeperunitlengthoffirefront(kW/m)(Byram,1959).Itiswidelyutilizedbecauseitisrelativelyeasytomeasure,anditisknowntobesignificantlycorrelatedtobiologicallyimportantfireimpactsincludingtreetop-killandmortality(Alexander,1982).AcontinuousrelationshipbetweenfireintensityandmortalityparametersisdeterminedfrommeasurementsofstemmortalityandFLIfromtheNhambitafireexperiments(Saitoetal.,inprep):𝑎𝑥=−7.025∙ln(𝐹𝐿𝐼)−13.112(13)VM0029,Version1.0SectoralScope14Page66𝑏𝑥=2.119∙ln(𝐹𝐿𝐼)−12.451(14)𝑏𝑥_𝑠𝑎𝑡=−123.5∙𝐹𝐿𝐼−0.498(15)AminimummortalityrateissettothevalueofMiforallfireintensities.Thisresultsinacontinuousrelationshipbetweenfireintensityandtop-killasdescribedinFigure4.Figure4:Fire-InducedStemTop-KillRateswithStemDBHandFirelineIntensity(FLI)A.3.3.2FireFrequencyFirefrequencyisinputasamodeldriver,andoccursatrandomintervals.Ateachyear,earlyandlatefireoccurswithprobabilitiesP(early)andP(late)areinputasamodeldriversfromwhichtheincidenceoffireisrandomlydeterminedforeachpatcheachyear.Fireincidencesaremodeledasindependentrandomevents,withnodependencyacrosstimeandspace(individualpatcheshavenospatialrelationshipwithotherpatches).A1.3.3.3FireSeasonalityItiswidelyrecognizedthatlowintensityearly-seasonfiresaremuchlessdamagingthanhighintensitylate-seasonfires(Campbell,1996,Hoffaetal.,1999,GoldammerandDeRonde,2004,Robertson,1993).GapFirestratifiesfiresintoearlyandlatecategories,eachwithacharacteristicrangeofintensities.FireintensityisdescribedbytwoWeibulldistributions,specifiedbyparameterskearly,λearly,klate,λlate.Thesetwodistributions(showninFigure6below)incorporatethevariabilityoffireintensitythatisexpectedofwildfiresandthedifferencesbetweenaverageearlyandlatefireintensity.VM0029,Version1.0SectoralScope14Page67ForeachmodeledfireincidenceinapatchtheFLIisrandomlygeneratedfromtheapplicableWeibullprobabilitydistributionfunction.A1.3.4RegenerationTop-killedstemsaboveaminimumsizeclass(Sresprout)haveaprobabilityofresproutingwhenkilled(1-Smort),whereSmortisprobabilityofrootstockmortalitybasedondatafromtheNhambitafireexperiments.Itisassumedthatsproutsreach2cmDBHintheirfirstyearofgrowth.Additionallythereisachanceofarecruitmenteventoccurring(Precruit),whereanumberofnewseedlings(Snew)areestablishedinapatch,reaching1cmDBHintheirfirstyearofgrowth.Theinitialgrowthofseedlingsisassumedtobesmallerthansproutsduetotheestablishedsproutrootstocksprovidingenergyforincreasedgrowth.A1.3.5OutputFollowingthegrowth,mortalityandregenerationateachpatchateachyear,carbonstocksaresummedusingtheallometricrelationbetweenstemDBHandabovegroundbiomass.Althoughallstemsaremodeled,themodelonlycalculatesthebiomassofstemswithDBH>5cmduetouncertaintiesinthedynamicsofsmallstemsandtomatchmeasurementprotocolsusedattheKilwaPSPs.AllometricmodelsareusedtoproduceestimatesofLAIandrootstockbiomass,whichinconjunctionwithbasalareaandstockingdensitycountsmaybeusedasmodelvalidation.A1.4ParameterEstimationandModelSensitivityParametervalueswerederivedfromtheliterature,fieldmeasurementsandassessedwithreferencetolocalknowledgeandprofessionalexperience.Allparametersareunchangedfromtheoriginallypublishedmodel(RyanandWilliams,2010)withtheexceptionofmaximumphotosyntheticrate(Pmax).ModelsensitivitywasevaluatedwithrespecttochangestonominalparametervaluesasdescribedintableA1.Wherethemodelisfoundtobesensitivetoaparameterchange,anassessmentismadeabouthowwellconstrainedtheparameterisbyavailabledata.Modelparametersareconsideredintwocategories:parametersdetermininggrowthandthosecontrollingmortalityandregeneration.VM0029,Version1.0SectoralScope14Page68TableA1:ModelParameterValues,TheirSource,NominalValuesandSensitivitySxIntenseFireNoFireParameterdescriptionParameternameNominalparametervalue,PnS0.5S0.75S1.5S2S0.5S0.75S1.5S2SourceofnominalparametervalueGrowthParametersFractionofGPPusedforautotrophicrespirationRa0.54.53.63.21.8-3.6-3.4-2.6-1.5(Waringetal.,1998)ExtinctioncoefficientforBeer-Lambertlawk0.5-0.1-0.60.00.0-0.1-0.20.00.0(NormanandCampbell,1991)AmountofCallocatedtofineroots,asafractionofallocationtoleavesCfr10.80.31.00.7-0.8-0.7-0.6-0.6(Hendricksetal.,2006,Castellanosetal.,2001)Maximumrateofphotosynthesis(µmolC·m-2·s-1)Pmax12-3.1-3.4-4.3-4.82.62.83.74.3(TuohyandChoinski,1990,Tuohyetal.,1991,Woollen,2013)PARintensityatwhich0.5P-maxisobtained(µmol·s-1·m-2)kp2503.11.91.51.1-2.7-2.1-1.1-0.9(TuohyandChoinski,1990,Tuohyetal.,1991,Woollen,2013)Leafcarbonperleafarea(gC·m-2)LCA501.61.31.61.3-1.4-1.4-1.2-1.1(Nottingham,2004,Chidumayo,1997)Mortality&RegenerationParametersIntrinsicmortalityrateMi0.020.2-0.10.40.4-0.8-0.9-0.7-0.7Estimated(RyanandWilliams,2011),similartoDesanker&VM0029,Version1.0SectoralScope14Page69Prentice(1994)Earlyburnintensityscalekearly10650.20.10.50.4----Rothermelmodel,similartoHoffaetal.