简介：Thispaperfocusesontheuseofmodelsforincreasingtheprecisionofestimatorsinlarge-areaforestsurveys.Itismotivatedbytheincreasingavailabilityofremotelysenseddata,whichfacilitatesthedevelopmentofmodelspredictingthevariablesofinterestinforestsurveys.Wepresent,reviewandcomparethreedifferentestimationframeworkswheremodelsplayacorerole:model-assisted,model-based,andhybridestimation.Thefirsttwoarewellknown,whereasthethirdhasonlyrecentlybeenintroducedinforestsurveys.Hybridinferencemixesdesignbasedandmodel-basedinference,sinceitreliesonaprobabilitysampleofauxiliarydataandamodelpredictingthetargetvariablefromtheauxiliarydata.Wereviewstudiesonlarge-areaforestsurveysbasedonmodel-assisted,modelbased,andhybridestimation,anddiscussadvantagesanddisadvantagesoftheapproaches.Weconcludethatnogeneralrecommendationscanbemadeaboutwhethermodel-assisted,model-based,orhybridestimationshouldbepreferred.Thechoicedependsontheobjectiveofthesurveyandthepossibilitiestoacquireappropriatefieldandremotelysenseddata.Wealsoconcludethatmodellingapproachescanonlybesuccessfullyappliedforestimatingtargetvariablessuchasgrowingstockvolumeorbiomass,whichareadequatelyrelatedtocommonlyavailableremotelysenseddata,andthuspurelyfieldbasedsurveysremainimportantforseveralimportantforestparameters.

简介：Background:Informationonabove-groundbiomass(AGB)isimportantformanagingforestresourceuseatlocallevels,landmanagementplanningatregionallevels,andcarbonemissionsreportingatnationalandinternationallevels.Inmanytropicaldevelopingcountries,thisinformationmaybeunreliableoratascaletoocoarseforuseatlocallevels.ThereisavitalneedtoprovideestimatesofAGBwithquantifiableuncertaintythatcanfacilitatelandusemanagementandpolicydevelopmentimprovements.Model-basedmethodsprovideanefficientframeworktoestimateAGB.Methods:UsingNationalForestInventory(NFI)datafora~1,000,000hastudyareainthemiomboecoregion,Zambia,weestimatedAGBusingpredictedcanopycover,environmentaldata,disturbancedata,andLandsat8OLIsatelliteimagery.Weassesseddifferentcombinationsofthesedatasetsusingthreemodels,asemiparametricgeneralizedadditivemodel(GAM)andtwononlinearmodels(sigmoidalandexponential),employingageneticalgorithmforvariableselectionthatminimizedrootmeansquarepredictionerror(RMSPE),calculatedthroughcross-validation.Wecomparedmodelfitstatisticstoanullmodelasabaselineestimationmethod.Usingbootstrapresamplingmethods,wecalculated95%confidenceintervalsforeachmodelandcomparedresultstoasimpleestimateofmeanAGBfromtheNFIgroundplotdata.Results:Canopycover,soilmoisture,andvegetationindiceswereconsistentlyselectedaspredictorvariables.ThesigmoidalmodelandtheGAMperformedsimilarly;forbothmodelstheRMSPEwas-36.8tonnesperhectare(i.e.,57%ofthemean).However,thesigmoidalmodelwasapproximately30%moreefficientthantheGAM,assessedusingbootstrappedvarianceestimatesrelativetoanullmodel.Afterselectingthesigmoidalmodel,weestimatedtotalAGBforthestudyareaat64,526,209tonnes(+/-477,730),withaconfidenceinterval20timesmoreprecisethanasimpledesignbasedestimate.Conclusions:OurfindingsdemonstratethatNFIdatamaybecombinedwithfreelyavailablesate

简介：Background:Inrecentdecades,nativeAraucariaforestsinBrazilhavebecomefragmentedduetotheconversionofforesttoagriculturallandsandcommercialtreeplantations.Consequently,theforestdynamicsinthisforesttypehavebeenpoorlyinvestigated,asmostfragmentsarepoorlystructuredintermsoftreesizeanddiversity.Methods:Wedevelopedadistance-independentindividualtree-growthmodeltosimulatetheforestdynamicsinanativeAraucariaforestlocatedpredominantlyinsouthernBrazil.Thedatawerederivedfrom25contiguousplots(1ha)establishedinaprotectedarealeftundisturbedforthepast70years.Theplotsweremeasuredat3-yearintervalsfromtheirestablishmentin2002.Alltreesabovea10-cmdiameteratbreastheightweretagged,identifiedastospecies,andmeasured.Becausethisforesttypecompriseshundredsoftreespecies,weclusteredthemintosixecologicalgroups:understory,subcanopy,uppercanopyshade-tolerant,uppercanopylight-demanding,pioneer,andemergent.Thediameterincrement,survival,andrecruitmentsub-modelswerefittedforeachspeciesgroup,andparameterswereimplementedinasimulationsoftwaretoprojecttheforestdynamics.Thegrowthmodelwasvalidatedusingindependentdatacollectedfromanotherresearchareaofthesameforesttype.Tosimulatetheforestdynamics,weprojectedthespeciesgroupandstandbasalareasfor50yearsunderthreedifferentstand-densityconditions:low,average,andhigh.Results:Emergentspeciestendedtogrowinbasalarea,irrespectiveoftheforestdensityconditions.Conversely,shade-tolerantspeciestendedtodeclineovertheyears.Underlow-densityconditions,themodelshowedagrowthtendencyforthestandbasalarea,whileunderaverage-densityconditions,forestgrowthtendedtostabilizewithin30years.Underhigh-densityconditions,themodelindicatedadeclineinthestandbasalareafromtheonsetofthesimulation,suggestingthatundertheseconditions,theforesthasalreadyreachedit

简介：Thusfar,measurementsandestimationsofactualevapotranspiration(ET)inextremelyaridareasarestillinsufficient.BasedonsuccessiveobservationsfromJune–September2014,wesimulatedETofaPopuluseuphraticaOliv.forestduringthegrowingseasoninanextremelyaridregionofnorthwestChinausingtheShuttleworth–Wallace(S–W)model.SimulatedETvalueswerecomparedtothoseoftheeddy-covariance(EC)methodona1hinterval.Witharootmeansquareerror(RMSE),relativeerror(RE)andmeanabsoluteerror(MAE)of0.192,3.100and0.165mmh-1,respectively,modelperformancewasnotsatisfactory.Inparticular,ondayswithstrongwinds(Sep.11–13),deviationsbetweensimulatedandobservedETvaluesincreasedto0.275,0.878and0.251mmh-1,RMSE,REandMAErespectively.Thesevaluesweresignificantlygreaterthanthoseinotherstudyperiodsandweremostlikelyowingtosharpincreasesinwindspeed.Asaresult,thereweresubstantialadvectiveeffects,whichisnotconsistentwiththeassumptionoftheS–Wmodelthattherearenoadvectiveeffectsormesoscalecirculationpatternsinducedbysurfacediscontinuities.