(1999),Robertson(1993)Earlyburnintensityshapeλearly1.86-0.5-0.40.0-0.1----Rothermelmodel,similartoHoffaetal.(1999),Robertson(1993))Lateburnintensityscaleklate34983.32.72.01.7----Rothermelmodel,similartoRyan&Williams(2011),Sheaetal.(1996)Lateburnintensityshapeλlate2.780.2-0.30.50.2----Rothermelmodel,similartoRyan&Williams(2011),Sheaetal.(1996)DBHatwhichseedlingdevelopsrootstockandtheabilitytoresprout(m)Sresprout0.020.1-0.40.10.0-0.10.00.10.0Estimated(RyanandWilliams,2011)Numberofseedlingsestablishedinarecruitmentyear(ha-1)Snew5000-0.20.20.20.00.0-0.10.00.0Estimated(RyanandWilliams,2011)ProbabilityofarecruitmentyearoccurringPrecruit0.03-0.1-0.50.20.2-0.1-0.10.10.0Estimated(RyanandWilliams,2011)ProbabilityofarootstockfailingtoresproutfollowingfireSmort0.04-0.1-0.80.00.10.00.10.00.0(RyanandWilliams,2011)Sensitivityanalysiswasconductedonanensembleof5000patchesunderanintensefireregime(FRI=1.11,p(burnearly)=0.33,p(burnlate)=0.66)andintheabsenceoffire,withtheresponsevariablesetas10-yearchangeinabovegroundbiomassofanaveragebiomass(21tC•ha-1)woodland.ModelsensitivityisspecifiedasSx=([Radj-Rn]/Rn)/([Padj-Pn]/Pn),wherexisthefactorbywhichthenominalvalueischanged,Radjistheresponseforthemodelrunwiththeadjustedvalue,Rnistheresponsewiththenominalvalue,andPnandPadjaretheparametervaluesforthenominalandadjustedcasesrespectively.VM0029,Version1.0SectoralScope14Page70A1.4.1AllometricEquationsTreeallometrydescribesthequantitativerelationsbetweenkey,easytomeasurecharacteristicsoftreesandothermoredifficulttoassessproperties.AllometricrelationshipsderivedfromdestructivemeasurementsoftreesinmiombowoodlandinNhambitaareusedextensivelyinthemodel.Theallometricmodelsusedaresummarizedintable3,andtheirderivationdescribedindetailinRyanetal.(2011).TableA3:Allometricequationsusedinthemodel,derivedfrommiombowoodlanddatainNhambitaEquationformsarespecifiedas:Power,y=axb,Linear,y=ax+b.nisthenumberofsamplesusedtofitthefunction,andR2givestheproportionofvariancedescribedbytheallometricmodel.A1.4.2GrowthParametersOftheparameterscontrollingpatchgrowth,modelpredictionswerefoundtobesensitivetorespiratoryfraction(Ra),maximumphotosyntheticrate(Pmax)andhalfsaturationofphotosyntheticrate(kp).Thesethreeparametersrelatetotheproductivityofmiombowoodlandsbycontrollingcarbonassimilationandgrowthrate.Respiratoryfractiondeterminestheproportionofcarbonassimilatedthroughphotosynthesisthatislosttotheatmospherethroughplantrespiration.Respiratoryfractionispoorlyconstrainedwithinsavannahwoodlands,asitisverydifficulttomeasureinsitu.However,itiswellconstrainedgloballyataround50percent(Waringetal.,1998).Asignificantdeviationfromthisvalueinmiombowoodlandsisnotexpected.Maximumphotosyntheticrateandhalfsaturationdescribetherateofphotosynthesisthatisexpectedunderdifferentlightconditions.Maximumrateofphotosynthesisoccursunderoptimallightconditions,andthisvalueislikelytolieintherange9-15µmolC·m-2·s-1(TuohyandChoinski,1990,Tuohyetal.,1991,Woollen,2013).Thoughindividualparametervaluesdeterminingmiombowoodlandgrowthareuncertain,therearegooddataonmiomboregrowthfollowingwoodlandclearancewhichcandemonstratethattherateofgrowthintheabsenceoffireisrepresentedappropriatelybythemodel(seeSection5.1).Dependent(y)variableIndependent(x)variableFormabnR2notesCstockofstem(kgC)DBH(m)Power42222.6290.93Cstem:Croot+stemDBH(m)Linear0.320.6230.26Heightoftreetop(m)DBH(m)Linear,withsaturation42.6800.67whereDBH>60cm,height=25mHeightofcanopybase(m)DBH(m)Linear,withsaturation22.3800.58whereDBH>67cm,canopybase=15m.Leafarea(m)BasalArea(m2)Linear1330100.29Plot-scaledataVM0029,Version1.0SectoralScope14Page71A1.4.3Mortality&RegenerationParametersAGBchangewasfoundtobelargelyinsensitivetoparameterscontrollingmiomboregeneration.Overthetime-scalesofaREDD+project,mechanismsofregenerationarenotexpectedtobehugelyimportantasbiomasschangeisconcentratedinlargestemswhichregenerateoverseveraldecades.Overlongertimescalesitisexpectedthatfirewillreducepressureonseedlingsandsproutsbyexposingstemstolowerintensityandlessfrequentfires.Theliftingofthis‘demographicbottleneck’shouldimprovethecapacityofwoodlandstoregenerate,thoughtheimpactsofthesechangesmayonlybeobservableinthedecadesfollowingfiremanagement.Stemmortalityisofconsiderablygreaterimportancetomodeloutput,particularlyfire-inducedtop-kill.Thissignalstheimportanceoffiretotheproductivityofmiombowoodlands.Dataonfiredynamicsinmiombowoodlandsisscarce,thoughthereismuchinformationonfirebehaviorinsavannahgrasslands,particularlyfromSouthAfrica(Table2).Fireintensityinmiombowoodlandsvariesfromverylow(<300kW/m)tohighintensity(>6,500kW/m).Thereisalsothepotentialforextremelyhighintensityfires(>10,000kW/m),thoughthesehavenotbeendetectedinmiombowoodlandsbypublishedfireexperiments.OfparticularsignificancearetheresultsofHoffaetal.(1999),whereaseriesofearly-seasonfiresinmiombowoodlandsshowedaprogressionfromverylowintensitiesthroughtothehigherintensitiesobservedbythemiddryseasonasdescribedinfigureA5.FigureA5:FireIntensityMeasurementsInZambiaFromHoffaEtAl.(1999),ShowingaTrendofIncreasingIntensitythroughtheFireSeasonRobertson(1993)alsorecordedthedifferencesbetweenearlyandlateburns,measuringearlyburnstobeofverylowintensity(<300kW/m)andlateburnstobeconsiderablymoreintenseandmorevariable(500-5000kW/m).RyanandWilliams(2011),settingfiresinthelatedryseason,exploitedwithinfirevariabilityatdifferenttimesofdaytoachieveawiderangeoffireintensities.Firesearlyinthemorningwereofsimilarintensitytoearlyburns(360kW/m),withfireintensityreaching6,600kW/matthehottestpartoftheday.Thoughcurrentlyavailableliteraturecannotprovidealltherequireddetailaboutseasonalvariabilityinfireintensity,itdoesgiveasuitableenvelopeofprobablefireintensitiesinmiombowoodlands,asseenintableA2below.VM0029,Version1.0SectoralScope14Page72TableA2:ObservationsofFireBehaviourinSouthernAfricanSavannahandMiomboWoodlandVegetationFuelTypeLocationROS(m/min)FlameLength(m)Intensity(kW/m)SourceMiomboMozambique3-55360-6600(RyanandWilliams,2011)SavannahSouthAfrica0.6-10228-17905(Govenderetal.,2006)SavannahZambia1.2-105.60.3-4.243-9476(Hélyetal.,2003)MiomboZambia6-480.7–3.225-6553(Hoffaetal.,1999)SavannahSouthAfrica12-601-6475-6130(Sheaetal.,1996)MiomboZambia18-483-51734-4061(Sheaetal.,1996)SavannahSouthAfrica(Stocksetal.,1996)Block5597.23.510906Block5637.81.74048SavannahSouthAfrica0.5-6(Trollopeetal.,1996)Head4.2-28.893-3644Back0.0-1520-160MiomboZimbabwe100-300500-5000(Robertson,1993)EarlyLateAcknowledgement:BillHigham.Furtherdescriptorsoffireseasonalitymaybegeneratedusingwidely-usedmodelsoffireintensity.TheRothermelmodelisasemi-empiricalmodeloffirespreadthathasbeenwidelyusedtounderstandvariationinfireintensityinawiderangeoffiresystems(Rothermel,1972).Itsequationslieatthecenterofmanymodernfiremodels,includingtheUnitedStatesDepartmentofAgriculturesystemfortheevaluationofoperationalfirehazard(Pyneetal.,1996)andintheSPITFIREmoduletotheLPJdynamicglobalvegetationmodel(Thonickeetal.,2010).Itsmodestinputrequirementsandadaptabilitymakeitanappropriatemeansofapproximatingseasonalfireintensityvariationinmiombowoodlands.TheRothermelmodelwasparameterizedusingvaluesfromtheliteratureaswellasfielddatafromKilwa,anddrivenbymeteorologicaldatafromtheECMWFERA-interimclimatereanalysisdatasetandfuelcuringestimatesgeneratedfromtheMODISarchive.Themodelarrangementisdescribedinappendix2.TheRothermeloutputwasusedtogeneratedistributionsdescribingtheintensityofearlyandlateburnsasdescribedinfigureA6below.Basedoninputdatafrom2000-2012andanearly/latecut-offdateoftheendofJune,twoWeibulldistributionswerefittedthatqualitativelyfitexpectedfireintensities,withgoodmatchesfortheexpectedseasonalfireintensityrange.Foreachearly/latefireevent,themodelissettorandomlysampleafireintensityfromthesedistributions.VM0029,Version1.0SectoralScope14Page73Figure6:NormalizedHistogramsofEarlyandLateFireIntensitiesPredictedtytheRothermelModel,FittedWithWeibullDistributionsDistributionsarespecifiedbyscale(k)andshape(λearly)parameters,withvaluesofkearly=1065,klate=1.86,λearly=3498andλlate=2.78.A1.5ModelEvaluationA1.5.1PatchGrowthAssumptionsrelatingtogrowthparameterswerevalidatedusingchronosequencefielddatafromNhambita.Achronosequenceisaseriesofforestsiteswithsimilarattributesthathavebeenregrownfollowingclearanceandabandonment,allowingtheanalysisofforestprocessesthatoccuroverlongtimescales,inthiscaseforestregrowth.Thechronosequencepredictsapproximately0.7tC/hatobeaccumulatedeachyearonabandonedlandwhichhasnotbeenobservedtoburn.Thissituationwassimulatedbyrunningthemodelfromastart-pointwithnolargestemsthrough25yearsintheabsenceoffireasdecriedinfigureA7.Themodelsuggestsasigmoidalgrowthform,relatingtoexponentiallyincreasingstandproductivitywithbiomassuptoasaturationpoint.Averagegrowthratesover25yearsproducedbythemodelamounttoabiomassincrementof0.63tC/ha,arateslightlylowerthanobservedinthechronosequence.Themodelpredictsbasalareaappropriately,withgrowthratematchingobservations.Themodelislessabletopredictstockingdensity,likelyduetotheinitializationof‘bare’patcheswithastockofsaplings,highlightingalimitationinthemodelrepresentationofregeneration.ThisdatarepresentsaconservativeestimateofproductivityastheannualincreaseinbiomassisVM0029,Version1.0SectoralScope14Page74belowthatobservedinmiombowoodlands.Additionally,chronosequenceobservationsareassumedtobeinthetotalabsenceoffire,whichisunlikelytobethecaseinreality.Figure7:ModelOutputComparisontoChronosequenceDatafromNhambitaBiomassincreasefollowingabandonmentispredictedaccurately(RMSE=5.3tC/ha,R2=0.66),asisbasalarea(RMSE=1.7m2/ha,R2=0.84),thoughthemodelisnotabletoveryeffectivelycapturetrendsinstockingdensityoftrees>5cmDBH(RMSE=250ha-1,R2=0.53).A1.5.2FireIntensityItisnotpossibletofullyevaluateperformanceoftheRothermelmodelduetothelackofappropriatevalidationdatafrommiombowoodlands.However,weareabletodemonstratethatthepredictedintensitiesforearlyandlatefiresarereasonable,andthusresultinaconservativerateofstemtop-kill.Areviewoftheliteratureonmiombofiressuggeststhatearlyburnshaveanintensityoflessthan500kW/m,whereastheRothermelmodelpredictsfireintensitytofrequentlyriseabove1000kW/m.(ForaREDDproject,whichexpectstoseeanincreaseinearlyseasonfires,thishigheraverageintensityisconservative.)Lateseasonfireshaveawiderrangeofintensities,from500kW/mtoover6,000kW/m,averagingaround4,000kW/m.TheRothermelmodelpredictsamedianlatefireintensityofaround3,000kW/m,andthatfireintensityveryrarelyrisesabove6,000kW/m.(AgainforaREDDproject,thisisconservativesinceitwillresultinusingalessseverebaselinescenariothanislikelytobethecase.).VM0029,Version1.0SectoralScope14Page75TheRothermelmodelestimateoffireintensityisbaseduponafixedearly/latecut-offdateoftheendofJune.Inreality,climaticvariabilitywillshiftthisdateannually,butpurposiveearlyburningcangenerallybeexpectedtoproducelowerintensityfireswithconsequentlowertreemortality.A1.5.3ExclusionofDeadWoodPoolAnalysisofthedeadwoodpoolshowsthatitcanbeconservativelyexcludedasfollows.Thetwoscenarios(baselineandproject)aretreatedseparately.Sinceexcludingthepoolisconservativeforbothcomponents,excludingitmustbeconservativeoverall.A1.5.3.1BaselineScenarioAtthestartofprojectactivitiestherewillalreadybesomedeadwoodpresentintheforest.Thisdeadwoodpoolwilldecayoveraperiodoftimereleasingitscarbonintotheatmosphere.Therateofdecaywillbeaffectedbythefireregime;underthebaselinescenariothisdoesnotchange.Theaboveshowsthatharshfireregimessuchaspertaininthebaselinescenarioleadtoadegradationincarbonstocks,whichscaleswithstartingbiomass.Thus,asthestandingbiomassdeclines,futureadditionstothedeadwoodpoolwillbelowerthanpreviousadditionstothepool.Sincetherateofdecayoutofthepoolisconstant,thecarboncontentofnewadditionseachyearwillbelowerthantheamountlostduetodecayofexistingdeadwood.Andthusexcludingthedeadwoodpoolconservativelyestimatesactualcarbonemissionsinanygivenyear.A1.5.3.2ProjectScenarioExcludingthedeadwoodpool,andthustreatingtreemortalityasleadingtoinstantaneousemissionsratherthantheslowdecaythatactuallytakesplace,willalwaysbeconservativeintheProjectScenario.A1.5.4AssessmentofModelConservativenessKeytothismodellingframeworkistheassertionthatmodelpredictionsresultinconservativeestimatesofabovegroundbiomasschangeinallcases.Fromthemodelparameterizationitcanbeconcludedthatmodeloutputswillresultinconservativeestimatesofbiomasschange.A1.5.4.1DegradationTreetop-killfromhighintensitylate-seasonfiresdominatesishigherinthebaselinescenariothanintheprojectscenario,soinordertobeconservativethemodelshouldunder-estimatetheselosses.Section5.2describeshowthefireintensityusedbythemodelislowerthanthatobservedinreality.Hencethemodelisconservativewithrespecttodegradation.5.4.2AccumulationAccumulationwillbehigherintheprojectscenariothaninthebaselinescenario,soinordertobeconservativethemodelshouldunder-estimatethesegains.Section5.1showsthatmodelledwoodlandproductivityisbelowthatobservedinmiombowoodlands,andthereforeestimatedbiomassaccumulationratesarelikelylowerthanthosethatwillbeobservedinreality.Hencethemodelisconservativewithrespecttoaccumulation.VM0029,Version1.0SectoralScope14Page76A1.6UsingtheModelThemodelhasbeenportedintoMicrosoftExcelasafamiliarenvironmentforuserstoexaminedetailedmodeloutputs.Thespreadsheetprovideddoesnotcontainanyofthemodelitself,butisaninterfaceforsettingmodelparametersandvisualizingoutputs.Thespreadsheetallowsforsimpleuseofcommonlyusedmodelfunctions,andalsoanalysisoftheimpactofparameterchangesandtheexplorationofrawdataoutputs.Herewedescribeuseofthemodelforpredictionofcarbonstoragechangesfromamodifiedfireregime.Themodelisprovidedinacompressedfolder(‘GapFire.zip’),andshouldbeunzippedtotheC:\drive.Open‘GapFire.xlsm’toaccessthemodelspreadsheet,dismissingthesecuritywarningthatappearsatthetopoftheworkbook(‘EnableContent’).Theprimarycontrolsforthemodelarenumberofensembles(‘patches’),projectstartdate,initialbiomass,andfrequencyofearlyandlatefiresinthebaselineandprojectscenariosasdescribedinfigureA8below.Increasingthenumberofpatcheswillresultinalessvariableoutput,butwillincreasemodelruntime.Aminimumof10,000patchesshouldbeusedtoachieveareliablemodelrun,with100,000patchesrecommended.Initialbiomasscanbealteredtoreflectthemeanstartingbiomassofpatches.Duetotherandomallocationofstemstoeachpatchtheactualinitialbiomassmaybeslightlydifferenttothatselected,butsuchdifferencesareminimalwhenalargenumberofpatchesarerun.FigureA8:PrimaryinputstothemodelforuseincarbonaccountingFirefrequencyshouldbeinputbasedonhistoricalfiremappingandongoingmonitoring.Themodelshouldbestartedfromtheyearthedataforthecarbondensitymapwasobtained;thismaypredatethestartofprojectactivitiesforafewyears;thenumberofyearsafterthemapdatemustbeenteredastheProjectStart.Thebaselinefirescenariomustthenbecopiedintotherelevantcellsoftheprojectfireregime.Theactualfirefrequencyforeachyearofprojectactivitiesasdeterminedbymonitoringmustthenbeinputintotheappropriatecells.Yearswhichareleftblankrelatetofutureyears;themodelwillcomputethemeanfirefrequenciessincethestartofprojectactivitiesandusethemtopredictperformanceuptoyear10.Notethattheprobabilityofnofireisequalto1–pEarly–pLate.Torunthemodel,theparameterinputfileshouldfirstbeupdatedusing‘UpdateParameters’,whichwritesselectedparameterstoacommaseparatedvariables(.csv)file.ThemodelisstartedusingVM0029,Version1.0SectoralScope14Page77‘RunGapFire’andoncecompletethespreadsheetshouldbeupdatedwithnewmodeloutputsusing‘UpdateResults’asdescribedinfigureA9.FigureA9:ModelControls,Initiating1)WritingofModelInputFile,2)ModelRun,and3)SpreadsheetUpdateModeloutputsaresummarisedinatableshowingabovegroundcarbonstorageinthebaselineandprojectscenarios,alongwithestimatesofavoidedemissions,sequesteredcarbonandnetchangeasdescribedinfigureA10.ThissummarycontainstheinformationrequiredforaREDD+projecttoaccountforabovegroundbiomasschange.FigureA10:SummaryofCarbonStorageChangePredictedbytheModelAgraphicalsummaryofcarbonstoragechangeintheprojectandbaselinescenariosisalsogenerated,aswellasestimatesofcarbonstoragechangebetweenthetwoscenariosandtherateofcarbonaccumulationasdescribedinfigureA11below.Avoidedemissionsarecalculatedasthedifferencebetweeninitialbiomassandbaselinepredictedbiomass,andsequesteredcarboniscomputedastheincreaseinabovegroundbiomassintheprojectscenariosincemodelinitialization.Alldataispresentedinarawformattoallowformoredetailedanalysisofpotentialprojectimpacts.Furtherinformationispresentedonthe‘Parameters’worksheet,wheremodelparametersandvariablescanbealtered.Rawmodeloutputsaredisplayedinthe‘DATA’worksheet,whereadditionalmodeloutputscanbeexplored.Foramoreadvanceduser,modelcodeissuppliedwritteninFortran90,whichmaybecompiledtotheproject’sspecification.VM0029,Version1.0SectoralScope14Page78Figure11:ModelOutputGraphicalandTabularSummaryReferencesAlexander,M.E.1982.Calculatingandinterpretingforestfireintensities.CanadianJournalofBotany,60,349-357.Bond,W.J.&Midgley,J.J.2001.Ecologyofsproutinginwoodyplants:thepersistenceniche.TrendsinEcology&Evolution,16,45-51.Byram,G.1959.Combustionofforestfuels.In‘Forestfire:controlanduse’.(Ed.KPDavis)pp.61–89.McGraw-Hill:NewYork.Campbell,B.M.1996.TheMiombointransition:woodlandsandwelfareinAfrica,Cifor.Castellanos,J.,Jaramilllo,V.C.J.,SanfordJR,R.L.&Kauffman,J.B.2001.Slash-and-burneffectsonfinerootbiomassandproductivityinatropicaldryforestecosysteminMexico.ForestEcologyandManagement,148,41-50.Chidumayo,E.N.1997.Miomboecologyandmanagement:anintroduction,IntermediateTechnologyPublicationsLtd(ITP).Chidumayo,E.N.2004.DevelopmentofBrachystegia-Julbernardiawoodlandafterclear-fellingincentralZambia:Evidenceforhighresilience.AppliedVegetationScience,7,237-242.Desanker,P.V.1996.DevelopmentofaMIOMBOwoodlanddynamicsmodelinZambezianAfricausingMalawiasacasestudy.ClimaticChange,34,279-288.VM0029,Version1.0SectoralScope14Page79Desanker,P.V.&Prentice,I.C.1994.MIOMBO–AvegetationdynamicsmodelforthemiombowoodlandsofZambezianAfrica.ForestEcologyandManagement,69,87-95.Goldammer,J.G.&DeRonde,C.2004.WildlandfiremanagementhandbookforSub-SaharaAfrica,AfricanMinds.Govender,N.,Trollope,W.S.&VanWilgen,B.W.2006.Theeffectoffireseason,firefrequency,rainfallandmanagementonfireintensityinsavannavegetationinSouthAfrica.JournalofAppliedEcology,43,748-758.Hély,C.,Alleaume,S.,Swap,R.&Justice,C.2003.SAFARI-2000characterizationoffuels,firebehavior,combustioncompleteness,andemissionsfromexperimentalburnsininfertilegrasssavannasinwesternZambia.JournalofAridEnvironments,54,381-394.Hendricks,J.J.,Hendrick,R.L.,Wilson,C.A.,Mitchell,R.J.,Pecot,S.D.&Guo,D.2006.Assessingthepatternsandcontrolsoffinerootdynamics:anempiricaltestandmethodologicalreview.JournalofEcology,94,40-57.Jackson,J.F.,Adams,D.C.&Jackson,U.B.1999.Allometryofconstitutivedefense:Amodelandacomparativetestwithtreebarkandfireregime.AmericanNaturalist,153,614-632.Johnson,E.A.&Miyanishi,K.2001.Forestfires:behaviorandecologicaleffects,AcademicPressSanDiego,CA.Jones,H.G.1992.Plantsandmicroclimate:aquantitativeapproachtoenvironmentalplantphysiology,CambridgeUniversityPress.Miller,C.&Urban,D.L.2000.ModelingtheeffectsoffiremanagementalternativesonSierraNevadamixed-coniferforests.EcologicalApplications,10,85-94.Mlambo,D.&Mapaure,I.2006.Post-fireresproutingofColophospermummopanesaplingsinasouthernAfricansavanna.JournalofTropicalEcology,22,231-234.Norman,J.M.&Campbell,G.S.1991.Canopystructure.Plantphysiologicalecology.Springer.Nottingham,A.2004.ThecharacterisationoffoliarN-concentrationsandpresenceofN-fixersinMiombowoodland,Mozambique.MRes,UniversityofDundee.Pyne,S.J.,Andrews,P.L.&Laven,R.D.1996.Introductiontowildlandfire,JohnWileyandSons.Robertson,F.E.1993.Early-burningintheBrachystegiawoodlandoftheParksandWildLifeEstate.ZimbabweScienceNews,27,68-71.Rothermel,R.C.1972.Amathematicalmodelforpredictingfirespreadinwildlandfuels,USFS.Running,S.W.,Nemani,R.R.&Hungerford,R.D.1987.Extrapolationofsynopticmeteorologicaldatainmountainousterrainanditsuseforsimulatingforestevapotranspirationandphotosynthesis.CanadianJournalofForestResearch,17,472-483.VM0029,Version1.0SectoralScope14Page80Ryan,C.M.&Williams,M.2010.Howdoesfireintensityandfrequencyaffectmiombowoodlandtreepopulationsandbiomass?EcologicalApplications,21,48-60.Ryan,C.M.,Williams,M.&Grace,J.2011.Above‐andBelowgroundCarbonStocksinaMiomboWoodlandLandscapeofMozambique.Biotropica,43,423-432.Saito,M.,Poulter,B.,Bellassen,V.,Yue,C.,Ryan,C.M.,Williams,M.,Luysseart,S.,Ciais,P.&Peylin,P.inprep.FireregimesandvariabilityinabovegroundwoodybiomassinanAfricansavannawoodland.Shea,R.W.,Shea,B.W.,Kauffman,J.B.,Ward,D.E.,Haskins,C.I.&Scholes,M.C.1996.FuelbiomassandcombustionfactorsassociatedwithfiresinsavannaecosystemsofSouthAfricaandZambia.JournalofGeophysicalResearch:Atmospheres,101,23551-23568.Shugart,H.H.&Smith,T.M.1996.Areviewofforestpatchmodelsandtheirapplicationtoglobalchangeresearch.ClimaticChange,34,131-153.Stocks,B.,VanWilgen,B.,Trollope,W.,Mcrae,D.,Mason,J.,Weirich,F.&Potgieter,A.1996.Fuelsandfirebehaviordynamicsonlarge-scalesavannafiresinKrugerNationalPark,SouthAfrica.JournalofGeophysicalResearch,101,23541-23,550.Sutherland,E.K.&Smith,K.T.2000.Resistanceisnotfutile:theresponseofhardwoodstofire-causedwounding.In:Proceedings:workshoponfire,people,andthecentralhardwoodlandscape.GeneralTechnicalReportNE-274,USDepartmentofAgriculture,ForestService,NortheasternForestExperimentStation,NewtownSquare,PA,2000.111-115.Trollope,W.,Trollope,L.,Potgieter,A.&Zambatis,N.1996.SAFARI-92characterizationofbiomassandfirebehaviorinthesmallexperimentalburnsintheKrugerNationalPark.JournalofGeophysicalResearch,101,23531-23,539.Tuohy,J.&Choinski,J.1990.ComparativephotosynthesisindevelopingleavesofBrachystegiaspiciformisBenth.JournalofExperimentalBotany,41,919-923.Tuohy,J.M.,Prior,J.A.&Stewart,G.R.1991.PhotosynthesisinrelationtoleafnitrogenandphosphoruscontentinZimbabweantrees.Oecologia,88,378-382.Waring,R.,Landsberg,J.&Williams,M.1998.Netprimaryproductionofforests:aconstantfractionofgrossprimaryproduction?TreePhysiology,18,129-134.Williams,M.1996.Athree-dimensionalmodelofforestdevelopmentandcompetition.EcologicalModelling,89,73-98.Woollen,E.2013.CarbondynamicsinAfricanmiombowoodlands:fromtheleaftothelandscape.PhD,EdinburghUniversity.Designcredits:Figure1,Horhew&redheadstock(deviantART.com).VM0029,Version1.0SectoralScope14Page81APPENDIX2:THEROTHERMELMODELThisappendixwaspreparedbySamBowersandProf.MathewWilliamsoftheUniversityofEdinburgh.Rothermel’smodelsimulatesfireasaquasi-steadystateseriesofignitionsinaspatiallyuniformfuelbed.Rateoffirespreadismodelledastheratioofpropagatingheatfluxtotheenergyrequiredtodryoutandigniteunburnedfuels.Firerateofspreadisinfluencedbythegeometry,compositionandmoisturecontentofthefuelbed,andmultiplicativefactorsdescribingwindspeedandslope.ThecompleteRothermelmodeloffirespreadis:𝑅=𝐼𝑅𝜉(1+𝜙𝑤+𝜙𝑠)𝜌𝑏𝜀𝑄𝑖𝑔(1)whereR=rateofspread(mmin-1),IR=reactionintensityoftheflamingfront(kJm-2min-1),ξ=propagatingfluxratio(dimensionless),φw=windcoefficient(dimensionless),φw=slopecoefficient(dimensionless),ρB=fuelbedbulkdensity(kgm-3),ε=effectiveheatingnumber(proportionoffuelraisedtoignitiontemperature)(dimensionless),andQig=heatofpre-ignition(quantityofheatrequiredtoignitefuel,afunctionofmoisturecontent)(kJkg-1).Reactionintensity(IR)istheproductofseveralfactorsthatrelatetorateofenergyrelease,specifiedby:𝐼𝑅=Γ′𝑊𝑛ℎ𝜂𝑀𝜂𝑠(2)whereΓ’=optimumreactionvelocity(min-1),Wn=netfuelloading(kgm-2),h=lowheatcontent(kJkg-1),ηM=moisturedampingcoefficientandηs=mineraldampingcoefficient.Reactionvelocityiscalculatedasafunctionofsurfaceareatovolumeratio(σ).ForfurtherinformationontheformulationoftheRothermelmodel,refertoRothermel(1972).Importantly,theRothermelmodelcanbeappliedtoestimateoffirelineintensitybycombiningmodelledrateofspreadwithcombustioncompleteness(Cf),hereestimatedusingthemethodofPetersonandRyan(1986):1-hrfuels:𝑖𝑓𝑚𝑓𝑚𝑒𝑥𝑡≤0.18,𝐶𝑓=1.0𝑖𝑓0.18<𝑚𝑓𝑚𝑒𝑥𝑡≤0.73,𝐶𝑓=1.2−0.62𝑚𝑓𝑚𝑒𝑥𝑡𝑖𝑓𝑚𝑓𝑚𝑒𝑥𝑡>0.73,𝐶𝑓=2.45−2.45𝑚𝑓𝑚𝑒𝑥𝑡(3)10-hrfuels:𝑖𝑓𝑚𝑓𝑚𝑒𝑥𝑡≤0.12,𝐶𝑓=1.0𝑖𝑓0.12<𝑚𝑓𝑚𝑒𝑥𝑡≤0.51,𝐶𝑓=1.09−0.72𝑚𝑓𝑚𝑒𝑥𝑡𝑖𝑓𝑚𝑓𝑚𝑒𝑥𝑡>0.51,𝐶𝑓=1.47−1.47𝑚𝑓𝑚𝑒𝑥𝑡(4)VM0029,Version1.0SectoralScope14Page82100-hrfuels:𝑖𝑓0𝑚𝑓𝑚𝑒𝑥𝑡≤0.38,𝐶𝑓=0.98−0.85𝑚𝑓𝑚𝑒𝑥𝑡𝑖𝑓𝑚𝑓𝑚𝑒𝑥𝑡>0.38,𝐶𝑓=1.06−1.06𝑚𝑓𝑚𝑒𝑥𝑡(5)Wheremf=moisturecontentoffuel(dimensionless)andmext=fuelmoistureoffireextinction(dimensionless).Fuelsaresplitintothreesizeclasses:1-hour-<0.64cm,10-hour–0.64-2.5cm,100-hour–2.5cm-7.5cm,whicheachhavecharacteristicsurfacearetovolumeratioanddryingrates.The1-hrfuelclassissplitintolive(greengrass)anddead(drygrass,litter)classes,whichhavedistinctbehaviors.Theinputofeachfuelclasstotheoverallrateofspreadiscalculatedbyweightingthedifferentfuelclassesbytheirsurfaceareatovolumeratios.TheRothermelmodelrequiresanumberofinputstopredictfireintensity,summarizedintableA4below.Thesehavebeenselectedfromacombinationoffieldmeasurementsandstandardizedparametersfromtheliterature.Inadditiontofixedfuelparameters,anumberofinputswillbeexpectedtochangeseasonally,namelyfuelcuring,fuelmoistureandwindspeed.Thesedriversareestimatedasfollows:FuelCuring:Annualcuring(drying)ofgrassfuelsisoneofthekeyinfluencesonannualfireintensityvariationinmiombowoodlands(Hoffaetal.,1999).TheproportionofgrassthatiscuredisapproximatedusingNDVImeasurementsfromMODIS,coveringareasofKilwawithinthe5-35tC/habiomassrange.AstherelationshipbetweenNDVIandfuelcuringisofteninconsistentbetweenlocations,a‘relativegreenness’valueiscalculatedbasedonthemaximumandminimumrecordedNDVIvalueineachpixel(j):𝑅𝐺𝑖,𝑗=𝑁𝐷𝑉𝐼𝑖,𝑗−𝑁𝐷𝑉𝐼𝑚𝑖𝑛,𝑗𝑁𝐷𝑉𝐼𝑚𝑎𝑥,𝑗−𝑁𝐷𝑉𝐼𝑚𝑖𝑛,𝑗(6)whereiisthetimeofmeasurementandNDVImax,jandNDVImin,jarethemaximumandminimumNDVIvaluesrecordedateachpixeloveraspecifiedtimeperiod.Thistimeperiodwassetto3yearstomaximizethechancesofcapturingextremesofNDVIwhilstminimizingtheimpactoflandcoverchangesonthepredictionsoffuelcuring(Newnhametal.,2011).ThismethodissimilartothatusedinAustralia’sGrasslandFireIndexandtheUnitedStatesNationalFireDangerRatingSystem(NFDRS)(Burganetal.,1998).FuelMoisture:Fuelmoistureofdeadfuelvariesthroughtheyearbasedonrainfalleventsandsubsequentdrying.DeadfuelmoistureisestimatedusingthemethodofThonickeetal.(2010),whichisbaseduponcalculationoftheNesterovIndexoffiredanger(Nesterov,1949):𝑁𝐼(𝑑)=∑𝑇𝑚𝑎𝑥(𝑑)∙(𝑇𝑚𝑎𝑥(𝑑)−𝑇𝑑𝑒𝑤(𝑑))(7)whereTmax(d)isthemaximumtemperatureandTdew(d)isthedewpointatdaydandsummationisoverthenumberofconsecutivedaysthatprecipitationremains<3mm.Whenadayreceives>3mmprecipitation,theNestrovIndexresetsto0.TemperatureandprecipitationdataaretakenfromtheVM0029,Version1.0SectoralScope14Page83ERA-interimclimatereanalysisdataset,andTdew(d)isapproximatedasTmin(d)–4(Runningetal.,1987)whereTministheminimumdailytemperature.Moisturecontentforeachfuelclassiscalculatedas:𝜔𝑖=𝑒−𝛼𝑖∙𝑁𝐼(𝑑)(8)whereωi=relativefuelmoistureofsizeclassi,andαiisappliedtothethreefuelclassesininverseproportiontothesurfaceareatovolumeratios(α1h=1.0⋅10−3,α10h=5.42⋅10−5,α100h=1.49⋅10−5).Livefuelmoistureisestimatedfromdegreeofcuring,whereamaximumgreennessrelatedto230percentmoisturecontent,andminimumcuringto30percentmoisturecontent,wherelivefuelsbehaveinthesamemannerasdeadfuels.Asadynamicfuelmodel,fuelistransferredfromthelivetothedeadclassinproportiontothequantityofthegrassfuelthatiscured(ScottandBurgan,2005).WindSpeed:Windhasaconsiderableimpactonfirebehavior,withstrongerwindsrelatingtoamoreintensefire.Windspeedat10misusedfromtheERA-interimdataset,andadjustedbyafactorof0.45torepresentthereductionofwindspeedwithinwoodlands.VM0029,Version1.0SectoralScope14Page84TableA4:RothermelmodelparametersResponsevariablesaremedianearly/latefireintensity.ModelsensitivityisspecifiedasSx=([Radj-Rn]/Rn)/([Padj-Pn]/Pn),wherexisthefactorbywhichthenominalvalueischanged,Radjistheresponseforthemodelrunwiththeadjustedvalue,Rnistheresponsewiththenominalvalue,andPnandPadjaretheparametervaluesforthenominalandadjustedcasesrespectively.EarlyLateParameterDescriptionParameterNameNominalValueS0.5S0.75S1.5S2S0.5S0.75S1.5S2SourceFuelmass(kgm-2)●grass●1hr(other)●10hr●100hrWliveW1-hrW10-hrW100-hr0.590.110.050.060.80.10.0-0.10.70.10.0-0.10.90.20.0-0.10.80.20.0-0.11.00.00.00.01.00.00.00.01.00.00.0-0.11.00.10.00.0Fieldmeasurements(Kilwa).Surfaceareatovolumeratio(cm-1)●grass●1hr(other)●10hr●100hrσliveσ1-hrσ10-hrσ100-hr59.7465.53.580.980.20.10.00.00.20.20.00.00.50.30.00.00.50.30.00.00.50.00.00.00.50.10.00.00.30.10.00.00.70.20.00.0Fieldmeasurements(Nhambita).(σlive,σ1-hr)(ScottandBurgan,2005)(σ10-hr,σ100-hr)Fuelbeddepth(m)δ0.451.01.01.01.01.11.11.11.1Fieldmeasurements(Kilwa)Windadjustmentfactorwadj0.451.11.21.41.51.21.31.51.6Estimated,similartoThonickeetal.(2010)Maxlivingfuelmoistureωmax120%-0.9-0.9-0.5-0.3-0.3-0.3-0.2-0.2ScottandBurgan2005)Minlivingfuelmoistureωmin30%-0.1-0.1-0.1-0.1-0.1-0.1-0.1-0.1ScottandBurgan2005)Fuelheatcontent(kJkg-1)h18,6221.51.72.42.71.51.72.53.0ScottandBurgan2005)VM0029,Version1.0SectoralScope14Page85Fuelparticledensity(kgm-3)ρp5130.10.10.00.00.10.10.00.0ScottandBurgan2005)Minimumfirelineintensityforcontinuedcombustion(kW/m)Imin500.00.00.00.00.00.00.00.0EstimatedPage86Byrunningthemodelonadailytimestepforallyearsofavailabledata(2000-2012),aclearfireseasonispredictedbeginninginmid-May,lastingthroughtoDecember(Figure12).FireintensityispredictedtoremainlowinMay/June,peakinSeptemberandreducetowardstheendoftheyear.Themodelsuggeststhatfireintensityishighlyvariableatanygiventimeofyear,thoughaverageintensityoflateseasonfiresaresignificantlygreaterthanearlyseasonfires.Bysplittingthesepredictedfireintensitiesinto‘early’(day≤180)and‘late’(day>180)seasonfires,twodifferentfireintensityPDFswereproducedfigure12.Figure12:PredictionsofFireSeasonalityfromtheRothermelModelIndividualyears(2000-2012)areplottedingreyandthemeanofallyearsplottedinred.ReferencesBurgan,R.E.,Klaver,R.W.&Klaver,J.M.1998.Fuelmodelsandfirepotentialfromsatelliteandsurfaceobservations.InternationalJournalofWildlandFire,8,159-170.Hoffa,E.A.,Ward,D.,Hao,W.,Susott,R.&Wakimoto,R.1999.SeasonalityofcarbonemissionsfrombiomassburninginaZambiansavanna.JournalofGeophysicalResearch:Atmospheres,104,13841-13853.Nesterov,V.1949.Combustibilityoftheforestandmethodsforitsdetermination.Moscow:Goslesbumaga.Page87Newnham,G.J.,Verbesselt,J.,Grant,I.F.&Anderson,S.A.2011.Relativegreennessindexforassessingcuringofgrasslandfuel.RemoteSensingofEnvironment,115,1456-1463.Peterson,D.L.&Ryan,K.C.1986.Modelingpostfireconifermortalityforlong-rangeplanning.EnvironmentalManagement,10,797-808.Rothermel,R.C.1972.Amathematicalmodelforpredictingfirespreadinwildlandfuels,USFS.Running,S.W.,Nemani,R.R.&Hungerford,R.D.1987.Extrapolationofsynopticmeteorologicaldatainmountainousterrainanditsuseforsimulatingforestevapotranspirationandphotosynthesis.CanadianJournalofForestResearch,17,472-483.Scott,J.H.&Burgan,R.E.2005.Standardfirebehaviorfuelmodels:acomprehensivesetforusewithRothermel'ssurfacefirespreadmodel.TheBarkBeetles,Fuels,andFireBibliography,66.Thonicke,K.,Spessa,A.,Prentice,I.,Harrison,S.,Dong,L.&Carmona-Moreno,C.2010.Theinfluenceofvegetation,firespreadandfirebehaviouronbiomassburningandtracegasemissions:resultsfromaprocess-basedmodel.Biogeosciences,7,1991-2011.Page88DOCUMENTHISTORYVersionDateCommentv1.08May2015Initialversion